Startup College Startups 101

Module 5 of 9 · 18 minutes

Build and launch your MVP

A one-page MVP plan with a launch date, two or three channels, a finished launch post and the numbers you will watch in the first week.

Scope the smallest product that tests your riskiest assumption

This module takes you from a problem you've validated to a first product in real users' hands. You'll decide what to build and what to leave out, choose how to build it, set a launch date, pick your channels, write the launch post and decide what to measure in the first week.

A minimum viable product, or MVP, is the smallest version of your product that lets real people do one job, so you can find out whether your riskiest assumption holds. Think of it as an experiment with real users, not a cut-down copy of the product you imagine.

Start from the assumption. Your customer conversations should have produced a list of things that must be true for the business to work. Pick the one most likely to be wrong and most damaging if it is. Common examples:

  • People will switch from what they use today, such as a spreadsheet, a group chat or a paper diary.
  • They, or their employer, will pay for it.
  • They'll come back without being reminded.
  • You can reach them at a reasonable cost.

Then write down the one job your MVP does, in a sentence, for example: a parent can book and pay for their child's next piano lesson from a link the teacher sends. If you can't describe it in one sentence, the scope is still too big.

The smallest test is sometimes barely software at all. You can do the job by hand behind a simple front end: a form you answer personally, a spreadsheet you update yourself, messages you send manually. If people use and value the manual version, you've learned something real before writing much code.

Cut features ruthlessly

List every feature you have in mind. For each one, ask: would the test still work without this? If the answer is yes, move it to a list called "not yet". Keep that list. It stops good ideas from creeping back in, and parts of it will become your plan for later.

Usually safe to leave out: passwords and user accounts (a sign-in link sent by email will do), settings, admin dashboards (use a spreadsheet), integrations, a mobile app (a website that works on phones is enough) and automated billing (send invoices by hand).

Not safe to cut: the core job working reliably, a way for users to reach you, an honest description of what the product does and doesn't do, and care with people's data. Privacy laws may apply wherever your users live; the EU's GDPR, for example, can cover companies based outside the EU. Find out what applies to you before launch.

A booking tool for piano teachers started with fourteen planned features. The founder kept three: a booking page for each teacher, card payments and a reminder email the day before each lesson. Calendar sync, group lessons, invoices and a teacher app went on the "not yet" list. The first version was ready in three weeks, not the four months the full list would have taken.

Choose how to build it

You have three broad options. Pick the one that gets a working version to users soonest and most cheaply, in a form you can change yourself next week.

No-code tools. Website builders, form tools, spreadsheets, automation services and online-store platforms can be combined into surprisingly capable products without programming. They are often the right choice for a first version, especially for booking, directory, marketplace and internal-tool ideas. You may hit limits later and need to rebuild. That's acceptable: rebuilding after you've learned what people want costs less than building the wrong thing well.

A simple stack you already know. If someone on the team can code, use the tools they know best, not the ones they'd like to learn. Rely on established hosted services for the hard, risky parts such as payments, sign-in and email. AI coding assistants can help a less experienced builder get further, but someone still has to understand, test and secure what ships, especially anything that handles money or personal data.

Hiring someone. Freelancers or an agency make sense when the product needs skills nobody on the team has, but every change becomes slower and more expensive. If you hire, keep the scope to your MVP list, pay by milestone, make sure the contract says your company owns the work, and keep the domain, code, hosting and app-store accounts in your company's name from the first day.

Set a launch date and choose your channels

Pick a launch date two to six weeks away and tell someone who will hold you to it. A date forces decisions: anything that won't be ready moves to "not yet". Launching here simply means opening the product to people outside your team, and you can launch again later, to new audiences, as the product improves.

