HOW THE WORK HAPPENS.

//Three phases, priced separately#

Discovery, build, launch. Each is estimated and invoiced on its own, which is not a billing convenience — it's an admission that an estimate for phase three written before phase one exists is fiction. By the end of discovery I know enough to quote the build accurately. Before it, I'd be adding a margin to cover my own ignorance and calling it a price.

//What I need from you#

One person who can make decisions. Not a committee that reviews fortnightly — a named individual who can answer a question within a day. This single factor predicts project outcomes better than budget, team size or technical complexity in my experience. The projects that went badly all had ambiguous ownership.

I also need access to real data and the real content early. Building against placeholder content is building against a fiction that flatters the design, and the reckoning always comes eventually. It's cheaper in week two than in week nine.

//Where my estimates go wrong#

Consistently in two places. Content migration always takes longer than anyone plans, because the old content is messier than the audit found. And third-party integrations expand — a payment provider, an analytics vendor, a consent platform each turn out to have an undocumented quirk. I now pad both deliberately and say so rather than discovering it under deadline.