Decide What the First Version Has to Prove, Then Build Only That
Most conversations about a minimum viable product start with features. The useful ones start earlier: whether this should be built yet, and what specific thing it has to demonstrate before the next tranche of money or the next customer commitment arrives. We scope backwards from that, quote it in sterling excluding VAT, and finish it.
The Cheapest Project Is the One You Decide Not to Start
A software firm telling you not to build software is unusual enough to be worth explaining. A first version is not only a cost; it is three or four months of founder attention, followed by a maintenance obligation that does not end. If the question you are trying to answer can be answered without code, answering it without code leaves you with more money and a considerably better brief.
So the first call is largely diagnostic. What decision are you trying to reach? Who has to behave differently for you to call it a success? What would have to be true for you to abandon the idea? Founders who can answer those three get a better build, because the scope more or less writes itself. Founders who cannot usually discover around week seven that they are paying to explore a question rather than answer one.
How a build actually runsFour Tests That Cost Less Than a Single Sprint
None of these replace building. All of them sharpen what gets built, and occasionally one ends the project early, which is a good outcome bought cheaply.
A Page and a Small Ad Budget
One landing page describing the product as though it already exists, with a real enquiry form behind it. A few hundred pounds of search or LinkedIn spend tells you what one interested person costs to reach, which is the number that decides whether the business model works at all.
Run It By Hand First
Deliver the service manually for ten customers using a shared inbox, a spreadsheet and a phone. Letting agents, recruiters and private clinics can often run an entire offering this way for a month. You learn the real workflow, the exceptions, and which single step people would pay to remove.
Existing Tools, Wired Together
A database tool, a form builder and an automation service will carry a surprising amount of an internal process for a few months. It will not scale and it will irritate you, and both are useful: the points where it breaks are precisely the features worth paying to build properly.
Something Signed
A letter of intent, a paid pilot, or a purchase order conditional on delivery. In UK business-to-business software this is the most persuasive artefact you can bring to a build and to a funding conversation. It also forces a specific scope, because somebody has now written down what they expect.
If two or three come back positive, the brief is usually obvious. If all four are ambiguous, a fixed-scope build is an expensive way of continuing the ambiguity.
An MVP for Investors and an MVP for Customers Are Different Builds
They overlap in the middle and diverge sharply at the edges. Knowing which one you are commissioning prevents the most expensive kind of scope argument.
| Question | Built to convince investors | Built to serve customers |
|---|---|---|
| What it has to prove | That the difficult part is genuinely solved, and that the market responds when it is put in front of them | That somebody can complete the whole job unaided, on a bad day, without ringing you |
| Where the money goes | The core mechanism, a credible demonstration path, and analytics that stand up to questioning | Onboarding, empty states, error recovery, permissions, and the admin screens that let you put things right |
| Acceptable rough edges | Manual steps behind the scenes, a small number of supported scenarios, limited device coverage | Very few — an unrecoverable error in front of a paying user costs more than the feature saved |
| Data and compliance | Still real: UK GDPR applies from the first live user, whoever that user is | Real and visible: subject access, deletion, retention, and a privacy notice somebody might actually read |
| How success is judged | Evidence assembled for a specific meeting on a specific date | A usage pattern that holds for several weeks without you pushing it along |
| Typical fixed-scope range | Roughly £12,000 to £30,000 excluding VAT | Roughly £20,000 to £45,000 excluding VAT |
Ranges are indicative and depend on platforms, screen count, and whether payments, real-time features or an admin area are in scope. The cost calculator narrows it in about a minute, and the pricing page explains what sits behind the number.
What the Next Cheque Expects to See
UK early-stage money arrives on its own timetables and to its own evidence standards. Scoping a first version without knowing which one you are aiming at is how builds get finished a month after the decision they were meant to inform.
