Hsu Ken Ooi

Control, Influence and Agency

Sat, 30 May 2026 00:00:00 GMT

Something clicked recently when thinking about the difference between control, influence and agency. People think a lot about the first two. Almost nobody thinks clearly about agency.

Control

Having control over something is surprisingly rare.

If I asked you whether you control your body, you'd say yes. You can wave your arms, walk across the room, stand up whenever you want. But if you're sick and I ask you to make yourself not sick, you can't do that. If you're swimming in a frozen lake and you want to be warm, you can't will your body temperature up.

So how much control do you actually have?

It's the same at work. You might think you control your career. You pick what to work on, how hard to work, when to switch jobs. But you don't control whether your company gets acquired, whether your industry gets disrupted, whether your manager decides she doesn't like you. You control your inputs. You don't control the outcomes.

The pattern repeats everywhere. You have control over a much smaller slice of your life than you think. And thinking you have control when you don't is painful. If you're like most people, you'll exhaust yourself trying to control things only to discover it's futile.

Influence

What you actually have is varying degrees of influence.

Influence means you can affect an outcome but can't determine it. You do things that make a result more or less likely, but the final outcome depends on other people, other forces, other variables you can't see.

A manager trying to get their team to hit a deadline has influence. They can set clear priorities, remove blockers, keep morale up. But they can't force each person to do great work. Everyone on the team has their own motivations, their own problems, their own capacity on any given day. The manager's job is to push the probability of a good outcome higher. That's influence.

A founder pitching investors has influence too. They can tell a compelling story, show strong metrics, build a relationship over time. But the investor has their own thesis, their own portfolio constraints, their own mood that morning. The founder can do everything right and still get a no.

The hard part about influence is you can't measure it precisely. With control, the feedback is clean. You flip the switch, the light turns on. With influence, the feedback is noisy. You did ten things, the outcome went your way, and you don't know which of those ten things mattered. Or you did ten things, the outcome went badly, and you don't know if it would've been worse without them.

Most of your life operates in this zone. Not control but not helplessness either. You have influence. The sooner you accept that most outcomes are probabilistic, the sooner you can stop demanding clean feedback and start playing the long game.

Agency

Agency is the belief that you have influence. It's the psychological layer on top of whatever influence you actually have.

Belief changes behavior. Take two people who get laid off from the same company on the same day. Same job market, same industry, roughly the same skills. One updates their resume the next morning, reaches out to ten contacts, starts interviewing within a week. The other sits at home for a month telling themselves the market is terrible and nobody's hiring. Same influence over what comes next. Different agency. The first person doesn't know their outreach will work. They just believe it might, so they act.

The person who acts isn't smarter or more connected. They just produce more attempts. More attempts means more chances for something to work. Agency doesn't change the odds on any single attempt. It changes how many attempts you make.

Well-calibrated agency looks boring. You do the things that tilt odds in your favor. You follow up. You prepare. You adjust based on feedback. No single action looks decisive, but the cumulative effect is large. You believe your actions matter, so you keep acting.

Poorly calibrated agency looks dramatic. You swing between "I can make anything happen" and "nothing I do matters." You bet everything on one big move instead of doing the ten small things that compound. When it doesn't work, you blame the system or double down on the wrong lever. The problem isn't your influence. It's your belief about your influence.

People with no agency are miserable and bitter. They complain about how unjust the world is. They only feel bad for themselves. Unsurprisingly, they never amount to much and are generally miserable to be around. Not because they lack ability. Because they've decided ability doesn't matter.

The opposite failure exists too. You can have lots of agency and very little actual influence. You believe you can change things, but the situation genuinely doesn't give you room. You keep pushing, keep trying different approaches, keep getting the same result. Without recalibrating, you burn out. Or worse, you blame yourself for outcomes you never had the influence to change.

Conclusion

Oddly, people tend to overestimate how much control they have but underestimate how much influence they have.

Overestimating control leads to frustration. You try to force outcomes that can't be forced. Underestimating influence leads to passivity. You stop trying because you've told yourself nothing you do matters.

Underestimating your influence is probably a defense mechanism. If you admit you have influence, you have to take responsibility when things don't work out. You can't blame the system or bad luck. The outcome is at least partly on you. That's uncomfortable, so most people opt out of believing it.

The trick is to hold a roughly accurate view of your agency. Overestimate it wildly and you'll burn out chasing outcomes you can't reach. Underestimate it and you won't try.

But if you had to be wrong in one direction, you'd want to be wrong on the side of overestimating. A little. Not so much you're fooling yourself. Just enough to keep you acting when a perfectly rational person might stop. The person who sends one more email, makes one more ask, runs one more experiment. They're probably overestimating their influence a little. And they tend to get more out of life than the person who sized things up accurately and decided it wasn't worth the effort.

If you had to pick a bias, pick that one.

Non-AI, AI Companies

Sat, 30 May 2026 00:00:00 GMT

For every sufficiently promising technology, companies will claim to be focused on it whether they are or not. It happened with internet companies, social companies, crypto companies. It's happening with AI right now. That isn't particularly interesting. What's interesting is AI is the first of these technologies to change how companies operate.

If you time travelled 26 years into the past (2000), to work at a company, you might snicker at how bad the software was and how slow the hardware was, but how you worked (e-mail, taking notes in meetings, etc.) and how the company operated (individual contributors, layers of management, etc.) would be very familiar.

I suspect if you time travelled 26 years into the future (2052), how you worked and how the company operated would be unrecognizable. A high resolution picture of what this will look like is anyone's guess, but my guess is the low resolution picture is smaller, flatter companies and many more of them.

I don't think you'd need to travel that far into the future to see it. Some companies are already operating this way.

This operational shift is already underway, and it's so meaningful it's worth arguing for a second dimension of what it means to be an "AI company":

The Product – Is AI the core of the product you're selling?
The Company – Are you using AI to meaningfully operate your business?

The Product

This dimension is the same as in previous technology cycles, so it's straightforward. You're either selling (1) the technology itself or (2) a product built on the technology. The model companies (Anthropic, OpenAI, DeepSeek) sell the technology. The application layer (Cursor, Harvey, Perplexity, Sierra) sells products built on it, with varying levels of depth and defensibility.

For the rest of the post, I'll primarily refer to products built on the technology. The number of companies that sell the technology itself (foundational models) is small, and few new ones get started at this stage.

The Company

This dimension is unique to AI. The metric for whether you're "using AI to meaningfully operate your business" is whether your ratio of output per employee (revenue, contracts reviewed, properties managed) is significantly higher than competitors in the same industry. Not 20% difference an order of magnitude different. The kind of difference that can only be achieved by being structurally different.

That doesn't happen by telling everyone at your company to use AI or buying them Claude licenses. Individual adoption is required but not sufficient. What generates a significantly higher difference is building organization-wide infrastructure: shared prompt libraries and agent configurations, business processes owned end-to-end by AI with human oversight, and workflows where AI is the default rather than the exception.

Here are a few recent examples.

Shopify – Shopify's CEO Tobi Lütke mandated "reflexive AI usage" as a baseline expectation. Teams must demonstrate why AI can't do the work before requesting additional headcount. AI proficiency is part of performance reviews.
Crescendo – Crescendo acquired BPO (business process outsourcing) firm PartnerHero and rebuilt operations around AI. They now automate up to 90% of frontline support tickets and run at 60%+ gross margins, roughly 4x a traditional BPO.
Anthropic – Anthropic shipped 120 features in 90 days in early 2026, roughly one every 18 hours. Their CPO confirmed that the vast majority of Claude Code is now written by Claude Code itself. The product builds the product.

Looking at these two dimensions together, and since they are independent, we get these four quadrants.

	Selling AI	Not Selling AI
AI Operated	True AI	Stealth AI
Traditional	Fake AI	Traditional

True AI (AI, AI)

These are companies that sell a product built on AI and use AI to meaningfully operate their business. When it's real, the output shows it: small teams shipping at a pace that would have required 10x the headcount five years ago.

But most companies who claim to be here aren't. For a surprising number of them, it's more aspirational than representative. More on this in the next section.

Fake AI (AI, Non-AI)

These are companies that sell a product built on AI but are NOT using AI to meaningfully operate their business.

These companies are problematic for a couple of reasons.

Incoherence – They believe in AI enough that it's the competitive advantage of their product but apparently not enough to use it meaningfully themselves? It's like a priest who doesn't practice.
Cursory Understanding – AI is moving so fast if you don't use it daily, in a meaningful way, you won't understand it well enough to build a company around it.
Bad Signal – AI is widely available to almost anyone. If you're an AI company (or any company) that doesn't use AI, it's because you're choosing not to. As a founder, you need every competitive advantage you can get. It's a bad sign if you're turning one down.

For most of these companies, nothing nefarious is going on. Often, the founders don't know enough about what's possible to realize how they use AI is cursory at best. When we talk to these founders, it's easy to tell. If they talk about using ChatGPT, we know they aren't that sophisticated. If they can't talk in detail about their agent harness, how they're managing shared context, etc. we know they can't be doing meaningful work with it.

If deep down, you think your company might be in this quadrant, start building things with AI, even if it's just for fun. The best way to understand the technology is to use it.

Traditional (Non-AI, Non-AI)

These are companies that do NOT sell a product built on AI and are NOT using AI to meaningfully operate their business.

Not much to say here except that the vast majority of businesses exist in this quadrant. Wherever you are right now, look at the businesses around you. They are probably all traditional businesses (restaurants, banks, etc.). If you're in tech, or read what companies get covered in the news, you mainly hear about tech companies and forget that almost all businesses are traditional businesses.

Stealth AI (Non-AI, AI)

These are companies that do NOT sell AI but are using AI to meaningfully operate their business.

True AI companies get most of the attention and rightfully so, but this quadrant has a lot of quiet potential. The ceiling on the value of these companies is an order of magnitude smaller than the ceiling of True AI companies (I don't think any of these companies can be worth the $1T Anthropic and OpenAI are being valued at) but my hunch is there's room for a lot more of them.

There are two ways to start these types of companies: (1) start the company like you would any other company and (2) start by buying an existing business that fits the target. The former is standard. The latter is interesting.

Buying businesses and using technology to make them run more efficiently isn't new. PE firms like Vista Equity Partners built $100B+ businesses doing software roll-ups: buying enterprise software companies, standardizing operations, and expanding margins. It works, but only when (1) the work was structured enough (data, workflow, etc.) for software to handle and (2) the opportunity was lucrative enough to build software for.

AI removes both constraints. AI handles unstructured work, and it's made software so cheap to build, you can apply it to a wider range of opportunity sizes.

A few recent examples.

