Product

What to build first: cutting scope without gutting the product

The hardest decision in a first release is what to leave out. A practical method for finding the one workflow that has to work, and for deciding what can be done by hand instead of built.

2 min readBy Mayur Kale

Key takeaways

  • A first release should do one thing completely rather than five things partially. Partial features generate feedback about your bugs, not about your idea.
  • Ask what a customer is paying for. Everything that is not that can usually be deferred, faked or done manually.
  • Manual is a legitimate implementation. Doing onboarding, billing or moderation by hand teaches you what to automate later.
  • Cut by user type before cutting by feature. Serving one persona properly beats serving three badly.
  • Write down what you deliberately excluded. Undocumented exclusions get quietly re-included by whoever is closest to the work.

Every founder has been told to cut scope. Almost nobody is told how, so the usual outcome is a list where everything has been trimmed by twenty percent, which produces a product where nothing quite works, at eighty percent of the cost.

Cutting properly means removing whole things, not shaving all of them.

Start from the moment of value

There is a single point in your product where a user gets what they came for. The report appears. The booking is confirmed. The document is signed.

Everything on the direct path to that moment is in scope. Everything else is a candidate.

This sounds obvious and is surprisingly clarifying, because most feature lists are assembled by imagining a mature product and working backwards. Working forwards from a first-time user's actual journey produces a much shorter list, and a different one.

Then ask: could a person just do this?

For each remaining item, ask whether a human could do it by hand at the volume you realistically expect in the first three months.

Things that are almost always fine to do manually at first:

  • Onboarding, a call is better than a wizard, and you learn more
  • Billing, an invoice works fine below a few dozen customers
  • Moderation and approvals, a person with a shared inbox
  • Reporting, a query and a spreadsheet, sent on Fridays
  • Support and refunds, obviously
  • Matching, routing, triage, a person makes better decisions than your first algorithm would

Cut by user type before feature

The highest-leverage cut is usually not a feature. It is a persona.

Most product plans serve three groups: the primary user, someone who administers them, and someone who reports on both. That is three products in a trenchcoat, and the second two can nearly always wait.

Serving one persona completely gives you users who love the product. Serving three partially gives you three groups who find it almost useful, which is not a position anyone builds from.

A method that works

  1. Write the user's journey as sentences. "A landlord logs in, sees which properties have open issues, and assigns one to a contractor." Concrete, not abstract.
  2. Mark the moment of value. One sentence in that list.
  3. Delete every sentence not needed to reach it. Be aggressive; you are not deleting them from the product, only from this release.
  4. For each survivor, ask if a person could do it. If yes, that is the implementation.
  5. Cut to one user type. Whoever gets the value first.
  6. Write down what you removed and why. This is the non-goals list, and it is the artefact that stops all of it quietly reappearing.

Step six matters more than it looks. Scope does not return through big decisions, it returns through a hundred small reasonable additions, each of which nobody would have refused on its own.

What good looks like

A first release that does one thing completely, for one kind of user, with an ugly admin process behind it and a founder personally onboarding every customer.

That is not a compromise. It is the version that gets you a real answer fastest, and it is what most successful products actually looked like at the point nobody was watching.

Frequently asked questions

How do I decide which features go in an MVP?

Work backwards from the moment of value, the point at which a user gets the thing they came for. Everything on the direct path to that moment is in scope; everything else is a candidate for deferral. Then ask of each remaining item whether a person could do it by hand at your expected volume. If yes, that is your first implementation.

What if customers expect a feature we have not built?

Expectations vary far more than founders assume, and the way to find out is to ask rather than to build. Where a genuine expectation exists, meeting it manually is usually acceptable to early customers, people will tolerate a slow or human-powered process from a small product far more readily than a broken automated one.

Isn't it risky to launch with too little?

The risk is real but smaller than the alternative. A narrow product that works gives you clear signal about whether the core idea has value. A broad product where everything half-works gives you noise, costs several times more, and takes months longer to reach the same conclusion.

How do we know when to stop cutting?

When removing the next thing would break the core promise. If a user could no longer complete the journey that made them sign up, you have cut into the product rather than around it. Everything up to that line is negotiable.

References & further reading

  1. [1]
  2. [2]

Turn the trend into shipped software

I contract as a senior fullstack and data engineer, building the production systems your customers and your reporting depend on.

Explore services↗