Scoping is the work. Not the sales call.
Most studios treat scoping as unpaid overhead to get through before the real project starts. We think it is the project — the part that decides whether everything after it goes well.
Nobody’s estimate is wrong on purpose.
01 — The problemProjects don’t overrun because someone did bad arithmetic. They overrun for four reasons, and all four are findable in advance.
The integration
“It just needs to talk to our existing system” is four weeks or four months, and nobody knows which until someone looks.
The edge case
The 3% of records that don’t fit the model. They’re never 3% of the work.
The assumption
Everyone pictured something slightly different and nobody wrote it down.
The obligation
Regulatory, contractual or accessibility requirements found after the architecture was set.
What we actually interrogate.
In order, and in the open — you’re in the room for all of it. The output isn’t a proposal written at you; it’s a document the two of us built.
- 01The business caseWhat has to be true for this to be worth building at all — and what happens if you build nothing.
- 02The users, specificallyNot “customers”. The three or four actual roles who will touch this, and what each is trying to finish.
- 03The critical journeysThe handful of paths that must work perfectly. Everything else is negotiable, and most of it should be.
- 04The hard edgesIntegrations, data you don’t control, compliance, and anything with a third party’s name on it. This is where the time goes.
- 05The cutWhat release one contains. Usually a third of what you arrived with, and better for it.
- 06The numberA range, a timeline, and every assumption it rests on, written down where you can argue with it.
1. Scope
The platform replaces the current process, which currently runs across email and . Release one covers request intake, assignment, estimate approval and report generation. Explicitly not in release one: client self-service, automated , historical data migration beyond 12 months.
2. Hard edges
Report output must satisfy submission format — externally defined, not negotiable. Two user populations with different auth requirements. The integration is undocumented; assume discovery time.
3. Assumptions this estimate rests on
Report templates supplied as by end of week one. One nominated decision-maker available for weekly review. No change to the requirement mid-build.
Sample artefact — real structure, client details redacted
What you actually walk away with
03 — What you receiveNot a slide deck. A working document, written the way we’d write it for ourselves — because it is what we’d write for ourselves. It belongs to you: take it to another developer, use it to raise money, or put it in a drawer.
- ✦Product definitionScope, users, journeys, and the explicit list of what is not in release one.
- ✦Technical approachArchitecture, data model sketch, integrations, and the decisions that are expensive to reverse.
- ✦Risk registerWhat could move the number, how likely it is, and what we’d do about it.
- ✦Costed planPhases, a timeline, and a range — with the assumptions attached to each line.
It will. Here’s what happens.
04 — When the scope moves- ✦You hear about it the day we doNot at the end of the sprint, and not in a status report. If something has changed, you’ll know immediately.
- ✦It comes with options, not just bad newsAbsorb it, defer something else, or extend. Costed, so it’s your decision to make.
- ✦The document gets updatedThe scope is a living artefact. If it stops matching reality, it stops being useful.
- ✦We say when we got it wrongSome overruns are ours. When they are, we say so, and we don’t bill you for our own mistake.
Useful, and not the point.
05 — Where AI actually helpsWe use AI heavily, and it genuinely compresses the work — exploring approaches, drafting and stress-testing scope documents, generating throwaway prototypes to test an idea in hours rather than days, and moving faster through the mechanical parts of a build.
What it does not do is decide what to build, or take responsibility for the result. The judgement is the job. AI makes a good scope faster to produce; it does not make a bad scope good, and an agency that leads with it is usually selling speed because it hasn’t got anything else.
Start with a Scoping Sprint.
A fixed-price, fixed-scope engagement that turns your idea into a document you own — including the number. If you never work with us again, you still leave with something useful.
Tell me what you’re trying to build.
Half an hour. No deck, no discovery theatre. Bring an idea on the back of an envelope or a system that is falling over — either way you will leave the call knowing what I think it takes, and whether I’m the right person to do it.
Zeki Mirza — Founder
Every call is with the person who does the work