Scoping a SaaS MVP: what to build first and what to leave out

How to decide what goes in the first version of a SaaS product: the one-workflow test, what to do by hand, what never to skimp on, and a scope template.

A SaaS MVP should contain one complete workflow that solves one painful problem for one type of customer, plus the few foundations you can’t fake: secure sign-in, correct billing, separation between customers’ data and a way for customers to get their data out. Everything else is a candidate for later, for doing by hand, or for never.

The hard part isn’t knowing this. It’s saying no to the features that feel essential in a planning meeting and turn out to be used by nobody.

  • Define the job the customer is hiring your product to do, then build only what that job needs.
  • Apply the one core workflow test: can a customer go from sign-up to the result they care about without you?
  • Do things by hand behind the scenes until volume forces you to automate them.
  • Don’t cut corners on security, data export or billing accuracy. Those mistakes cost customers’ trust, and trust is hard to win back.

What an MVP is, and isn’t

Eric Ries, author of The Lean Startup, defined the minimum viable product as “that version of a new product which allows a team to collect the maximum amount of validated learning about customers with the least effort”. The key word is learning. An MVP is an experiment to find out whether customers will use and pay for something, not a cut-down version of your five-year roadmap.

Y Combinator’s Michael Seibel gives a plainer version in his talk on planning an MVP: it’s “the first thing you can give to the very first set of users you wanna target, in order to see if you can deliver any value at all to them.” He adds that you should be able to build it “in weeks, not months”.

For B2B SaaS that’s often too short if you count billing and security properly. The principle still holds: the smaller the first version, the sooner you find out if you’re right.

Start with the job to be done

Ask what the customer is trying to get done when they reach for your product, in their words. Say you’re building software for independent letting agents. The job isn’t “manage properties”. It might be “chase late rent without awkward phone calls” or “get gas safety certificates renewed before they lapse”. Each of those is a different product with a different MVP.

A useful format: “When [situation], I want to [action], so I can [outcome].” Write three or four of these from your customer interviews and pick the one that came up most often with the most pain attached. If you haven’t done those interviews yet, start with validating the idea before scoping anything.

The one core workflow test

Trace the path a new customer takes from first visit to the outcome they care about. For the letting agent: sign up, import tenancies, the system sends rent reminders on the right dates, the agent sees who has paid. If any step on that path is missing, the MVP doesn’t work. If a feature isn’t on that path, it probably isn’t in the MVP.

Then ask, for each item on the list:

  • Would a customer refuse to pay without it?
  • Can it be done by hand, by you, for the first 20 customers?
  • Does it help you learn whether the core idea works?

Only the first question puts something firmly in scope. A “yes” to the second means it moves to your manual task list.

Must-have versus nice-to-have

Usually must-have Usually nice-to-have at first
The core workflow, end to end Dashboards and custom reports
Sign-in, password reset, basic roles Configurable permissions per customer
Correct billing and invoices Coupons, referral schemes, usage tiers
Data separation between customers Single sign-on for enterprises (until one asks)
Export of the customer’s own data A public API and integration marketplace
Basic audit log of key actions Native mobile apps (if a mobile web page will do)
Email notifications for the core workflow Theming, white-labelling, dark mode

The right-hand column isn’t “never”. It’s “when customers pay for it or ask for it repeatedly”.

What to fake or do by hand at first

Paul Graham’s essay “Do things that don’t scale” notes that “you can sometimes get away with doing by hand things that you plan to automate later.” Two well-known patterns:

  • Concierge. You deliver the service yourself, openly, and the software is thin or absent. You onboard each customer personally, import their spreadsheet for them, send the weekly report by email. You learn exactly what they value before you code it.
  • Wizard of Oz. The customer sees what looks like an automated product, but a person does the work behind the screen. A “smart” matching feature might be you reading each request and picking the match. It tests whether customers want the result before you build the clever part.

Good candidates for manual work: data import, onboarding, report generation, account changes, refunds, and anything AI-driven where a person can produce the output while you check whether customers value it. Be honest in your terms about what happens behind the scenes, especially where personal data is handled by people rather than code.

Graham’s essay also describes Stripe’s “Collison installation”: when anyone agreed to try the product, the founders would say “Right then, give me your laptop” and set them up on the spot. That kind of hand-holding is the job at MVP stage, not a distraction from it.

What not to skimp on

Security

One leak of one customer’s data into another customer’s account can end a young B2B product. Build tenant separation properly from the start, use an established sign-in library rather than writing your own, encrypt data and keep dependencies patched. Buyers will ask about this sooner than you expect; the questions are covered in security and compliance for a UK SaaS.