Crosby – AI-first law firm backed by Sequoia and Bain, with a median contract review time of 58 minutes. They built from zero with AI as the default operating model.
Long Lake – Backed by General Catalyst, is rolling up HOA management companies: 18 acquisitions, $1.28B raised, $100M in EBITDA in under two years. They bought businesses with proven revenue and used AI to transform margins.
Crete Professionals Alliance – Backed by Thrive Capital and Bessemer, is rolling up U.S. accounting firms: 20+ acquisitions, $300M+ in annual revenue, with OpenAI-powered tools saving audit teams hundreds of hours per month.

Customers of these companies have no idea AI is involved. They shouldn't. What they notice is faster turnaround times, better service, and competitive rates.

The obvious counter-argument is if the moat is "we use AI to run operations better," what happens when every competitor has access to the same AI tools?

A few reasons why I'm not that worried about it.

Slow Adoption – Traditional companies are slow to adopt even when the competitive advantage is obvious. Think about how many companies still aren't fully using software (especially in Southeast Asia) 20 years after it became widely available. The tools aren't the bottleneck. The organizational capacity to use them is. More on this in the next section.
Effort to Implement – Making AI work in a specific industry isn't plug-and-play. It requires deep domain context, refined workflows, and operational knowledge that compounds. The moat isn't "we have AI." The moat is "we've spent years making AI work in this specific business."
Data Compounds – The longer you operate in a specific domain with AI, the more proprietary data you generate: edge cases, customer patterns, performance benchmarks. That data feeds back into the AI and makes it better. A new entrant can't shortcut years of accumulated data.

Why Traditional Companies Stay Put

Why don't Traditional companies just adopt AI themselves? The tools are available. The cost savings are clear. Four structural forces make internal transformation nearly impossible for most of these companies.

Risk Asymmetry – Nobody gets fired for doing things the way they've been done. People get fired for trying something new and having it fail. The rational career move is to wait.
Incentive Mismatch – The payoff of margin expansion accrues to the company. The risk, time, and political capital accrue to the person who pushed for it. The math doesn't work for the individual.
Authority Gap – Most people don't have the authority to push change of this magnitude, including most people near the top. The number of people who both understand AI and have the authority to implement it at scale is usually zero.
Security – Data, IP, and regulatory compliance are legitimate concerns. They're also the most convenient veto for people whose real objection is one of the first three.

These four forces don't just slow AI adoption. They explain why software didn't fully eat these industries either. The transformation, when it comes, mostly comes from outside.

Migration

The quadrants aren't static. Companies move, and in the cases of Traditional and Fake AI, they should. The interesting paths are all about getting out of those two quadrants.

Traditional → Stealth AI – The most valuable move in the framework. This happens internally if a traditional company has strong leadership and absolute belief in making the leap. If not, that's where the opportunity for acquisitions and roll-ups is. New ownership brings the authority, incentive alignment, and risk tolerance to change. It's also the path for greenfield builders who start an AI-native business in a Traditional industry from scratch.
Traditional → Fake AI – The temptation path. Bolt an AI label onto the marketing without changing operations. Easiest move to make, worst place to end up. You attract AI-level competitive scrutiny without the operational advantage to survive it.
Fake AI → True AI – This is easier than it sounds, which is what makes it frustrating that more companies don't do it. The tools are available. The playbooks exist. The barrier isn't technical. It's the same inertia and poor judgment that put you in the Fake AI quadrant in the first place.
Fake AI → Stealth AI – If you're honest about being in the Fake AI quadrant, this might be the better path. Drop the AI product and apply your operational AI knowledge to a traditional industry. The bar for Stealth AI is competence with AI, not breakthroughs. Most of the time, you don't need to be True AI, and the Stealth AI opportunity is much larger.

The Southeast Asia Opportunity

SEA has a lot of Traditional companies for two reasons.

Regional Fragmentation – Ten countries, dozens of languages, radically different regulatory and operating environments.
Lower Technological Sophistication – Most SEA businesses are small, owner-operated, and never had the scale pressure, capital, or vendor ecosystem to digitize. Many skipped the software era entirely.

These typically look like impediments to building startups in Southeast Asia, but for Stealth AI opportunities, they're advantages. Think about the companies that do things like property management, logistics, freight forwarding, accounting, BPO, customer support, F&B operations, healthcare administration.

These are the industries where the work is labor-intensive, the businesses are fragmented, and AI-enabled consolidation has the most room to run. The fragmentation that usually looks like a problem is the setup: the same AI-enabled operational playbook that consolidates a property management business in Jakarta works in Kuala Lumpur, Manila, and Ho Chi Minh City, just with different execution.

For founders in SEA, the question is which Traditional industries to move into the Stealth AI quadrant before someone else does.

Single User Software

Sat, 30 May 2026 00:00:00 GMT

No one loves their task manager.

Surprisingly, it's because software is expensive to make. For software to be economically viable, a lot of people need to buy it. To make something a lot of people need to buy, it needs to work for a lot of people. To make something work for a lot of people, you have to compromise on how it works.

That's why you end up with the same task manager used by freelancers for client work, engineers for sprints and students for homework assignments. The features that would make it great for freelancers (like per-client billing) are useless to the others. So they get cut. It reminds me of these composite images created by averaging thousands of faces. A face created from many but of no one in particular. That's your task manager.

Some products handle this compromise better than others. If the product has a universal task (search, maps, calculators) almost everyone wants the same experience. If you designed a calculator specifically for freelancers, engineers and students, they would all be the same.

If the product has a personal workflow (task manager, note taking, bug tracking), almost everyone wants a different experience. If you designed a task manager specifically for freelancers, engineers and students, they would all be different.

That's been the deal for software since it existed. Everyone got the average.

This has all changed in the last 6 months. AI coding tools (Claude Code, Codex, etc.) have gotten dramatically better. Companies are forcing engineers to adopt them. Non-engineers have started using them too. Creating software is now cheap.

Since it's cheap, it's now economically viable to build a product for a small group of people. That allows you to design specifically for how those few people work. Instead of a single task manager for freelancers, engineers and students, there can be one for freelancers, another for engineers and another for students.

Why stop there? If the cost of writing software approaches zero, it becomes economically viable to do it for a single person. If software is built just for you, there's no need to make design compromises for anyone else. It will fit your exact specifications. If you don't love that product, I don't know what to tell you.

Who would write software just for you? Well... you would. People who build software (designers, engineers, product managers) are already doing it but the tools are good enough for anyone to use. Automate your own workflows, build internal or personal tools you wish existed, make a tracker for whatever you're tracking now in a spreadsheet.

Every weekend, I build something new. This post is about task managers because that's what I was working on when I had the idea. It runs in the terminal, built for agents and has exactly the right level of structure (areas, initiatives, projects, tasks). It works exactly the way I want. If it doesn't, I change it so it does.

I love my task manager. You would probably hate it. That's the point.

The AI-Pilled Advantage

Thu, 16 Apr 2026 00:00:00 GMT

At the moment, being AI-pilled is the biggest competitive advantage a founder can have. Someone could be 2x better at making decisions about their startup but if you're competing against someone that can do 10x more, you lose almost every time.

The magnitude of this advantage is even larger in SEA than it is in SF. In SF, it's quickly become table stakes for founders but the adoption is much slower in SEA. That's bad for the community as a whole but fantastic if you're one of the few that's taking advantage.

When do you become AI-pilled? There's currently 5 levels of AI sophistication and, in my experience, it happens at Level 3.

Basic – You use an LLM (ChatGPT, Claude, etc.) as mostly a Google replacement. You ask for a bolognese recipe, what time it is in San Francisco, etc.
Context – You've picked a specific tool and are starting to give it context in the form of projects or memory. You're thinking in terms of a system rather than individual, single prompts.
Agentic Execution – You start using LLMs to do things for you, not just answer questions. Often it starts with automating simple workflows (read my e-mail and summarize) but eventually, it works on almost anything.
Agentic Engineering – You've automated simple workflows and now that you've had a taste of what they can do, you're drunk with power. You want to automate everything. Anything you can't automate (yet), you want to build something that augments yourself so you can do it faster and better. To do this, you need to start developing your agent harness. You give it more tool access (CLIs, MCPs, etc.), a memory and context management system (mempalace, Kaparthy LLM wiki, etc.), task orchestration (Linear), guardrails and oversight, etc.
Agentic Employees – So far, all of the above is used by a single person. For your startup to get maximum advantage, you need agents that work with your organization. They sit in Slack and have their own e-mail. They have their own tasks assigned to them, they write updates and take feedback. For all intents and purposes, they are agentic employees.

To give you a sense of how much we believe in this at Iterative, we're actively working on ways to evaluate people on which level they are (application only sophisticated agents can apply to?) and how to prioritize for them in our investment process.

I'll have to write another post on my experience going through these levels since December (I'm working on 5 now) but if we've been quiet, it's because we've been having passionate internal debates on how Iterative should change given these changes.

Should You Stop Working on Your Startup?

Thu, 06 Nov 2025 00:00:00 GMT

This is the toughest question founders face, and given the current funding environment, I get it a lot.

When I was a founder, I never had a good way to think about it. Due to sunk cost fallacy and survivorship bias, answering it for yourself is both emotional and intellectual.

Now that I primarily work with founders rather than being one myself, I have a clearer way to think about it.

If you're considering whether to stop working on your startup, follow this decision tree.

Am I growing quickly?
If not, am I learning quickly?
If not, do I have ideas for promising experiments?
If not, do I still have conviction?
If not, stop working on your startup

If the answer is yes to any of these questions, continue working on your startup but focus on that step.

Let's go through each in detail.

1. Growth

If you're growing quickly, keep working on your startup. There's a saying in sports that "winning cures everything." For startups, it's growth. But if you're growing quickly, you're probably not asking this question.

A quick note on how fast is fast. It depends on the stage, but for our program (and YC), the target is 5% to 7% weekly growth (not cumulative). If you're very early, grow faster. If you're later stage, you will probably grow slower.

2. Learning

If you're not growing quickly but you're learning quickly, keep working on your startup.

Typically, the learnings you're looking for are...

Target Users – What specific group of users (job industry, company size, etc.) strongly experience the problem you're solving?
Value Proposition – For that narrow set of users, what do they care about (saving $, making more $, etc.)?
Channel – For that narrow set of users, how do you reach them (e-mail, associations, etc.)?

If you're learning about your users quickly and turning those learnings into growth exploits, you should start growing quickly soon.

3. Experiments

If you're not growing or learning quickly but have many experiment ideas, keep working on your startup.

Promising experiments are typically...

Specific to Your Startup – They should be experiments uniquely relevant to your startup, resulting from an observation or insight about your users or product usage.
Narrowly Focused – You won't learn much from "Launch on Product Hunt" but from "Cold outreach to 20 accountants at firms with 5 to 20 employees on cost savings."
Well-Formed – Promising experiments consist of an action, expected outcome, and insight (often the hypothesis). They validate a hypothesis. A useful test for a well-formed experiment is to express it as "if we do [action] then [expected outcome] because [insight]."

