One accountable person
A named delivery lead from proposal to a year after launch. Not a rotating account manager, and not a shared inbox.
How we work
A process you can hold us to.
Five stages, each ending in something you can open and check. The standing rule underneath them is simple: you should never have to ask what is happening on your project.
Why the process looks like this
Most agency processes are drawn to look reassuring on a slide. Ours came from thirteen years of noticing where projects actually go wrong, and putting a checkpoint at each of those places.
Projects fail at the start, not the end. The expensive mistakes are almost always decisions made in week one on incomplete information: the wrong platform, a content model that cannot hold the real catalogue, a redirect map nobody checked. So we spend real time on discovery and we write the findings down, in a document you keep whether or not you hire us. If discovery says you do not need the thing you asked for, we will tell you, twice this year that has meant talking a client out of a rebuild they had already budgeted.
A plan nobody can read is not a plan. You get a scope document with the assumptions listed next to the price, a delivery schedule with dates rather than "phases", and a definition of done for each deliverable. Something always changes on the way, and when it does we quote it as a written change request before the work starts, rather than as a surprise on the invoice.
Build and run are the same team. In most agencies the build team disappears at launch and a support desk inherits a system it has never seen. We do not split those roles. The engineers who ship your site are the ones who run it afterwards, which is why post-launch fixes take hours instead of weeks and why nobody bills you for "familiarisation".
Measurement decides what happens next. Before anything goes live we agree what success looks like as numbers: organic sessions to money pages, cost per qualified enquiry, checkout completion rate, Core Web Vitals. Then we instrument for those numbers. After launch the monthly report starts with them, says plainly whether they moved, and proposes the next thing to try. If one went the wrong way, that is the opening line, not a footnote on page nine.
Improvement is a loop rather than a phase. A website is never finished; it is only current. The last stage feeds straight back into the first, and the backlog is scored on impact against effort so next month's work is chosen by evidence rather than by whoever shouted loudest. You can see that backlog, and you can reorder it.
Underneath all five stages, three habits do most of the actual work: one named person accountable for your account, everything written down where you can see it, and bad news delivered early. The last one is the hardest and the one clients thank us for most.
Discovery
2 wks
written brief
Sprints
2 wks
each, staged
Post-launch
30 days
fix window
Every stage ends with something you can open and check
The five stages
What happens, and what you get.
Timings below are for a typical marketing website or a retainer's first quarter. Larger platform work stretches stages 1 to 3; the shape stays the same.
01
DiscoveryWeek 1–2
We audit what exists: technical crawl, analytics, rankings, competitors, and the interviews nobody enjoys but everybody needs. We look at your sales data, not just your traffic, because the goal is enquiries that close.
You get- A findings document with prioritised problems, not a 90-page crawl export
- A competitor and keyword landscape with the realistic opportunities marked
- A recommendation on platform and approach, including the option of doing nothing
- Yours to keep, whether or not you continue
02
PlanWeek 2–3
Scope, sitemap, content model, wireframes for the templates that matter, and the measurement plan. Everything is written down and priced with the assumptions visible, so the number does not move unless the scope does.
You get- A scope document with a definition of done per deliverable
- A dated delivery schedule and the list of what we need from you, when
- The measurement plan: which numbers we are moving and how they are tracked
- A fixed price, valid for 30 days
03
Build and runWeek 3–10
Two-week sprints with a staging link that is always current. For retainers this is the campaign work itself: pages shipped, ads live, content published. You see progress continuously instead of at a reveal.
You get- A staging environment updated every sprint, with a written change log
- A fortnightly call with the engineer doing the work
- Two rounds of consolidated revision per deliverable, built into the price
- A launch checklist: redirects, tracking, performance budget, accessibility pass
04
MeasureFrom day one, monthly after launch
GA4, Search Console and call or form tracking configured before launch, not after. The monthly report opens with the numbers we agreed in stage two and says whether they moved.
You get- A dashboard you can open yourself, in accounts you own
- A one-page monthly report: what moved, what did not, what we are doing about it
- 30 days of post-launch defect cover at no charge
- An honest note when a result came from seasonality rather than from us
05
ImproveOngoing
The backlog is scored on impact against effort and reordered every quarter with you in the room. Tests are run one variable at a time so we can tell what actually caused the change.
You get- A visible, reorderable backlog, you can move things up
- A quarterly strategy session with the next 90 days agreed
- Experiment write-ups, including the ones that failed
- Platform, security and dependency updates applied in staging first
The rules under it
Four things that do not change, whatever the project.
Written down, where you can see it
Scope, decisions, change requests and reports live in a shared space you keep access to, including after the engagement ends.
Bad news early
If a date is going to slip or a tactic is not working, you hear it in the week it becomes clear, with what we propose to do instead.
You own everything
Accounts, code, designs and data are in your name from day one. On the final payment the IP assigns to you, and we hand over inside 10 working days.
Questions
How the process works in practice.
How long does a typical website project take?
Eight to twelve weeks for a marketing site of 8–20 pages, discovery to launch. Ecommerce builds and platform migrations run 12–20 weeks depending on catalogue size and integrations. The biggest variable is not our capacity. It is how fast content and approvals come back from your side.
What do you need from us to keep it on schedule?
One named decision-maker who can approve work, content and logins within five working days of a request, and feedback consolidated into a single response per review round. When we are blocked for more than ten working days we re-plan the dates and tell you why.
What happens if we want to change the scope mid-project?
You can, and most projects do. We quote the change as a written change request with its own price and schedule impact, you approve it, and only then does it enter a sprint. Nothing gets built and invoiced without that approval.
Do we get access to the staging site during the build?
Yes, from the first sprint. It is updated every two weeks with a change log, so you are never waiting for a big reveal and we are never a month down a path you would have vetoed.
What happens after launch?
Thirty days of defect cover at no charge, then either a care plan or a monthly programme; both are priced on the pricing page. Either way the same engineers stay on the account. There is no support desk that has never opened your codebase.
Can we start with just discovery?
Yes, and several clients do. Discovery is priced and delivered on its own, the document is yours to keep, and you are free to take it to another agency. We would rather you make the decision with good information than sign a build you were not sure about.
Start here
Stage one is a conversation.
Thirty minutes, no deck. We will tell you what we would look at first and roughly what it would take, before anyone signs anything.