Angels and Pre-Seed Funds
They are buying the founder and the wedge. What helps is a working core they can hold, two or three honest numbers, and a plain statement of what the money buys next. A long feature list does not help. Expect diligence on who owns the code, which is why repositories and cloud accounts sit in your company name from week one rather than ours.
SEIS and EIS-Backed Rounds
A large share of UK angel money arrives through these schemes, and the paperwork has its own rhythm. Advance assurance from HMRC is worth holding before the conversation starts, and it takes weeks rather than days. In practice the build budget is often released in one lump after a close, so a scope that can start small and extend cleanly matters more than usual.
Grants and Innovate UK Style Programmes
Competitions open and close on published dates, assessment takes a further period after close, and awards are typically paid quarterly in arrears against costs already incurred. That is a cash-flow pattern rather than a budget. We usually propose a small self-funded slice that evidences the technical claim, with the larger scope timed to begin once the award is confirmed.
One practical consequence of all three: put the date of the decision in the scope document. If the round closes in March, a build that lands in April is a different product, and we would far rather cut features in January than miss it. That single line has reshaped more of our scopes than any feature discussion.
Build, Measure, Decide — And Be Willing to Decide No
The loop is only worth running if the last stage is genuine. Teams that treat the decision point as a formality have bought a slow way of building the thing they always intended to build.
Build one slice, not one feature
The smallest complete path a real person can walk end to end, with the measurement for your question already inside it
Put it in front of people who are not you
A pilot group, a handful of customers, one branch or one depot — anybody whose praise is not socially obligatory
Measure the number you named in advance
Named before the build, so the result cannot be quietly reinterpreted afterwards into whatever the data happened to show
Decide: continue, change direction, or stop
Written down, with a date against it. Stopping is a legitimate outcome and the only one that saves a whole budget
Fortnightly sprints mean this loop can run three or four times inside a typical first version, rather than once at the end when nothing can be changed cheaply.
What Earns a Place in the First Release
A rough rule: if removing it would not change what you learn, and a person could do it by hand for the first fifty customers, it waits.
One Complete Journey
Sign in, do the main thing, see that it worked. One user type, one path, no branches. If you have three kinds of user, version one usually serves the one whose behaviour you most need to observe.
An Admin View You Can Fix Things From
Consistently underestimated. Without somewhere to look up a record, reset a state or correct a mistake, every support issue becomes a developer request and the founder becomes the helpdesk inside a fortnight.
The Data Decisions, Settled
What personal data you hold, where it lives, how long you keep it and how a deletion request gets serviced. UK GDPR applies to your first ten pilot users. Retrofitting this later is one of the genuinely expensive rewrites.
Measurement That Answers Your Question
Not a dashboard of everything. Two or three events that tell you whether the behaviour you predicted is happening, plus crash reporting so you learn about failures before your users describe them to you.
Payments, Only If Payment Is the Question
If you need to know whether people will pay, take money in version one — Stripe for cards, GoCardless or Direct Debit for recurring business-to-business billing. If pricing is settled and the question is usage, invoicing by hand for a quarter is entirely respectable.
A Route to a Human
Early users forgive a great deal if somebody answers. They forgive nothing if the only visible contact route is a form that appears to go nowhere. One in-product link to a monitored inbox is enough.
How a Fixed Scope Gets Written Down
A fixed price only means anything if the scope behind it is specific enough to argue with. This is the sequence we use, and it is the same whether the figure is twelve thousand pounds or fifty.
A Half-Hour Diagnostic
What decision the build has to inform, who has to change their behaviour, and the date that matters. You get our honest view on whether to build yet, including on the occasions when that view is no.
A Written Scope, Screen by Screen
Screens, user types, integrations and rules, with the assumptions stated openly. Anything ambiguous is flagged as an assumption rather than absorbed quietly, because silent assumptions are where fixed prices go wrong.
One Sterling Figure, Staged
A single price excluding VAT, split across milestones with payment against each. VAT is added at the prevailing rate on invoices. Purchase order references and 30-day terms are entirely normal here.