Action generates information. Promising experiments generate learnings that lead to growth.

4. Conviction

If you're not growing or learning quickly, and short on promising experiment ideas, ask yourself if you still believe in what you're doing.

This is a challenging question because emotions and biases (sunk cost fallacy, confirmation bias, etc.) cloud your judgment.

To counteract this, I recommend a thought exercise. Imagine starting a new company and picking a problem to work on. Given what you know now and other problems, would you choose the problem you're currently working on?

If yes, I'd consider whether you're being honest with yourself.

If the answer is no, I'd sit with that for a few days. For every startup I've worked on, even the successful ones, my answer has been yes. It's normal and natural. As a founder, you'll go through periods of doubt about your startup for various reasons.

But when that feeling persists for months, take it seriously.

5. Stop Working on Your Startup

If you stop working on your startup, you've decided to (1) work on a new problem or (2) stop completely.

If it's (1), the journey continues and I hope you find renewed energy and conviction from working on something new.

If it's (2), I'd like to mention a few things.

Obligation – If you have investors, you might feel like you've let them down. You haven't. When Iterative invests in a founder, we expect you to put a good faith effort into your startup. You can't guarantee success, and I don't expect you to. I've been a founder long enough to know that's impossible. We're betting on your startup, just like you are, and when it doesn't work out, we're wrong too.
Failure – You might feel like a failure. Welcome to the club of everyone who's tried something new. I'll teach you the secret handshake next time I see you.
Rest – Take time off. The biggest mistake founders make after shutting down their startup is to start another one immediately. As a founder, the only time you really get to rest is in between startups, so please do it. The world will always have problems for you to solve. It needs you at your best.

When you're ready, apply to Iterative. We like founders who have failed before. As I often say, the best advice for first-time founders is to be a second-time founder.

Dogmatic vs Scientific Founders

Wed, 29 Oct 2025 00:00:00 GMT

All founders are optimists. If you don't think things can improve, you wouldn't build anything new.

Unfortunately, optimism is necessary but insufficient to be a successful founder. A founder’s success often depends on their source of optimism.

Dogmatic Founders

Dogmatic founders are optimistic in the way a priest is. They know they’re building something people want because they believe they are.

This is the strongest but least effective form of optimism. If your optimism is driven by a belief, it's inexhaustible but myopic. When faced with contradictory evidence, dogmatic founders double down rather than adapt. They see rejection of their product not as valuable feedback, but as a sign that users "don't understand it yet."

Signs you’re a dogmatic founder.

Confirmation Bias – They interpret feedback through their existing beliefs. Positive feedback confirms they're right. Negative feedback means users don't understand the vision. There's no situation where they're wrong.
Incremental Experiments – Convinced they know what users want, their experiments tend to be small and incremental.
Inefficient Learning Loop – They interpret everything through their existing beliefs, rarely gaining new insights. Even with clear evidence their approach isn't working, they attribute failure to external factors rather than their own assumptions.

Dogmatic founders become detached from reality, rarely adapt, and as a result, rarely succeed.

Scientific Founders

Scientific founders are optimistic in the way a scientist is. They have a hypothesis that they’re building something people want, but derive their optimism from (1) evidence supporting their hypothesis and (2) their ability to change it to fit the evidence.

Signs you’re a scientific founder.

Look for Negative Evidence – They seek evidence that proves them wrong. They are mindful of confirmation bias, so they investigate why people aren't using their product or what features users dislike.
Everything is an Experiment – They treat each feature, marketing campaign, and customer interaction as a learning opportunity. They form clear hypotheses and design quick tests. Most importantly, they're open to being wrong.
Optimize for Learning – Their primary goal isn't to be right, but to learn fast. They measure success not by how often they're right, but by how quickly they can improve the accuracy of mental model on what people want.

Scientific founders are constantly adapting and as a result, more likely to succeed.

Choose your religious beliefs however you’d like, but if you’re working on a startup, I recommend being a scientist, not a priest.

Why Office Hours Are Effective

Thu, 02 May 2024 00:00:00 GMT

At the end of every batch, we ask founders "What was the most valuable part of the program?" The answer is always Office Hours. If you've experienced it, it's obvious why but I wanted to see if I could explain it to people who haven't experienced it.

Prioritization

A common reason startups fail is they work on the wrong things. There’s so many things to work on, founders don’t spend enough time figuring out what they should work on.

I suspect this happens for a few reasons. First, it takes a surprising amount of work to figure out what to prioritize. Second, figuring out what to prioritize doesn’t feel like work in the same way as e-mailing a potential customer, changing the website, etc.

Office Hours helps with this by starting each week with “What’s your biggest growth blocker?” Often, the response is, “I don’t know” so we spend our time trying to find the answer. Working on your biggest growth blocker is almost always the thing you should work on.

Focus

Another common reason startups fail is they try to do too many things simultaneously. When a startup spreads itself thin, quality of execution decreases, and the likelihood of failure increases.

Worse still, if a project fails due to poor execution but is perceived as inherently flawed, founders will dismiss a viable opportunity. For a startup, where viable opportunities are rare, false negatives are killer.

I suspect the tendency to do too many things is related to being unable to prioritize effectively. Instead of figuring out which 3 of the 5 things they should work on, they try to do all 5.

Office Hours helps with this by giving founders feedback and a sounding board. It’s surprising how much clearer things can be to someone who’s been a founder but is not emotionally attached to the startup and involved on a daily basis.

Accountability

One of the first challenges I faced as a founder was the lack of a feedback loop. Without it, I’d find myself making excuses or ignoring when things didn’t go well. It’s like doing poorly on a test. Ideally, you would analyze each wrong answer to understand the mistakes and make a plan for improvement, but I don’t know anyone that actually did that. It’s hard to inspect your own failure.

Office Hours address this by establishing a weekly feedback mechanism for your startup. In the beginning, we work with you to set weekly targets. Each week after that, we ask you to show whether you’re above or below those targets. If you’re above target, we explore why, what you learned about your users and how to build on what’s working. If you’re below target, we discuss the reasons, what you learned about your users, and what changes to make so you will be above target the following week.

Ambition

Founders are often not as ambitious as they should be. This happens for a few reasons.

First, their ambition constrained by the ambition of those around them. If you meet someone with 10,000 users, that might seem ambitious. If you later meet someone with 10 million users, the 10,000 users doesn’t seem as ambitious anymore.

Second, the fear of failure causes founders to be less ambitious. You are less likely to fail if you aim for 10,000 users than 10 million.

Third, in Southeast Asia, proclaiming ambitious goals is culturally frowned upon. People often think you’re stupid or arrogant.

Office Hours addresses this by having founders work with more experienced founders who are more ambitious and successful than in their existing circles. This is, in large part, why people move to SF.

Empathy

Office Hours focused on growing your startup, but sometimes, the main blocker is founder morale. In those cases, the camaraderie of other founders, especially experienced founders is invaluable. It reminds you that you're not alone in your struggles, and what you're doing is hard. Some of the most rewarding Office Hours I've had were the founders opening up about struggling, me sharing a personal anecdote about being in a similar situation, having a laugh about it and telling them it'll be fine.

This is one of the reasons why only someone who has been a founder is allowed to be a Visiting Partner. People who haven't been founders will understand these challenges intellectually but because being a founder is such a unique experience, there is a different level of empathy that comes from someone who's been through it themselves.

Apply to Iterative. If you think Office Hours might be helpful to your startup, and you're interested in Iterative, you should apply by EOD tomorrow (Friday, 3 May). I'll be doing Pre-Batch Office Hours with everyone we invest in who applied before that deadline.

5 Most Common Startup Pitch Mistakes and What to Do

Thu, 25 Apr 2024 00:00:00 GMT

Iterative Winter 2024 is having its Demo Day on tomorrow (Wednesday, 8 May). To help companies prepare, we've hosted pitch practices every few days where companies pitch to other founders, Brian and I. They get immediate feedback, learn from each other's pitches, iterate over a few days and pitch again. You'd be surprised how much pitches improve by doing this.

Having now done this for 8 batches and 100+ companies, these are the 5 most common startup pitch mistakes and what to do.

Act 3 Problems Are Act 1 Problems

This is a phrase I got from fiction writers. It means if the ending of your story sucks, it's probably because the beginning sucks. Applied to startup pitches, the beginning of the pitch should always be centered around the problem your startup is solving. If you don't communicate that well (typically because it's too vague) then the rest of your pitch won't be very effective. The problem, and how you define it, is the context necessary for investors to understand whether your solution is significantly better, your market is sufficiently large, you're the right people to do it, etc.

What's Your Secret?

Everyone loves a good secret. At Iterative, we define secrets as something a startup (1) knows that very few people know or (2) believes that very few people believe. For Airbnb, it was that people were willing pay strangers to let them stay in their house. For Uber, it was that strangers would pay to have other strangers drive them around. Nobody believed it at the time but they turned out to be right and since they were right when everyone else was wrong, they become valuable companies. The most interesting part of the pitch is the secret. It's what makes investors sit up in their seat and pay attention. Remember that investors spend hours a day listening to pitches. If you say something they've never heard before and it's outlandish enough to be interesting but not so much so that it's stupid, you'll have their attention. You know you've got them when they say something like "Wait, that can't be right. Can it?"

Explain Your Growth

If you’re growing quickly, can you explain why? If you can’t explain why, it suggests that you don’t understand what’s happening so when the growth slows (it always does) then you won’t be able to start it again or build off of what’s happening. Similarly, if you’re growing slowly, you need to be able to explain why you’re growing slowly, what you’re going to do differently and why you think it’s going to change your trajectory.

Raising $X to Reach Y Milestone

Good investors don’t care about the % of the $ you’re fundraising is going to marketing, engineering, etc. What they care about is they’re going to give you $X of money and you’re going to turn that into some significant business milestone. We tell all of our companies to phrase their fundraising ask this way. We’re raising $1M to get to $2M in ARR. As an aside, you typically shouldn’t ask for more $ than the milestone. It’s weird to ask for $1M to get to $250K in ARR.

So What?

After a few of the pitches, I found myself thinking "So what?" The pitch was mechanically good. The delivery was good, the pitch was well structured, etc. but I just didn't care about the company or what they were doing. If you're trying to fundraise, that's bad.

When you hear the best founders pitch, it sounds like they're starting a rebellion. There is something wrong with the world and they're going to do something about it. The pitch feels more like a manifesto. A good way to build this into your pitch is to answer the question "If your startup is successful, how will the world have changed and why will we be better off?" If you can't answer that question, you probably (1) haven't thought hard enough about what you're doing.

