Contents · Project leader guide

Project leader guide · Chapter 2

Build less, learn more

Start with the smallest version that teaches you something, treat everything as an experiment, tackle your biggest risk first and iterate fast.

Start with the smallest version

With very few exceptions, you want to build less. Whatever you’re planning, try to cut its scope by 90% and get that version out.

Planning a website to automate a process? Run the process by hand with a spreadsheet and your first users. Building is slow. You don’t want to spend weeks on a scheduling website only for the first call to go wrong because the process doesn’t work. Do it manually and learn as much as you can.

Say you’re building a tool that reads people’s bank statements to find products to boycott. Take your own statements and go through them by hand to work out what you’d boycott. Anything you’d learn from building the tool, you can learn this way first.

Put another way: you don’t yet know what to build. You may think you do, but you won’t really know until people use it, and then you’ll find out how many of your assumptions were wrong. So build small, get feedback and iterate.

This applies to almost everything. If you need people to take part in something, ask them whether they would before you build it. If they wouldn’t, find out why. There’s something critical there that you need to learn first, or you’ll waste everyone’s time building the wrong thing.

Your smallest version may not be an app at all. It could be a spreadsheet, an SMS system or a Signal group.

Treat everything as an experiment

The worst thing you can do is decide on a vision, build it, and only then show it to users. Most people who do this find users don’t want what they built, and that those users could easily have told them so months earlier.

Your path to market, your target users and the medium you use are all experiments. You build a small version because you need to experiment. Don’t commit to any of it. If you see signs that something isn’t working, there are two likely reasons:

  • the strategy is wrong, and you need a different one; or
  • the strategy is fine, but the way you carried it out (the tactics) was poor.

Take ethicsVC. It started as a way to get venture capitalists to sign a letter supporting a ceasefire. We learned that wasn’t going to happen, and learned how VCs thought about it. So it moved to naming and shaming, and found that many people were angry about VCs’ complicity. But it only found an effective strategy when the student movement took off. Even then, getting students to adopt it was hard. Was the strategy wrong, or the tactics? The team tried many tactics, learned from each and got feedback from students along the way, until people were excited about what they were doing.

The good news is that experiments never fail. You learn from them and do something else.

Tackle your biggest risk first

Which experiments matter most depends on your risks. What is most likely to make everything fail? For a boycott scanning app, it might be whether a phone scanner can work at all: a technical risk. For ethicsVC, it was whether students would adopt its criteria: an adoption risk. For an AI chat project, it might be “who is this actually for?”.

Attack the biggest risk first, then the next. There’s nuance: if the first risk takes a month to test and the second takes six, start the second early and run them in parallel.

The risk is rarely “can we build it?”. Anyone can build a product with enough time.

It’s easy to fool yourself about your risks and work on what you’re comfortable with. But risk rarely lies where you’re comfortable, because that’s the part you already understand. Engineers in particular need to stop writing code and focus on getting users. It’s harder, but “will anyone care?” is a much bigger risk than “can I build it?”.

Use feedback, don’t follow it

The first time you speak to users, they’ll give you a sense of what they need. Don’t build what they tell you to build.

Instead, combine what several users know and believe with your own knowledge and intuition. You’ll have to set some people aside as outside your target group. Expect to set aside at least 20% of the people you speak to. It’s not unusual to set aside half, because you find a valuable niche in the other half.

Then work out your next step. It might be building a small version of your product, but it’s more likely to be setting up a manual process.

Iterate, iterate, iterate

Once you have something, show it to users and have them try it. Watch them while they use it: their face and the screen. You’ll learn everything that’s wrong, where you misunderstood the problem, and new facts they forgot to mention that now block them. They’ll probably be confused and ask lots of questions. Resist answering unless they’re so stuck you can’t learn anything more. Note everything that went wrong.

Then fix what was wrong, broken, confusing or didn’t make sense, and show it to them again.

The faster you iterate, the faster you reach your goal. The best teams iterate in minutes or hours. Aim for a day if you can. If each round takes three weeks, that’s a serious problem: getting something right usually takes dozens of rounds.

More to think about

Test the whole process

For some projects, especially processes, it helps to prove the whole thing works before committing to it. Apricot helps people in Gaza get hired remotely. We pushed them to get five people hired before doing anything else, so they understood every part of the problem: pitching, interviews, payments and internet access.

This doesn’t suit every project, but if yours has a well-defined scope, making sure you truly understand each part of it is very valuable.

Who exactly are you serving?

Are you building for organizers or for protesters? For social media activists or for specific journalists? Organizations or individuals? The answer changes how you build your product, what it does and how you reach people.

Your theory of change

A theory of change explains how your project brings about change for the movement. Spelling out exactly how you think change happens will help with every product and market decision. At any point you should be able to say: “We believe X is very important. Our project Y will have effect Z, which pushes X in the right direction.”

Have a project for Palestine?

If this sounds like the way you want to work, apply to the incubator. It takes 10 to 30 minutes, and we aim to decide within 10 days.