What should go into the first version of my product, and what should wait?

By Bonaventure Ogeto, Founder and CEO, Mctaba Labs
Published

The short answer

The first version should do one job for one kind of customer, from first contact to payment, and nothing else. Keep whatever that journey cannot happen without, including a way to take money and a way to reach the customer. Admin dashboards, apps for a second platform, reports, referrals and automation of rare cases all wait.

A founder's first feature list tends to run to about twenty items. Customer app, provider app, admin dashboard, wallet, ratings, referral codes, chat, notifications, reports, loyalty points. Each one felt necessary on the evening it was written down.

Then the quote comes back and it is three times the budget. The founder asks the developer what can be removed, and the developer, who does not know the customers, guesses.

There is a better way to cut, and you can do it yourself in an afternoon with a sheet of paper. It rests on one idea: a first version is not a small copy of the final product. It is the shortest complete path from a customer's problem to your M-Pesa line.

What is the one job the first version must do?

It is the single result your customer is paying for, stated in one sentence. Write it in this form:

When [a specific customer] needs to [do something], they can [get this result] and pay for it.

Take a product that is made up for this article. Call it Fundi Mfano (mfano is Swahili for example): a service that lets households in Nakuru book a vetted plumber. The founder's feature list has seventeen items. The job sentence has one:

When a household in Nakuru has a plumbing problem, they can book a vetted plumber for a set time and pay a booking fee.

Notice what is missing. Nothing about electricians, ratings, a wallet or other towns. One customer, one problem, one result, one payment. If your sentence needs the word "and" between two results, you have two products. Pick the one your early customers asked for.

How do I sort features into must, should and later?

Walk through the journey step by step, then put every feature through one test.

First, write the steps a customer takes from hearing about you to paying you. For Fundi Mfano:

  1. Sees a link on WhatsApp or a poster.
  2. Opens a page that explains the service and the price.
  3. Describes the problem and picks a time.
  4. Pays the booking fee.
  5. Gets a confirmation.
  6. The plumber is told where to go and when.
  7. The plumber arrives and does the work.

Now take each feature on the long list and ask: can the customer complete this journey and pay without it?

  • If no, it is a must.
  • If yes, but the founder would have to do something by hand, it is a should. It stays manual for now.
  • If yes, and nobody would notice it missing in the first month, it is later.

Here is the Fundi Mfano list after the test.

Feature Verdict Why
A web page that explains the service and the fee Must Step 2 cannot happen without it
Booking form with problem, location, time and phone number Must It is the job
Booking fee paid on M-Pesa Must No payment, no test
Confirmation SMS or WhatsApp message to the customer Must People do not trust a payment with no receipt
A simple list of bookings the founder can see Must Someone has to act on each booking
Telling the plumber about the job Should The founder phones the plumber for now
Matching a booking to the nearest plumber Should The founder picks from a list of ten by hand
Paying the plumber Should The founder sends M-Pesa at the end of each day
Vetting plumbers Should Done in person, on paper, before launch
Customer accounts and login Later A phone number identifies the customer
Plumber mobile app Later Plumbers answer calls and WhatsApp today
Ratings and reviews Later The founder calls each customer after the job
Wallet and loyalty points Later Nobody has booked twice yet
Referral codes Later Unknown whether anyone will refer
Electricians, cleaners and other trades Later A different job
Admin dashboard with charts Later A list and a spreadsheet are enough at this volume
Android and iPhone apps Later The web page works on every phone

Seventeen items became five things to build. The "should" column is the part that is easy to miss. Those are real needs, but a person can meet them while volumes are low. Doing them by hand for the first fifty bookings also teaches you how they should work before you pay to automate them.

What always belongs in a first version in Kenya?

Two things, whatever the product: a way to take payment and a way to reach the customer.

Taking payment. A first version that cannot take money cannot answer the only question that matters. In Kenya that usually means M-Pesa. You have a range of options, from simple to automated: a Till or Paybill number displayed with instructions, a payment link from a provider such as Paystack or Flutterwave, or an STK push through Safaricom's Daraja API that puts the payment prompt on the customer's phone. Choose the simplest one that your volume allows. But decide it now, because the payment step shapes the screens around it.

A way to reach the customer. Collect a phone number at the first step and use it. A confirmation by SMS or WhatsApp is not a nice extra here. Buyers are used to an M-Pesa message for every shilling that moves, and a payment followed by silence feels like a scam. The phone number also lets you call customers afterwards, which is the best research tool you have.