Yes, investors want to invest in startups but what they really want is to invest in rebellions. Google was a rebellion against information freedom and the advertising industry. Web3 is a rebellion against the traditional financial system. The internet was a rebellion against physical borders. All big ideas are rebellions.

What to Look for in Your First Technical Hire

Mon, 05 Feb 2024 00:00:00 GMT

I’m looking for a technical co-founder or hiring my first engineer. Since I’m not technical, I don’t know what qualities to look for. Any advice?

There’s a big difference between a technical co-founder and first engineer but putting that aside, here’s a list of things to look for.

Generalist – A strong indicator of whether a startup will be successful is how quickly they can test and iterate on their ideas. Engineers who are good at many things but maybe not the best in any single thing tend to be better at this. They’re also more capable of adapting with your startup. You might be building a web app now but in 3 months you might find yourself primarily focused on iOS and Android. Ideally you can get an engineer who has range (tough but valuable) or is happy to learn (more likely and still valuable).
Thoughtful – Key technical decisions such as what language to use, how to design the database, etc. have significant, long lasting impacts on your startup. You want to find someone who is thoughtful and considerate about these decisions. Using an exciting but brand new language might be exciting for them but good luck trying to hire people who know it or want to work with it. Doubly so if that language suddenly goes out of fashion.
Communication – Related to being thoughtful, a bad engineer will make key decisions without considering the trade-offs, different stakeholders, etc. An okay engineer will consider these things but not communicate that they did and why they made the choice they did. A great engineer will consider them, layout the options, trade-offs, their recommendation and have a discussion about it. You might not be technical but great technical people can bring you into the discussion by explaining it in a way that you understand and from the perspective of the company.
Useful > Interesting – There’s a type of engineer that’s primarily concerned with either (1) working with the latest technologies or (2) only on problems that are technically interesting. Stay away from them. If you hire one of these engineers, they might do what's technically interesting instead of what's good for the product or company. Most software development isn’t technically interesting. Building an activity hook that triggers an email campaign isn’t technically interesting or particularly fun to build but it sure does boost retention. Find engineers who prioritize building something people want regardless of how interesting it was to build.
Management (Optional) – There's a type of first engineer or technical co-founder that's an incredible individual contributor. The type that can solo build the first version of the product in a few weeks. Unfortunately, they often are not interested in being a manager. That can work but at some point you'll have more engineers and will need someone that can manage them. If that's not your first technical hire, you're going to have to find someone else.

In addition, there’s a few common mistakes you should be aware of.

Large Company Engineers – Before I get a bunch of hate mail, this is obviously not true of every large company engineer but it’s happened enough to call out. Engineers from large companies, especially non-technology companies (banks, etc.) but even places like Google, tend to be specialist. That makes sense when you’re operating at that scale and complexity. Unfortunately, very little of that expertize is necessary at an early stage startup. Furthermore, there tends to be more culture shock to how quickly and chaotic startups are. Lots of great engineers at large companies, just be mindful and don’t assume that an engineer from XYZ large company is an automatic win.
Outsourcing – Outsourcing your initial MVP can work but there’s a couple of things to consider. First, it’s likely you’ll have to throw away everything they did at some point. To agency, your product is a temporary project. Second, at some point you’re going to want to build an in-house team. Again, your ability to iterate quickly is a core competency. You don’t want to outsource core competencies. Third, it’ll likely take more work than you’re thinking. You’ll need to stay on top of them, communicate clearly and repeatedly what you want, etc.

Hopefully that's helpful. I sometimes get the feeling that non-technical people think technical people are magicians or something. They do things they'll never understand. It's simply not true. Writing software is a skill like any other and any reasonably intelligent person can do it. There's no need to be intimidated by it or the people that do it. If you don't understand, ask them to explain it to you. If they can't, that says as much about them as it does about you.

In SEA, you have to do both! You need to convince them that they should work at a startup and that they should work at your startup. Sad.

Fortunately, I think the strategy is similar in both places. The key is to not compete on the things you can’t win and focus on the things you can.

Don’t Compete On..

Here are the things that, as a startup, you can’t compete in.

Salary – Your startup won’t be able to pay as much as the larger companies. Frankly, you shouldn’t even try.
Stability – Your startup will never have the stability and security of larger companies. That’s the nature of a startup.
Prestige – Your startup will never have the prestige (recognizable name, fancy office, etc.) of larger companies.

Compete on..

Here are the things that, as a startup, you should compete on.

Growth – Your startup will give people an opportunity to grow and learn at a significantly faster rate. At a large company, they’ll be responsible for a small part of a big thing. At a startup, they’ll be responsible for a big part of a small thing. Turns out you learn much more in the latter than the former and if that small things becomes a big thing, your learning and career becomes a rocket ship.
Financial Upside – If your startup becomes a large company, they will make a life changing amount of money. Where they will likely never have to think about money again. That's rare and takes a long time at a large company. It's also rare at a startup but can happen very quickly. This is admittedly a tougher sale in Southeast Asia so we spend a considerable amount of time educating people on how to think about this. That includes explaining how equity works and making a spreadsheet for their equity so they can see how much it could be worth.
Meaningful Experience – “It’s a job” is something I hear a lot from people at large companies and almost never from people who work at a startup.
Your Scenario Is The Worst Care Scenario – If someone is good at what they do (you should only be hiring these types of people) and working at a large company, they can likely get that job or an equivalent job at any time. If that's true then their current situation is their worst case scenario. Even if your startup is a complete bust, they can always get their job back. If that's the case, why not go for it? I find this works well on the right type of people.

An interesting byproduct of framing your startup this way is it’s compelling to the type of people that you want to hire and not to the ones you don’t. Effectively, it’s a filter. I know there’s a tendency to want to convince everyone to work at your startup but if you have to twist their arm to come work with you, they’re probably not the type of person you want. Even if you do convince them, they’ll probably leave at the first sign of trouble.

The most effective way to convince people to work at your startup is through growth and opportunity. If that’s not appealing to them, they shouldn’t work at a startup anyway.

Is Your Startup Solving an Important Problem?

Tue, 18 Jul 2023 00:00:00 GMT

A startup exists to solve a problem. The problem the startup is trying to solve is the most important thing about the startup. It dictates the market size, the product, the type of people the startup needs, etc. Therefore, for a startup to succeed, it must solve an important problem.

How to Tell if a Problem Is Important?

A simple but effective test of whether a startup is solving an important problem is to ask these 2 questions.

How many people have the problem?
How painful is it for each individual person?

If we then enumerate the possible configurations, and give them labels for easy reference, we get the following..

Type 1: Many people have the problem and the pain for each is acute
Type 2: Many people have the problem and the pain for each is mildly annoying
Type 3: Few people have the problem and the pain for each is acute
Type 4: Few people have the problem and the pain for each is mildly annoying

In table form it's..

	Many people have the problem	Few people have the problem
Pain is acute	Type 1	Type 3
Pain is mildly annoying	Type 2	Type 4

Obviously the most important problems are Type 1 problems and the least important problems are Type 4 problems. Unfortunately, Type 1 problems are exceedingly rare and Type 4 problems are abundant.

What's not obvious is whether Type 2 or Type 3 problems are more valuable. Personally, I'd rather be working on a Type 3 problem than a Type 2 problem for a few reasons. First, it's easier to solve a problem for a small group of people who care about it than for a large group of people who don't really care. The latter doesn't care about the problem enough to talk to you, haven't thought about it that deeply so their feedback isn't as useful, etc. Second, it's easier to find more people who care about something than to convince people they should care about something more.

What Type of Companies Typically Work on Which Type of Problem?

All of the huge tech companies (Google, Apple, Amazon, etc.) work on Type 1 problems. What's interesting is it wasn't obvious to them (or others) that they were working on Type 1 problems when they started. When Google and Amazon started, it wasn't obvious that the internet would be used by everyone. When Apple started, it wasn't obvious that everyone needed a computer. You could argue that it wasn't until Apple started making phones that they were working on a Type 1 problem.

Type 1 problems almost never look that way when people start working on them. Instead, they look like a Type 2 problem that turns out to be more painful than people thought or a Type 3 problem that turns out to affect more people than people thought.

Type 2 problems are what most consumer startups work on, hoping they turn into Type 1 problems.

Type 3 problems are what most B2B companies work on. Almost by definition, if you're solving a problem for a type of business, you're not solving a problem for a lot of people. That's why the average person hasn't heard of even the most valuable B2B companies.

No startup thinks they are working on a Type 4 problem but most startups die because it turns out they are.

What Iterative Looks for When Investing in Founders

Thu, 25 May 2023 00:00:00 GMT

Maybe the most common question I get is what do we look for when deciding to invest in startups? The answer is surprisingly simple at a high level but exceedingly nuanced in practice. At a high level, we (and I suspect every investor) want to invest (1) in a startup working on an important problem and (2) in founders who can solve that problem. The question then becomes how do you assess each of those things? This is the first part of a two-part series answering that. I'll start by discussing what we look for in founders.

1. Does What They Say, Consistently

Often people will call this execution but I find that term too imprecise. What does it mean to be great at execution? My definition is simple: you are great at execution if you consistently do what you say you're going to do. If you say you're going to launch X feature by next week, you launch X next week. You say you're going to grow 10x in the next 3 months, you grow 10x in 3 months.

I can't overemphasize the consistent part of this. Lots of people can do what they say once in a while or for a short period of time. Very few people can do it consistently over years. I could, and probably should, write an entire post on how to do this but I'll save the suspense. There's no secret trick. People who do what they say consistently aren't better at setting expectations, they don't encounter fewer problems than other people, etc. They just get it done. They think carefully before they make a commitment, have a good plan to achieve it and when they run into problems, they do the math and find a solution.

2. Convincing

Most founders become founders because they want to build something or solve a problem. What they quickly realize is, to build or solve that problem, they spend most of their time convincing various people of various things. They need to convince users to use a product they've never heard of before. They need to convince people to leave their cushy jobs to join their wildly risky startup. They need to convince investors to invest in their startup that does something nobody has ever done before.

If a founder is not convincing, it will be almost impossible to build or solve the problem they're working on.

Note: I chose the word convincing, not charismatic. Most people think founders need to be charismatic but I disagree. Being charismatic is one way to be convincing but in my opinion, not the most effective. For example, doing what you say you're going to do is a far more effective way to be convincing.

3. Internally Motivated

We believe founders who are internally motivated (care about solving the problem, opportunity to learn, etc.) are more resilient than those who are externally motivated ($, being famous, etc.). Why? Being internally motivated makes you result independent.

The process of building a startup is tumultuous. The lows are lower than the highs are high and you spend considerably more time feeling like your startup is going to fail than succeed. That also means if you're externally motivated you're going to spend more time feeling like you're not going to get what you want out of starting a startup. You're going to quit.

