How we work
The engagement model, stated plainly
Four phases, each with an exit you can refuse to sign, and a set of commitments we would rather be held to than describe vaguely.
Most engagement models are written to be unfalsifiable. Ours is specific enough that you can tell when we have failed to follow it, which is the point of writing it down.
Engagement
The four phases
01
Discovery
Two weeks, fixed price. We interview your users and your team, read your code if there is any, and come back with a scope, a risk register and a number. You can stop here and take the output to another firm — several clients have.
- Costed roadmap
- Risk register
- Architecture outline
- Go / no-go recommendation
02
Foundation
Repository, pipeline, environments, design system and one vertical slice through the whole stack, deployed. If the architecture is wrong, this is where it becomes obvious, while changing it is still cheap.
- Deployed skeleton
- CI/CD pipeline
- Design system v1
- Architecture decision records
03
Build
Two-week iterations against a prioritised backlog. Demo every fortnight to whoever is paying, with the burn-up chart visible. Scope is reviewed at each demo, so change is a conversation rather than a change request.
- Fortnightly releases
- Live demo
- Updated burn-up
- Reviewed scope
04
Handover
Pairing, runbooks and a deliberate taper in our involvement until your team owns it outright. Acceptance is your engineers running a release without us in the room.
- Runbooks
- Onboarding sessions
- Support taper
- Signed acceptance
Commitments
How we behave when it is inconvenient
Say the difficult thing early
If an estate is worse than the brief suggested, or a use case will not pay back, we say so in week one. Delivering a project we know is wrong is the more expensive kind of politeness.
Leave nothing hostage
Documentation, runbooks and pairing are contractual deliverables. Our commercial interest in being hard to replace is real, and we have decided not to act on it.
Measure, then argue
Opinions about performance, quality and cost are cheap. We instrument first, so the disagreement is about what the number means rather than what it is.
Boring where it counts
Novel technology in the parts that make money is a decision that needs defending. We save the interesting choices for where they change an outcome.
Small batches, always
Large releases concentrate risk and hide causation. Everything we build ships in pieces small enough to reverse without a meeting.
The work is the credential
We would rather show you a system in production than a certification. Every claim on this site is one we can point at.
FAQ
Questions we are asked
Then we say so, and you have spent two weeks and a fixed fee instead of two quarters and a budget. This has happened often enough that we consider it a feature of the model rather than an awkward edge case.
Start with two weeks and a fixed fee
Discovery gives you a scope, a risk register and a number. You are free to take it elsewhere.