Data export

Customers trust a product they can leave. Under UK GDPR, individuals have a right to data portability, and the ICO describes this as receiving data in a “structured, commonly used and machine readable format”, giving CSV, XML and JSON as examples. Your business customers will also need to get records out when their contract ends: the ICO lists “end-of-contract provisions” among the minimum terms of a controller and processor contract. A CSV export per data type is a small build and answers both.

Billing correctness

Overcharging a customer once does more damage than a missing feature. Use an established billing provider, test upgrades, downgrades, failed payments and cancellations, and make sure invoices show VAT correctly. If you sell to consumers, note that new subscription contract rules are due to come into force in January 2027, according to the government’s August 2026 announcement. They require clearer up-front information, regular reminders, easier cancellation and a 14-day cooling-off period after a free trial ends or a long contract renews.

New UK subscription rules for consumer products
Jan 2027when the new subscription contract rules are due to start
14 dayscooling-off period after a free trial or a long-term renewal

Source: Prime Minister's Office, 9 August 2026

For how recurring card and Direct Debit payments work, see taking subscription payments in the UK. Our guide to SaaS pricing models covers how to present prices.

How to write a scope document

Seibel’s advice is to “time box your spec”, “write your spec” and “cut your spec”. A one or two page document is enough.

A one-page MVP scope

Step 1 of 6

Job

Describe the customer and the job they hire the product to do, in one paragraph.

Step 2 of 6

Workflow

Write the core workflow as numbered steps from sign-up to outcome.

Step 3 of 6

In and out

List what is in scope, what is done by hand and what is deliberately left out.

Step 4 of 6

Basics

Name the non-negotiables: security, data export, correct billing and accessibility.

Step 5 of 6

Success

Set the result and date that would prove the idea works.

Step 6 of 6

Deadline

Fix the date and agree that features get cut before it moves.

In more detail, include:

  1. The customer and the job, in one paragraph.
  2. The core workflow as numbered steps, from sign-up to outcome.
  3. In scope: each feature, with one line on why it’s needed.
  4. Done by hand: what you’ll do manually and when you’d automate it (for example, “after 30 customers”).
  5. Out of scope: the features you’ve decided not to build yet. Writing these down stops them creeping back in.
  6. Non-negotiables: security, data export, billing, accessibility basics.
  7. Success measure: what result, by what date, would tell you the idea works. For example, ten customers paying after their trial.
  8. Deadline, and the rule that features get cut before the date moves.

Share it with whoever is building and agree that any addition means something else comes out. For the wider sequence from idea to launch, see how to build a SaaS product.

Common questions

How do I know if my MVP is too small?

If customers can’t reach the outcome they care about without you stepping in at a point you didn’t plan for, it’s too small. If they reach it and still don’t come back, the problem is probably the idea or the customer, not the scope.

Should the MVP include AI features?

Only if AI is the core of the job. Otherwise, add it once the basic workflow is proven. AI features carry a running cost on every use, which affects pricing from day one. We explain how that bill is worked out in what AI costs to run.

Getting help with scope

If you want an outside view on what belongs in your first version, our SaaS product development team can review a draft scope with you.

Sources

→Keep reading

Related guides.

All insights
Software reviews

Cursor vs GitHub Copilot vs Claude Code: which AI coding tool fits your team?

Cursor, GitHub Copilot and Claude Code compared on published prices, plans, usage limits and developer survey data, with a best fit verdict for UK… Read
AI explained

RAG explained: how “chat with your documents” AI really works

Retrieval-augmented generation lets AI answer from your own files. How RAG works, what it costs, why it still gets things wrong and when to… Read
AI explained

Open-source AI models explained: Llama, Gemma, DeepSeek and the rest

What open-weight AI models are, how the Llama, Gemma, DeepSeek, Qwen, Mistral and gpt-oss licences differ, and when a UK business should use one. Read

→Get in touch

Got a question? Ask us.

Tell us what you’re trying to do. Every project starts with a free audit: someone who does the work replies with a few ideas, an honest view and a written plan with prices.

What happens next

  1. 1
    You send the formIt takes about two minutes.
  2. 2
    We do our homeworkWe look at your website and your market.
  3. 3
    You get a short planWhat we’d do first, with prices. Yours to keep.
  • Free, with no obligation
  • We won’t add you to a mailing list

We only use your details to reply to you. See our privacy policy.