A Change Process Agreed in Advance
New requests get assessed and priced in writing. You approve, swap for something of similar size at no cost, or park for version two. Settling this before the build removes the awkward conversation from the middle of it.
The Point an MVP Stops Being an MVP
There is a moment when the term starts doing damage, because it is being used to excuse things that now cost you customers. These are the signals, and we will say so when we see them.
When several of these are true at once, the honest next step is a short hardening phase rather than another feature sprint: the deliberately manual parts get automated, the architecture is sized for the coming twelve months, and monitoring and backups stop being aspirational. It is unglamorous work and it is considerably cheaper than the alternative, which is discovering the limits during your first serious month.
From there most clients move onto a maintenance and support retainer, and some take a dedicated team for a roadmap that keeps moving. If your first version came from somebody else and has already reached this point, project rescue starts with a fixed-price audit and a straight verdict.
The Pages Founders Read After This One
Cost Calculator
An indicative sterling range in about a minute, with a plain explanation of what pushes the figure up or down.
Open itDesign & Prototyping
A clickable version to sign off, with accessibility treated as a design input rather than a late audit.
Learn moreFractional CTO
Architecture calls, supplier oversight and technical diligence support for founders without a technical co-founder.
Learn moreWhat Founders Ask Us About First Versions
Our investors want to see traction. What counts as traction at UK pre-seed?
Less than founders expect, and more specific. At pre-seed most UK angels and early funds are looking for evidence that a defined group of people will change their behaviour because of your product: signed pilot agreements, a waiting list that converted, a fortnight of usage data from twenty real users, or one paying customer with a purchase order. Raw download numbers rarely move anybody. We scope the build so the one or two numbers your round will be judged on are instrumented from the first release rather than reconstructed afterwards.
We have applied for an Innovate UK grant. Should we wait for the decision before building?
Usually you build something small in the meantime, because grant competitions run to published opening and closing dates with an assessment period after close, and the money itself is normally paid quarterly in arrears against costs you have already incurred. That is a cash-flow shape, not a budget. The sensible pattern is a modest self-funded slice that proves the technical claim in your application, then a larger scope that starts when the award is confirmed and your quarterly claims can support it.
Can one build serve both investors and customers?
Partly, and it is worth knowing which parts pull in opposite directions. An investor build is judged on whether the hard bit is real and whether the numbers are believable, so it rewards a working core and clean instrumentation. A customer build is judged on whether somebody will use it on a Tuesday afternoon without help, so it rewards onboarding, error handling and support. Where those conflict we would rather build for customers and present the evidence to investors, because customer evidence survives the meeting and a demo does not.
What happens to the code if our round does not close?
It stays yours. Repositories, cloud accounts and store listings are in your company name from the first week and intellectual property transfers to you on final payment, so a round that stalls leaves you with an asset rather than a dependency. We will also tell you honestly what it costs to park a build: hosting can usually be reduced to a very small monthly figure, and a dormant codebase is realistically cheap to restart for about a year before framework and platform updates start to charge interest.
How do you stop an MVP quietly turning into a full product?
By writing down at the start what the version is meant to prove and by pricing every addition against that sentence rather than against enthusiasm. Each change request gets assessed and quoted in writing, and you can approve it, swap it for something of similar size at no cost, or park it for version two. The swap is the one most clients use. It is a dull mechanism and it is the main reason our first versions get finished.
Do we need the company and the IP sorted before you start?
You need a registered company at Companies House for the accounts to be opened in the right name, and that is genuinely worth doing before week one because moving an Apple or Google account between legal entities later is tedious. Everything else can run in parallel. We will sign a mutual NDA before a scoping call if you want one, and our standard terms assign intellectual property to you, which is the wording most UK investors and their solicitors will look for during due diligence.
Tell Us What the First Version Has to Prove
Send the idea, the decision it has to inform and the date that matters. You get a written scope and a sterling range excluding VAT within two working days — including an honest view on whether to build yet.