Choose two or three channels where your particular customers already spend time:

  • Communities they use. Industry forums, online groups, professional associations and local business networks. Read the rules, take part before you post, and ask moderators whether launch posts are welcome.
  • Show HN. A section of Hacker News for things people have made that others can try. The audience is mostly technical, so it suits developer tools and tech products. Its guidelines ask for something people can use right away, not a sign-up page or an article. Don't ask friends to upvote; the site penalizes it.
  • Product Hunt. A daily list of new tech products, followed by early adopters and other makers. It can suit consumer and productivity tools, and is less useful if your customers are, say, dentists. Read its current posting rules first.
  • Newsletters. Niche newsletters your customers read may feature new tools or sell small sponsorships.
  • Direct outreach. Personal messages to the people you interviewed, your waiting list and your own network. This is often where your first real users come from.

A big audience is not the same as the right audience. Ten visits from the people you built it for can teach you more than a thousand from other founders.

Write a launch post that asks for feedback

A good launch post is short and specific. Include:

  1. Who it's for, in the first line.
  2. The problem, in the words customers used with you.
  3. What the product does today, and what it doesn't do yet.
  4. How to try it, and whether it costs anything.
  5. The one piece of feedback you most want.
  6. A line about who you are and why you built it.

Leave out hype and invented numbers. Most communities can spot a sales pitch, while an honest note that the product is early and still missing things tends to invite help.

For independent piano teachers who arrange lessons by text message: over the past two months I talked to twenty teachers, and most lose more than an hour a week to booking and rescheduling. So I built a simple booking page. Parents pick a slot, pay by card and get a reminder the day before. It doesn't sync with calendars or handle group lessons yet. It's free while I test it, and the link is below. I'd like to know one thing: what would stop you from sending this link to your students' parents? I taught piano for six years before learning to build software, and I'll reply to every comment.

Adapt the length and tone to each channel, but keep the core the same: who it's for, what it does and what you want to learn.

Measure the first week

Before you launch, write down what you'll measure and what result would encourage or worry you. If you decide afterward, it's too easy to explain away disappointing numbers.

Track a handful of numbers daily in a simple spreadsheet:

  • Visitors per channel. Give each channel its own link so you can tell them apart.
  • Sign-ups, and what share of visitors they represent.
  • Activation: how many people complete the one job, such as booking a first lesson.
  • Return use: how many come back and do it again within the week.
  • Conversations: how many users you've talked to or exchanged messages with.

Upvotes, likes and page views feel good but say little about whether the product works. Activation and return use say much more.

Expect small numbers. Most launches are quieter than founders hope, and most startups fail, which is why cheap, fast tests matter. A quiet launch is still information: perhaps the wrong channel, an unclear post or a problem that isn't painful enough. Talk to everyone who signed up, especially those who didn't come back, and use what you hear to decide what to change before your next launch.

Do the work

Fill in the MVP launch plan builder below. It produces a one-page plan and your launch post, ready to share with a co-founder, an adviser or the person holding you to your date.

  • Write the assumption as a sentence that could turn out to be false.
  • Describe the one job in a sentence, from the user's point of view.
  • List what's in and what's out. If the "out" list is shorter than the "in" list, you probably haven't cut enough.
  • Say how you'll build it and who will do the work.
  • Choose a real date and two or three channels, each with a reason.
  • Draft the whole launch post, not notes for one.
  • List the numbers you'll watch, with the result that would encourage you next to each.

You're done when someone else could read the plan and know exactly what you're shipping, when, where you'll announce it and how you'll judge the first week. Then put the date in your calendar and start building.

Your document for this module

MVP plan and launch post. A one-page plan for the smallest product that tests your riskiest assumption, with a launch date, channels, the launch post itself and the numbers you'll watch in the first week. Share it with your team or with someone who will hold you to the date.

Open the mvp plan and launch post builder (+100 XP)

Check yourself

1. You have twelve features planned for your MVP. Which question best decides what stays in the first version?
2. You can't code, and you want to test whether small gyms will take class bookings through a shared link. What's the most sensible way to build the first version?
3. Your customers are independent bakery owners. Which launch plan is likely to teach you the most?
4. Which launch post is most likely to get useful responses?
5. In launch week you get 2,000 visits and 300 sign-ups, but only 15 people complete the core action and 3 come back. What's the most useful reading?