If you're internally motivated, you won't be bothered too much. If you started a startup to solve a problem or the opportunity to learn, those things will always be true, regardless of how it's going, as long as you don't quit. When things get really bad, and they're going to get really bad, that's what you're going to need to keep going.

Those are the 3 things we look for in founders. These qualities are required but not sufficient to be successful. It's not uncommon for someone who consistently does what they say, is convincing, and is internally motivated to still fail. Almost always it's because of the choice of problem they decided to work on. That's what we'll cover in the next post.

The Difference Between Bad, Good, and Great Startup Engineers

Fri, 28 Apr 2023 00:00:00 GMT

Given something that needs to be built.

A bad startup engineer will immediately start to build it. There’s very little consideration for how it should be built, ways it could be built and what the trade-offs are. You say make a button that when pressed does X. They make a button that when pressed does X. No questions asked. The project will get done in a week but it’ll be buggy. You will probably spend the next 3 weeks fixing bugs. In 3 months, they will tell you that they need to re-write the entire thing. At some point, another engineer will tell you a non-trivial amount of the code they wrote is copy and pasted from Stack Overflow. Your very promising startup will feel like it’s stuck in the mud. Constantly slipping and falling over itself.

A good startup engineer will consider how to build it but only in the most elegant, scalable, highest quality way possible. It will be a work of art but it will take a month to build and be complete overkill. Most recent Computer Science graduates are like this. They learned all the theory and can’t wait to put all of it to use. And by all, I mean all of it. Your very promising startup will feel like sitting in a car with a Grab driver who drives so slowly that you wonder if you’ll get pulled over. When merging, they wait until there’s no other visible cars on the road (maybe the planet) before merging. You will get there, and you will get there safely but by the time you do, dinner will be over and your friends long gone.

A great startup engineer will ask for context. Why are we building this? What are we trying to accomplish by building it? How do we know if it’s successful? After getting the appropriate amount of context, they will go off for a few hours and come back with options on how to build it. Typically those options will include the following..

Fast But Low Quality – If we cut corners, we can get it done in 3 days with minimal bugs but it’s not that extensible so if X happens, we’ll have to re-write it but I don’t expect X to happen for another 3 months.
Slow But High Quality – If we do it the “right” way, it’ll take us 3 weeks but we’ll never have to re-write it again. It’ll be extensible and scalable from the beginning.
Moderate and Moderate Quality – If we cut the right corners, we can get it done in a week. It’s relatively extensible and scalable. We might have to make improvements at some point but that shouldn’t be for at least a year.

These are the best engineers and it’s not close. I tear up a little bit thinking about them. Your very promising startup will feel like it’s being led by an experienced, capable guide in search of treasure. You still have to figure out where to go, what you’re looking for, etc. but there’s someone there who knows the terrain, knows how much food to pack and where the dangers are. With them, you feel like anything is possible and start dreaming even bigger.

Why are engineers like this so valuable to a startup? At a large company, the product has been mostly figured out. Engineers are mostly focused on scale and extensibility. At a startup, the product has not been figured out. Frankly, you don’t know if the product you’re working on now will be the product you’re working on in 3 months. 80% of the code you write will be thrown away. Not because it’s bad but simply because you’re trying to figure out what your users want. Startup engineers will build features one week only to remove them 2 weeks later.

In that type of environment, the most important thing for the startup, and by extensions the engineers at the startup, is iteration speed. Sometimes people interpret that as engineers who get things done through hacky code and lots of bugs. In fact, that drastically decreases your iteration speed. You end up spending so much time fixing bugs and it’s hard to tell if people want what you built because it’s so buggy.

To increase iteration speed, a startup needs to find the right balance between speed and quality of code. It’s knowing which corners to cut.

Should You Continue Working on Your Startup?

Mon, 06 Mar 2023 00:00:00 GMT

[Update] I wrote a new post on this topic here.

A persistent question founders ask themselves and probably the most difficult to answer. Here are the 3 questions I ask myself when trying to answer this question and what to do about it.

Is it growing fast?

Growing fast is the best indicator you might be onto something. How fast is fast is different depending on stage but our general guideline is 5% to 7% weekly growth. That's 12.6x to 33x annually. If you're just starting, it should be much faster.

Am I learning about my users?

You should be learning new things about your users every week. It should feel like every week you're getting a slightly higher resolution picture of who they are, what their problem is and what they care about.

Do I have lots of ideas for what to do next?

In the beginning, you have lots of ideas. As you work on them, you use up those ideas. When I'm onto something, I find I generate new ideas faster than I'm using them up. It's like being idea profitable.

If you are growing fast, learning about your users and idea profitable, you should obviously keep working on your startup. The reality is that's rarely the case so I thought it'd be interesting to look at the different permutations and what you should do (see graphic).

The way to think about this holistically is to remember that user learnings produce ideas and ideas generate growth. If you don't have growth, you need ideas. If you don't have ideas, you need to user learnings. If you don't have user learnings, talk to your users.

Does Story or Traction Matter More When Fundraising?

Thu, 11 Aug 2022 00:00:00 GMT

Fundraising is one of the most difficult things a founder has to do. And I suspect, given the current market conditions, it’s even more difficult now than in the previous seven to 10 years.

A common question we get is: “Which is more important: strong traction or a strong story?” But what is strong traction and a strong story — and why are they important?

Traction

A startup typically has strong traction if:

It is growing 5% to 7% weekly
40% of users are still using the product after six months

I say typically because this is greatly dependent on the company’s stage and industry.

Growth should be much faster for startups in the very early stages but slower for those in the series A phase and beyond. Similarly, consumer social products tend to have lower retention than B2B services.

Story

Meanwhile, a startup has a strong story if it has a compelling thesis about how the world should be and how it plans to get there. Typically, they start with a familiar observation, followed by something surprising (a secret), and end with a strategy that takes you from the familiar to the surprising.

Let’s use Airbnb as an example. The company observed two things:

During peak times, hotels are packed and become very expensive.
People want a more authentic, unique experience when traveling.

Its secret was..

People have plenty of room in their homes.
They’re willing to rent it out on a short-term basis.

Its strategy was..

Focus on large cities especially when there are events.
Target young people who are most cost-conscious and more willing to stay in other people’s homes.
Siphon traffic from Craigslist by posting its own listings.

Secrets are interesting and worth more discussion. By definition, these are things only a few people know or believe. Usually, these remain secrets because other people don’t know them, they are counterintuitive, or they are against conventional wisdom.

Airbnb’s secret was counterintuitive. Uber’s secret – that people would get in cars with strangers – is against conventional wisdom. Parents often tell their children not to get into cars with strangers.

Cases

Now that we know more about these two dimensions, let’s enumerate the different ways startups can manifest them.

Strong Traction + Strong Story

If you’ve got solid numbers and a compelling story, you’ll likely raise funds easily and at a high valuation (message me so I can invest).

An example of an early-stage company with both strong traction and a strong story is PayPal. Founded in 1998 and launched in 1999, the payments giant famously grew 10% daily (traction) through a referral scheme that was a huge success, even though it almost bankrupted the company.

PayPal was also one of the first internet payment products at the height of the internet bubble (story).

Weak Traction + Weak Story

If your startup is in this territory, fundraising will be almost impossible.

It’s hard to come up with examples because most of these companies don’t make it. However, I’d argue that Airbnb was like this in the early days.

In the beginning, the platform was only available for high-profile events. The first time it opened bookings, it had two guests stay for a design conference in San Francisco. The second time, nobody stayed. The third time was for the SXSW conference, and the only guests who booked were the founders.

That’s definitely not strong traction. At the time, the company was also still called Air Bed and Breakfast, and investors couldn’t get past the idea that people would want to sleep on air mattresses in other people’s homes.

Airbnb obviously gained traction and improved its story, but in those early days, it wasn’t an obviously promising startup.

Strong Traction + Weak Story

Now if you’ve got the numbers but have a weak story, you might be able to fundraise. The thing is, you might raise at a low valuation.

A weak story means investors won’t think your company can be that big, so if they invest (most won’t) it will be at a low valuation.

Direct-to-consumer companies often fall into this category. The good ones are able to build a real business with reasonable traction, but investors will always ask whether they’re scalable and if the market is big enough.

Weak Traction + Strong Story

If your company falls into this category, then it depends what stage you’re in. If the startup is at pre-seed or seed, you’ll probably be able to fundraise. If you’re at series A or above, you probably won’t.

Investors give you money in the early rounds to test your thesis. In the later stages, they need to see that your model works. Web3 companies often fall into this category. Many haven’t launched yet (no traction) or have very few users (weak traction), but if their thesis is correct, then they have the potential to be significant.

Observations

Looking at the outcomes, it is surprisingly more advantageous, especially early on, to have a strong story than strong traction – particularly in Silicon Valley and if you operate in frothy markets. The builder in me hates this, but in my experience, this does seem to be true.

Investors invest in the future, not the present. Traction is about the present and, at best, is evidence that your story is likely true. But if that story isn’t exciting or valuable, then it doesn’t really matter. After all, not all evidence is equally valuable.

With that being said, stories are fluid. It’s easier to turn a weak story into a strong one than to turn weak figures around. Often, companies with strong traction and a weak story either 1.) don’t understand why people are using their product or 2.) aren’t being imaginative enough. Having traction is incredibly important, but it’s not sufficient.

To be clear, I’m looking at this from the perspective of fundraising. Being able to fundraise successfully doesn’t mean your startup will be successful. As a founder, your job is to solve problems for your users. Fundraising is a way to get more resources so you can solve these problems better or for more people.

Founders often feel like their pitch is a work of fiction, and it makes them uncomfortable. I like to remind them that their job, as founders, is to turn that fiction into reality.

Learnings Compound, Money Doesn't

Sat, 06 Aug 2022 00:00:00 GMT

Things that compound are significantly more valuable than things that don't.

For startups, when you learn something about your users, that learning helps you every time you design a new feature, run a new marketing campaign, etc. for the lifetime of the company.

For startups, $ does not compound. In fact, it loses value over time because of inflation. So if you raise $1M and it's sitting in the bank, it's declining in value.

What should you do? The obvious answer is trade $ for learnings. Trade something that does not compound for something that does.

As examples, here are 2 ways you can make this trade.

Run Experiments – If you have 3 value propositions, and you're not sure which of them is more important, I'll run an ad for each. The results will tell me which is more important and to who. The demographic data with paid ads is great for figuring out personas.
Hire – There's a lot of nuance with this but generally, if you hire the right people, you'll be able to iterate faster. If you iterate faster, you learn faster.