Three smaller things nearly always belong too, because their absence breaks trust:

  • A clear statement of the price before the customer is asked to pay.
  • A visible way to get help: a phone number or WhatsApp line that a person answers.
  • A record of every transaction that you can look up when someone says "I paid".

If the product handles personal information, also tell people plainly what you collect and why. It costs one paragraph. For what else may apply to your product, read the guidance published by Kenya's Office of the Data Protection Commissioner.

What should wait?

Anything that serves a customer you do not have yet, or a problem you have not yet met.

  • The second platform. Build for the web or for one phone platform, not both.
  • The second type of user. If plumbers can be reached by phone, they do not need an app in month one.
  • Dashboards and reports. You can count fifty bookings in a spreadsheet.
  • Automation of rare cases. Refunds, disputes and cancellations will happen a few times a month at first. Handle each one by phone and write down what you did.
  • Growth features. Referrals, loyalty points and promo codes multiply demand that exists. They do not create it.
  • Settings and preferences. Every option is a decision you have pushed onto the customer and a screen you have paid for.
  • More languages, more towns, more categories. Each is a new market. Win one first.

"Later" does not mean "never". Next to each waiting item, write the evidence that would bring it back. "Plumber app: when I am phoning more than twenty plumbers a day." "Ratings: when customers ask who the plumber is before they book." That turns a wish list into a set of rules, and it stops the list from creeping back into the build.

How do I know the scope is small enough?

Use these four checks.

  1. The one-sentence test. You can describe the result without joining two results with "and".
  2. The demo test. You can show the whole journey, from link to payment confirmation, in under two minutes.
  3. The manual test. At least a few items on your original list are being done by a person, on purpose.
  4. The discomfort test. You feel a little embarrassed by how little it does. If you feel proud of how complete it is, it is too big.

If the scope fails any of these, go back to the table and move one more item to the right.

What does a smaller first version change about the build?

It changes the cost, the timeline and what you learn.

A scope of five must-haves can be quoted accurately, because there is little room for misunderstanding. It can be tested properly, because there are few paths through it. And it reaches customers sooner, which matters more than either, because everything on your "later" list is a guess until real people use the product.

Some of those guesses will be wrong. The plumbers may never want an app. Customers may care about a fixed price far more than ratings. A founder who built all seventeen features finds this out after the money is spent. A founder who built five finds out with most of the budget still available to build what customers asked for.

Key takeaways

  • Write the one job as a single sentence before you list any features.
  • Sort every feature with one test. Can the customer complete the job and pay without it?
  • Taking payment and a way to reach the customer belong in version one. In Kenya that usually means M-Pesa and a phone number.
  • Anything you can do by hand for the first fifty customers should stay manual.
  • Polish the one screen where the customer decides to pay, and leave the others plain.
  • Write a rule for each waiting feature that says what evidence would bring it back.

Frequently asked questions

How many features should an MVP have?
There is no right number, but there is a right shape. One complete journey for one type of customer, from finding you to paying you. If you can describe the result without joining two results with "and", the scope is probably about right.
Should the first version have user accounts and passwords?
Only if the job cannot be done without them. Many first versions work with a phone number and a one-time code, or with no account at all for the paying customer. Passwords, password resets and profile pages are real work that the customer does not value.
Does the first version need to look good?
It needs to look trustworthy, which is a lower bar. Clear words, a clean layout, your business name and a working phone number are enough. Spend design effort on the screen where the customer decides to pay and keep the remaining screens plain.
What if a customer asks for a feature that is on my later list?
Write it down with their name and ask what they do today without it. If three paying customers ask for the same thing unprompted, move it up. One request is an opinion. Three from people who pay is a pattern.
Can I launch without M-Pesa integration?
Yes, if payment still happens. A Till or Paybill number shown on the screen and a manual check of the confirmation message is a valid first version for low volumes. What you cannot do is launch with no way to pay at all, because then you are not testing whether people will pay.

Sources

  1. The Lean Startup: How Today's Entrepreneurs Use Continuous Innovation to Create Radically Successful Businesses, Eric Ries (book, 2011). Checked on 19 November 2026.
  2. Competing Against Luck: The Story of Innovation and Customer Choice, Clayton M. Christensen, Taddy Hall, Karen Dillon and David S. Duncan (book, 2016). Checked on 19 November 2026.