If that's the case, why not trade all your $ for learning right now? It turns out not all learnings are of equal value. I've talked about learnings as if they were disconnected, singular things. In reality, they are deeply interconnected and form a structure. Learnings that form the base of that structure (fundamental) are much more valuable than those at the periphery.

Learning if your target users are teenagers or professionals will fundamentally change what you do. That's how you can tell if something is a fundamental (and therefore valuable) learning. Does learning it fundamentally change what you do?

Given the broader market environment, everyone is being told to cut cost. Generally good advice but there's nuance. I'd still be spending to learn, especially if it's to learn fundamental things.

Founder Simulations

Wed, 06 Jul 2022 00:00:00 GMT

Great founders have this uncanny ability to run highly accurate, localized simulations about their business in their heads. What's surprising is none of them could do this when they started.

What's the benefit of being able to mentally run accurate simulations? You can predict the future. If you want to grow by 2x in the next month, you can come up with ideas on how to do it, run it through your mental simulator and know what will and won't work.

Nobody starts with this ability. In the beginning, everyone's mental simulator is laughably bad. Not only is it often wrong, it often doesn't yield a result at all. Should I do X? No idea.

A primary difference between good, okay and bad founders is how quickly they are able to improve their mental simulators.

Bad founders don't realize there's something to improve
Okay founders struggle to improve
Great founders are effective at improving

The trick is self-directed learning. A good founder will take a step back, analyse the mental model the simulator is built on, diagnose what they need to learn then devise and execute on a plan to learn those things.

Imagine a founder doing sales but is unable to close a customer.

An okay founder will try random things (different prices, customers, etc.) hoping they stumble into a solution. Maybe they get lucky but it's wasteful and there's no guarantee they'll get lucky next time.

A great founder will identify the potential causes, prioritize those they think are most likely then devise a plan to systematically test. There's no luck involved, it's an inevitable march to a solution.

Because the good founder is systematically testing, they learn more, faster. Therefore their mental model is improving faster and so is the accuracy of their mental simulator.

At some point, you too can evaluate the 14,600,005 possible outcomes to find the 1 way to beat Thanos.

Increasing a Startup's Chances of Being Successful

Wed, 06 Jul 2022 00:00:00 GMT

For a startup to win, it needs a sufficient number of successes along the way. The obvious question then becomes: how does a startup generate the maximum number of successes?

Let's define a success as anything that when accomplished, moves your startup closer to winning. It could be improving a feature that increases retention, getting a better understanding of a user segment, finding a new growth channel and so on.

# of Successes = Probability of a Success * # of Experiments
This formula is obvious but people rarely think about it. The higher your probability of success, the more successes you will have. The more times you try, the more successes you will have.

↑ Probability of Success
If your probability of success is low, you probably don't understand your users. If you did, it would be higher. Use metrics to identify what is and is not working. Talk to users to figure out why. Make changes and validate the results with metrics.

↑ # of Experiments
You can increase experiments by (1) hiring more people or (2) decreasing the cost of each trial so you can do more. A trick with (2) is finding low cost ways to achieve the same effect. This is highly contextual so here's an example:

↓ Cost of Experiment

A founder, who wanted to test if ads would be a viable way to monetize, asked for advice on getting advertisers. I told them to run the ads without their permission. They get the data they want without a sales process or having to build out an ads product.

More Experiments → Better Understanding of Users → Higher Probability of Success

More experiments lead to a better understanding of users, which results in a higher probability of success. Increasing experiments is key, especially for early stage startups.

How to Think About Product

Fri, 06 May 2022 00:00:00 GMT

For many, being a founder is the first time they've had to think about product strategy.

Here's how I think about it.

All products can be broken down into growth, activation, retention and monetization.

Growth – Are more and more people using your product? It's an indicator of whether there's a lot of people who want what you're building.
Activation – What % of users have actually tried the product? A person might have signed up for Grab but hasn't tried Grab unless they've taken a ride. It's an indicator of how effective your onboarding experience is.
Retention – What % of users still use the product after an extended period of time? An indicator of whether people like or find your product useful.
Monetization – How much profit can be generated by the product? An indicator of whether you can build a business around the product.

If more and more people (growth) are trying your product (activation), continually using it (retention) and you can generate profit (monetization) then you're onto something. Your job as a founder is to work on the pieces so this becomes true.

A few things to keep in mind.

Interconnected System – Products are an interconnected system composed of these 4 pieces. The performance of each piece impacts the performance of the others.
Higher Activation → Higher Retention – It's impossible to retain users who haven't been activated. Therefore by improving activation you often see an increase in retention.
Higher Retention → Faster Growth – This happens for 2 reasons. First, if you're losing more users then you're gaining, your active user base is shrinking. Second, if retention is high then people like your product and are more likely to tell others.
Faster Growth → Lower Retention – Often to grow faster, you target a new user segment that you might not understand as well or use your product differently. They might not like your product as much (it was originally designed for a different segment) so retention drops.

Your product strategy starts by stack ranking how important each is to the current state of your product. Is it more important to work on retention or monetization right now? This is my general rule of thumb.

Early Stage Startups – Retention > Activation > Growth >> Monetization. Focus on building something people want. That means working on retention and activation.
Seed Stage Startups – Growth > Retention > Activation > Monetization. Retention and activation should be high. Now focus on building something a LOT of people want.
Later Stage Startups – Monetization > Growth > Retention > Activation. Growth, retention and activation should be high. Now focus on building a business.

Think about the progression as prerequisites. If high retention and activation, then work on growth. If high retention, activation, growth then work on monetization. If at any point one dips, go back and work on it before progressing.

Invest in Slope

Sat, 13 Nov 2021 00:00:00 GMT

A benefit of studying mathematics is that you learn how to describe things and their relationships. For example, if you want to describe a line, you can with the equation y = mx + b, where b is the y-intercept and m is the slope.

I know what you're thinking, that's cool but other than passing an elementary school math test, how does knowing something like that help you? It turns out that knowing how to describe a line carries over to things that resemble a line. Like startups.

An obvious but overlooked difference between founders and investors is, founders are concerned with how to make their startup successful, investors are concerned with how to tell if someone else's startup will be successful. Having spent most of my time as the former but now the latter, I find that thinking of each startup as a line is a helpful mental model for assessing whether they will be successful. The first time we meet with a startup is x = 0, think of x as representing time, and their traction at that time is the y-intercept. The interesting question is what's their slope? Not an easy question to answer. The entire venture capital industry is dedicated to answering this one question.

Before attempting to answer it, let's talk about why it's the most important question.

Take the following 2 startups:

Startup A – 10,000 users at Week 0 (y-intercept) and is adding 100 users every week (slope)
Startup B – 1,000 users at Week 0 (y-intercept) and is adding 300 users every week (slope)

Plotting the lines for both yields this plot.

Week 0 – Startup A is 100 times larger than Startup B
Week 52 – Startup B surpasses Startup A in size
Week 104 – Startup B is 1.57 times larger than Startup A

Note: There's another type of startup we like even more. We'll discuss them at the end.

Back to the question of estimating the slope of an early-stage startup. It's a function of (1) the founders and (2) the market. Looking at it this way yields the following:

	Bad Founder	Good Founder
GOOD MARKET	Small Slope	Large Slope
BAD MARKET	Negative Slope	Medium Slope

Good founders in good markets are obviously good. Bad founders in bad markets are obviously bad. What isn't obvious is whether good founders in bad markets or bad founders in good markets are better? I'd argue good founders, bad markets > bad founders, good markets because good founders can find their way into a better market[^This is more true for early stage companies, maybe less true for later stage companies.]. Better founders pulled into a good market will likely outcompete the bad ones. As an investor, you hopefully invest as much as you can in good founders in good markets, some good founders in bad markets, a very few bad founders in good markets, and no bad founders in bad markets.

Founders

If you take the above to be true, this yields an interesting result. There's just one rule, invest in good founders.

How can you tell if the founders are good? In our experience, successful founders possess accurate mental models of how the world, as it relates to their business, works. Given a desired outcome (ie. 2x growth), they can run a simulation in their head, based on their mental model, and figure out what they need to do to produce it.

The secret is, no founders start this way. The world is too complicated. The difference between great founders and everyone else is how quickly they can update their existing model to be accurate. The best founders I know do this by (1) being deliberate and (2) iterating quickly.

By deliberate, I mean they have a clear articulation of their mental model, where it's weak and a plan for improving those parts. To illustrate, let's say someone was starting a ride-sharing company before they were ubiquitous. A version of the following would have been very impressive.

We believe 10x more people would use a taxi service 100x more often if (1) you could call them to you, (2) arrived in under 5 minutes and (3) was affordable. We believe (1) is a magical experience that's significantly better than existing taxis. We're the most worried about (3) and (2), in that order of priority. We're going to test (3) by subsidizing the cost for riders to measure the elasticity of price on usage. After that, we will test (2) by increasing drivers bonuses in a specific location to increase driver density. By increasing driver density, we can drive down the wait time to under 5 minutes and study the residual impact on rider satisfaction and retention.

Let's break this down. The first few sentences are a clear articulation of what they think will happen if they are successful and what 3 things they need to do to be successful. We can debate whether they are right but they have a clear position. They then go on to talk about which of the 3 things they are most worried about, and how they are going to figure them out.

By iterate fast, I mean they are updating their model daily or weekly, not monthly or quarterly. Remember that doubling your iteration speed is equivalent to being right twice as often. In our experience, it's much easier to go from monthly to weekly iteration than it is to be right 4 times as often.

Said another way, consider the following 2 founders:

Founder A – Runs experiments weekly and is right 10% of the time
Founder B – Runs experiments monthly and is right 20% of the time

On average, Founder A will have their first success in 10 trials (1/.10) or 10 weeks. On average, Founder B will have their first success in only 5 trials but, since they run experiments monthly, that's 5 months or 20 weeks. It pays to be fast.

Market

There are entire books written on what makes a market good and how to identify them but simplistically, I think about it along 2 axes: (1) how many people do you help? and (2) how much do you help each person? Thinking about it this way, yields the following:

	A little	A lot
MANY PEOPLE	Good	Best
FEW PEOPLE	Bad	Great

Products that help many people a lot are obviously great. Products that help a few people a little are obviously bad. You could argue that products that help a lot of people a little are better than products that help a few people a lot but in my experience the reverse is true. Practically, if you have a small number of people who care a lot about what you're doing, it's easier to build a product because you're building for a narrow set of use cases and your customers are by definition more likely to give you feedback. It's easier to monetize because they care a lot about it. It's easier to find more people who care a lot about something than it is to get an existing group of people to care more about something.

Non-Linear

So far, the lines we've been talking about are called linear. In mathematics, there are also non-linear lines where the slope is not constant. They can increase and decrease as they go along. They are a little more complicated but a more accurate representation of a startup's trajectory.

For example, say a startup announces a new round of funds and receives favorable press. Their slope will increase sharply for a short period of time but eventually revert to where it was originally.

The thing to remember about non-linear lines is they can grow significantly faster than linear lines. This is especially true if their rate of growth is accelerating[^For the really big math nerds, this is the slope of the slope or the derivative of the slope.]. Let's take another look at the two startups we mentioned at the beginning and a third startup whose trajectory is non-linear and rate of growth is accelerating.

Startup A has 1,000 users at Week 0 (y-intercept) and is adding 10 users every week (slope)
Startup B has 10 users at Week 0 (y-intercept) and is adding 30 users every week (slope)
Startup C has 10 users at Week 0 (y-intercept) and is growing 7% week over week (rate of change of slope)

Given these growth rates, here's how big each of these startups will be in the coming years.

Startup C will be orders of magnitude larger than Startup A or Startup B. Remember, if you're growing by the same number of users every week, your growth is actually slowing as a % of your user base (Startup A and Startup B). If you can maintain growth as a % of your user base, you're growing exponentially (Startup C). That's why success in our program is growing 5% to 7% weekly for the duration of the program. If they grow 7% for the duration of the program (12 weeks), they will be 2.25x larger. If they grow 7% for an entire year, they will be 33.7x larger.

Growing non-linearly is hard but not complicated[^This assumes you are sufficiently solving a real problem. Knowing whether you are or are not is a blog post all in itself.]. Think about your growth as the result of a set of growth initiatives. Every week, each of those initiatives will generate some amount of growth (users, revenue, etc.). When an initiative stops generating growth, you need to find another initiative to make up for the loss in production, otherwise your overall growth rate will decrease.

Given this, a good way to grow non-linearly is to work on growth initiatives that are (1) repeatable and (2) have low marginal cost. These types of initiatives generate growth each week with very little additional effort. That gives you time to work on other growth initiatives that will add to your overall growth. Content marketing and SEO are both examples of this but require significant effort before they reach this state. For some companies (B2B, D2C, some marketplaces), paid marketing might fit these criteria if you think of marginal cost as time instead of $ cost.

The best way to grow non-linearly is to work on growth initiatives that are (1) repeatable, (2) have low marginal cost and (3) compound. Initiatives that compound (ie. their growth generates more growth) generate an order of magnitude more growth than initiatives that don't. These are rare, though. In my experience, they are either product specific or referrals. When people talk about viral products, what they're really talking about are products with a high rate of referrals. Referrals are like growth recursion. Every referral has a chance to refer someone else and so on. The higher your rate of referrals, the more generations (and therefore users) are brought into your app.

In summary, if you're an investor, invest in slope especially if it's non-linear. If you're a founder, work on generating non-linear growth. If you can, you're probably onto something.

Thanks to Brian Ma, En Lerk Law, Marx Low, Yung Tyng Lee and Rachel Ng for reading drafts of this.

Starting Iterative

Fri, 27 Mar 2020 00:00:00 GMT

Brian, Hsu Han, and I are starting a YC-style accelerator called Iterative, focused exclusively on Southeast Asia. The first batch starts on 1 June 2020. If you’re working on something, you should apply.

Here’s how it works:

Invest $150K Upon Admission – Iterative will invest USD $150K upon admission.
Work Together for 3 Months – Startups will move to Singapore for 3 months to work with us intensively on their idea (growth, product, pitch, etc.).
Demo Day in Singapore and San Francisco – At the end of the 3 months, we’ll host a Demo Day with investors in Singapore and San Francisco.

Our goal is to actively guide founders to achieve their startups next major milestone, which is typically a raising their first or second round of funding. We’ve found that founders in Southeast Asia struggle with this today, not due to a lack of capital or hustle, but because of limited exposure to experienced operators. There are simply not enough experienced operators in the ecosystem yet, and those who are experienced are often too busy working on their own startups.

We’d like to close that gap. As founders, we’ve built startups from scratch; Decide (acquired by eBay), Weave (YCS14), Divvy Homes (recently raised $43M Series B from GIC). Whether it’s figuring out your growth framework, what to prioritize, pricing strategy, fundraising or any of the other hundreds of things founders have to think about, we would like help.

Why Southeast Asia?

We believe there’s an incredible potential in Southeast Asia. The Financial Times has a great article titled “The Asian century is set to begin” with this quote:

The world’s imminent entry to an Asian age is coming not just because of its two largest economies, but also thanks to growth among smaller and midsize countries.

It also has this graphic that illustrates how much each country is expected to grow in the next few years.

A few highlights from that graph:

Indonesia – Indonesia’s economy is already larger than Mexico, France and the UK.
Vietnam – Vietnam’s economy will overtake 17 countries by 2023 including Belgium and Switzerland.
Singapore – Singapore’s economy will overtake 11 countries by 2023 including Belgium, Ireland and Chile.

To continue its development, Southeast Asia needs smart, gritty, courageous founders to solve its most important problems. These founders are out there but need the right support to succeed. We want to provide that.

Does Southeast Asia Need Another Accelerator?

No, it needs a different type of accelerator.

1. Experienced Operators

Being a founder is really hard. Largely because it’s so different than other jobs. At most jobs, (1) the company is unlikely to die soon, (2) the business model works, you’re just responsible for growing a part of it and (3) you typically have a lot of support and direction from teammates, managers, etc.

None of those things are true as a founder. Unless you’ve been a founder, it’s unlikely you’ve developed the relevant skills or mental frameworks to deal with the challenges faced by founders. The process of becoming a better founder is more apprenticeship than university degree. You only really improve by having a support system, a mentor and doing it.

We’ve started from 0, multiple times. We’ve worked out of the proverbial and literal basement. We’ve been told by investors that our idea was too small, too big or just dumb. We’ve gone from thinking we would be the next Facebook to wondering what we were doing with our lives. We’ve been you.

2. Aligned Incentives

Unlike other programs in the region, we will invest our full amount (USD $150K) in every startup and at the start of the program, not the end. We believe it’s important to align incentives from the beginning. Both as a show of confidence in the founders and their startup and a commitment that we will be helpful from that moment onwards whether good or bad.

3. Learn from Your Batchmates

Early stage startups face the same types of problems (not growing fast enough, can’t raise funding, not sure what to prioritize, etc.). The solution to these problems are almost always specific to the startup but often it’s helpful to be around startups facing similar problems and hear how they are solving them.

That’s why we have weekly group office hours where each of the startups will share their goals, challenges, progress, plans and priorities with the rest of the batch. There’s an additional element of accountability that comes from wanting to report good news to the batch.

Since we invest our full amount, in every startup, at the start of the program, our batch sizes are limited. That incentivizes us (by design) to only admit the highest quality founders and startups that we can find. In turn, that means you will not only be learning from our partners but also your batchmates who are the highest quality founders we could find.

4. Support System

It’s worth repeating, being a founder is really hard. You will constantly feel like an imposter (after 12 years, I still do). At times, you will need to convince your investors, team members and potential hires about a vision that you have doubts about. Your relationships, inside and outside the startup, will suffer. You will also, occasionally, feel like you’re onto the next big thing.

It’s emotionally and physically exhausting. Startups have a funny way of chasing the good with the bad and the bad.. often with more bad.

That’s why we believe founders need a support system. A place where they can be open about their challenges, candid about their fears and celebrate their wins. It’s at these times where we can add the most value. Since we invest upfront, we hope our founders will feel safe enough to come to us when things are at their worst.

5. Mentor and Alumni Network

Arguably the most valuable part of YC is access to the mentor and alumni network. To be able to ask hundreds of current and past founders about hiring, growth strategies, etc. is incredibly powerful.

We want to build a similar type of network for Southeast Asia. We’re seeding that network with our own professional networks built over 12 years of doing startups in Seattle, San Francisco and Singapore. We’ll also be building private internal tools (similar to YC’s BookFace) for mentor and alumni network to discuss, share and be generally helpful to each other.

Building something? You should apply.

If you want to know more about the program (structure, process, etc), I recommend going here.

If you’re working on something, I highly recommend you apply here. If you’re unsure whether you’re at the right stage, I recommend you apply.

Origin of a Venture-Backed, Acquired Startup

Tue, 14 Jan 2014 00:00:00 GMT

Decide.com was acquired by eBay on September 6, 2013. For those unfamiliar with Decide, we invented technology that predicted the future price of consumer goods. Thinking about buying a Samsung television? We’d tell you whether the price would drop in the next two weeks. We helped you decide when to buy.

Seattle, Late 2009 — For the past 18 months, my 3 co-founders and I had worked out of my brother’s cold cold basement, without pay or insurance. We’d built a handful of products that hadn’t gone anywhere. Our future wasn’t bright.

An Observation

Brian noticed his girlfriend (now wife) Jessica, had looked at the same dress on Nordstrom’s website several days in a row.

Brian: “Why do you go to that page every day?”
Jessica: “I’m waiting for the price to go down”
Brian: “Oh, we can build you something that checks it for you”

A Prototype

Brian told Ian about the conversation and by the following Monday they’d built a prototype. They showed it to Hsu Han and I during our weekly meeting on my brother’s couch. Without thinking about the normal startup criteria like market size, defensibility, differentiation (it would’ve failed all of them), we hopped in. It sounded like fun.

The prototype was simple, you entered a link to something you wanted to track, it would check the price every day and email you when the price changed.

It needed a little work though. Their initial prototype was a white page, a logo and a input box. Like Google. Missing were the customary one sentence value proposition and information on how it worked.

After taking 9 months to release our first product, we made a rule that any new product had to ship the following Monday. Regardless of how buggy, unpolished and how many features it lacked.

We renamed it PriceYeti (is it the right price yet?), made some small improvements and got it out the door.

Here’s what it looked:

Cutting Corners

To adhere to our “it must ship in one week” rule, we got good at cutting corners. For this project, we decided to not invest in the scraper despite its importance. Accurately tracking the price of a product was what the site was about. It’d be terrible to send an email, notifying someone that the price of something they wanted had dropped, only to have them find out it hadn’t. Over-promising and under-delivering is typically a bad user experience.

Unfortunately, building an accurate scraper that would work on any site would’ve taken weeks if not months. It would also require constant maintenance since websites are constant changing. It would’ve been a huge investment. We needed to know if this was something people wanted first.

The fastest way to write software is to be the software.

Instead we made a decent scraper and had users verify the current price. When the scraper detected a price change, rather than emailing the user automatically, it published the change to an internal dashboard. We (mostly Ian, thanks Ian) would check the dashboard several times a day, verify the price changes and send the emails by hand. It’s not a scalable solution but it got us moving immediately.

When a Bug Isn’t a Bug

For most sites our semi-automatic system worked. Someone would track a product from Costco’s website. We’d diligently check the dashboard for price changes, verify and send out the email.

It didn’t work that well for Amazon though. The system would detect a price drop from $500 to $450 but when we manually verified the price it would be $475. We thought there was bug but couldn’t find it. To help us troubleshoot we had the scraper save the page when it detected a price change. That way we could see the same page the scraper saw. If the scraper said $450 but the price on the page was $475 we’d know there was something wrong on our end.

That’s Interesting..

The scraper wasn’t wrong. When we manually looked at the page, the price was in fact $450. What’s going on?

Turns out in the few hours between the scraper detecing the price change and one of us manually verifying it, the price had changed again. The reason it was almost exclusively an Amazon problem was because their prices changed the most frequently.

We had no idea prices moved so often. We started collecting prices more frequently and graphing them over weeks. We found that prices even went up sometimes. We’d assumed, like most consumers, that prices either stayed constant or slowly declined as new products came out.

It was becoming apparent that when you bought something, not just what and where, was important. You could get the same product for significantly less, if you bought at the right time. If only there was a way to predict whether prices were going to go up or down.

It’s Good to Be Lucky

We’d met Oren Etzioni 2 years earlier while Ian was a student in one of his Computer Science classes at the University of Washington. He’d politely told us the idea we were working on at the time, data mining resumes, wouldn’t work. It’d only be useful if you have hundreds of thousands of resumes. How are you going to convince people to give your resume until then? Classic chicken or the egg problem.

2 years later, we were sitting in his office again. Telling him he’d been right but we’d found something else he’d be interested in. We showed him what we’d learned, that prices changed often, that people could save money if they bought at the right time, that we’d been wondering if it was predictable.

We couldn’t have found a better person to talk to about what we’d learned. Besides being a Computer Science Professor specializing in Artificial Intelligence, Oren was a startup founder. His last startup (Farecast) predicted the price of airline tickets, was acquired by Microsoft and became Bing Travel.

I can’t overstate how fortuitous this was. The only person who’d started and sold a consumer product based on predicting the price of something happened to be in Seattle and we happened to have taken classes from him.

He thought our findings were interesting. As he liked to say “I’ve seen this movie before.” But first he wanted more data. Specifically answers to the following questions:

Does price volatility happen in many categories?
Are the price changes meaningful?
Are they reasonably predictable?

We spent the next few weeks gathering data to answer those questions. I’ve never been to graduate school but I imagine it felt a little bit like that. Checking in with a professor every few weeks, presenting your findings. Fortunately the answer to all 3 questions was yes. As we progressed, Oren joined as the 5th co-founder and the rest, as they say, is history.

Big Ideas Start Small

If you’re gonna take anything away from this story maybe it’s that big ideas start small. The idea that led to Decide was price alerts. A feature most shopping sites have. Not a change the world, kind’ve big idea.

And that’s okay. I see people try to brainstorm their way into the next big idea. Like you can sit down one day, decide you’re going to come up with the next big idea and will it into your consciousness. That’s not how ideas work.

We wouldn’t have stumbled into price predictions for consumer products unless we’d built a website about tracking prices. Sometimes jumping down the rabbit hole is the only way to see how deep it goes. Most of the time won’t go very far but every once in a while you’ll find an entire world.

Postmortem of a Venture-Backed, Acquired Startup

Thu, 17 Oct 2013 00:00:00 GMT

I’m obviously happy with the outcome. From my brother’s basement, we raised $16.5M, employed 30 people and most importantly built something people found useful. Only in my wildest dream would I have the audacity to think I’d build something that would be acquired by a company like eBay.

Although the outcome was positive, it’s helpful to reflect on the experience. Below are 3 lessons I learned from the experience. They’re not new but hopefully from within the context of a real startup they’re more tangible.

Scale Your Company with Your Product

We launched Decide.com on June 20th, 2011 with 20 people and 2 round of funding totaling $8.5M. That’s atypical, especially these days when investors want to see traction before investing. Instead of those resources helping us find market-fit it slowed us down.

We focused on company building:

Hiring. We hired 20 people in the 9 months following our first round of funding. That took lot of work. For every person you hire, you might interview 5 others, for each of those people you have a recap meeting with everyone that interviewed them, work on a offer for the person you want then sell them. That doesn’t take into account the work it takes to source candidates and fire people that don’t work out.

Organizational Structure. As we kept adding people, we had to figure out how they’d all work together. What teams should do what? Who leads those teams? How do those teams interact? Who needs to be in the meeting where we decide what we’re doing?

Once we got through those parts, it was harder to change directions:

Communication Overhead. The larger the team the more energy you need to spend communicating. Let’s say you want to change directions, something that happens often pre-market-fit. You have to convince everyone that there needs to be a new direction, what the new direction is and why this new direction will work when the last one didn’t.

Specialization. As companies grow, the people they hire become more specialized. If we had pivoted to target businesses (we talked about it) instead of consumers we’d need sales people (we didn’t have any) which probably meant we’d have to layoff other people (probably engineers) to hire them. This didn’t happen to us but it impacted some of our decisions so I felt it was worth mentioning.

Do whatever helps you reach market-fit faster. In my experience, that’s raising modestly (if at all) and keeping the team small. 2 or 3 people can get a lot done and is small enough where you don’t have to worry about structure or communication.

You wouldn’t scale your technology to support 10M users when you only have 100. Don’t scale your company that way either.

Hire Similar Minded People

Our hiring criteria can be summarized with 2 basic questions: (1) Are they great at what they do? (2) Do we like them? Turns out there should’ve been a 3rd question. (3) Do you generally agree with me?

We spent a lot of time debating rather than doing. What technology should we use? When should engineering get involved with the design process? What’s our design methodology? How long should our release cycles be? Should we even have release cycles? I could go on indefinitely.

Decide how you want do things then hire people that want to do things that way. There’s value in having a diversity of opinion but in a early stage startups, the benefits (moving fast) of hiring people that generally agree with you outweigh the benefits (diversity of opinion) of hiring people that don’t.

If you can’t hire anyone that agrees with you, re-evaluate how you want to do things.

Expanding Your Target Market Doesn’t Help You Find Market-Fit

We started with support for 3 categories (televisions, laptops and cameras) but expanded into all electronics within the first year. Looking for more growth, we talked about a few options:

Expand into more categories – Be relevant to more purchases. Number 1 thing our users asked for.
Expand beyond when to buy – Move up the buying funnel by helping people figure out what they should buy (product scores, reviews, comparison,etc.) in addition to when.
Double-down on when to buy – Make it so good, if anyone is buying in a category we cover, they have to use us.

We ended up doing a combination of 1 and 2. We built features (product recommendation, sentiment analysis, etc.) to help people figure out what to buy then expanded our coverage into appliances, home and garden, babies and kids. We looked for growth by expanding how often we would be relevant (more categories, cover more of the buying funnel).

Most of our traffic stayed in electronics and our experience became thin. We were in a lot of categories trying to do a lot of things. I would’ve liked to see us do option 3, double-down on when to buy. That was the part of the market nobody else was playing in. Get that experience to be amazing then roll it out into other categories.

Expanding your target market, like we did by supporting more categories and building when to buy features, doesn’t help you get to market-fit. In fact, I’d argue it makes it harder because there’s more needs to satisfy.

Said another way, is it better to start with a small group of people who love what you’re building or a large group of people who are indifferent? I think you’d rather have the small group that love you. It’s easier to turn a small group of rabid users into a large group than a large group of indifferent users into a large group of rabid users.

Keep your product narrow and focused on the small group of people most likely to love what you’re building.

Something positive

I don’t want to end this negatively because there’s a lot we got right. I just tend to learn more from failures than successes. Maybe I should write another post about worked for us. For now, let me tell you about the my proudest moment at Decide, when we lost almost all our revenue.

When we created the first version of our mobile app, we put Amazon’s prices next to everyone else’s just like we did on the web. Apparently Amazon doesn’t allow you to do that on a mobile device but it’s ok on the web. They want you to use their app to check their prices.

We received a notice from them informing us we weren’t compliant and unless we removed it they’d suspend our affiliate account. We weren’t making a lot of money but that account probably represented more than 80% our revenue.

We had 2 choices: (1) Comply and remove Amazon prices from our mobile app (2) Not comply, keep the prices and lose our biggest revenue stream. Proudly and with very little discussion, we decided not to comply. Most people want to buy from Amazon. Not being able to show those prices to our users wasn’t acceptable. In the 2 years since we got that notice we never complied. When faced with a decision where we had to choose between revenue or our users, we chose our users. I couldn’t have been more proud.

If you want to see what I’m working on now checkout CoffeeMe, or as we like to call it Tinder for professional networking.

Hsu Ken Ooi

Control, Influence and Agency

Control

Influence

Agency

Conclusion

Non-AI, AI Companies

True AI (AI, AI)

Fake AI (AI, Non-AI)

Traditional (Non-AI, Non-AI)

Stealth AI (Non-AI, AI)

Why Traditional Companies Stay Put

Migration

The Southeast Asia Opportunity

Single User Software

The AI-Pilled Advantage

Should You Stop Working on Your Startup?

1. Growth

2. Learning

3. Experiments

4. Conviction

5. Stop Working on Your Startup

Dogmatic vs Scientific Founders

Dogmatic Founders

Scientific Founders

Why Office Hours Are Effective

Prioritization

Focus

Accountability

Ambition

Empathy

5 Most Common Startup Pitch Mistakes and What to Do

Act 3 Problems Are Act 1 Problems

What's Your Secret?

Explain Your Growth

Raising $X to Reach Y Milestone

So What?

What to Look for in Your First Technical Hire

Top Posts and Book Recommendations of 2023

Top 10 Posts

Top 5 Books

Convincing People to Work at Your Startup

Don’t Compete On..

Compete on..

Is Your Startup Solving an Important Problem?

How to Tell if a Problem Is Important?

What Type of Companies Typically Work on Which Type of Problem?

What Iterative Looks for When Investing in Founders

1. Does What They Say, Consistently

2. Convincing

3. Internally Motivated

The Difference Between Bad, Good, and Great Startup Engineers

Should You Continue Working on Your Startup?

Is it growing fast?

Am I learning about my users?

Do I have lots of ideas for what to do next?

Does Story or Traction Matter More When Fundraising?

Traction

Story

Cases

Strong Traction + Strong Story

Weak Traction + Weak Story

Strong Traction + Weak Story

Weak Traction + Strong Story

Observations

Learnings Compound, Money Doesn't

Founder Simulations

Increasing a Startup's Chances of Being Successful

How to Think About Product

Invest in Slope

Founders

Market

Non-Linear

Starting Iterative

Why Southeast Asia?

Does Southeast Asia Need Another Accelerator?

1. Experienced Operators

2. Aligned Incentives

3. Learn from Your Batchmates

4. Support System