Straight answers.

Fixed scope for defined projects, monthly retainers for ongoing marketing, SEO and support. You get a written proposal with deliverables, timeline and total before any work begins, and we do not invoice for anything that was not agreed in writing. We do not publish a price list because the honest answer always depends on scope — but tell us your budget and we will tell you plainly whether it is realistic.

For builds, we are usually not the right fit below about $6,000, because the discovery and handover work we insist on does not compress well. Retainers start at three months. If your budget is smaller than that, say so — we would rather point you somewhere sensible than take work we cannot do properly.

Two to three weeks of research, architecture and prototyping, sold as a standalone piece of work. It ends with a plan detailed enough that you could hand it to any competent team — including one that is not us. Clients use it when the problem is clear but the solution is not.

Projects run to their scope. Retainers have a three-month initial term, because most of what we do needs that long to produce a meaningful signal, and are month-to-month with 30 days' notice afterwards.

Typically a deposit to book capacity, then payments at agreed milestones for projects, or monthly in advance for retainers. We invoice in your currency where we can and are happy to work with your procurement process.

A marketing website is usually 4-7 weeks. E-commerce runs 6-12 weeks depending on catalogue size and integrations. Mobile apps start around 12 weeks. AI automation pilots are deliberately short — we aim to have something running in 2-3 weeks so you can judge it on evidence rather than on a proposal.

Discovery can usually begin within one to two weeks. Full build capacity is typically three to five weeks out, because we deliberately do not over-book the team. If you have a hard deadline, tell us on the first call and we will be straight about whether it is achievable.

Roughly two to four hours a week during discovery and design, dropping to about an hour a week during build. The heaviest asks are content and decisions, both front-loaded. Projects that run late almost always do so because sign-off or content stalled.

Within a sprint boundary, easily — that is much of the point of working in fortnights. A fundamental change means re-scoping, which we quote honestly rather than absorbing silently and running out of budget at 80% complete.

Frequently, and it usually works well. We often take the specialist half — the AI work, technical SEO, a design system — while your team keeps what they already do well. We work inside your tools and rituals rather than imposing ours.

Yes, entirely. Repositories, hosting, domains, ad accounts, analytics and design files are created in your name or transferred to you at handover. We do not hold anything as leverage, and leaving us should be boring rather than painful.

Every build includes a 30-day warranty on defects against the agreed scope. Beyond that, most clients move onto a care plan covering hosting, updates, monitoring, backups and a monthly block of improvement hours. It is optional and month-to-month.

Usually. We start with a paid audit of the codebase, hosting and dependencies, then give you an honest read: maintain, refactor or rebuild. Sometimes the answer is rebuild, and we will show you the reasoning rather than simply assert it.

On a care plan: critical issues within 2 hours during business hours, everything else within one business day. You get a named engineer rather than a shared inbox, and a status page for anything we host.

Fair question. We only recommend AI where there is a repetitive, high-volume task with a measurable cost — quoting, triage, support answers, content operations. We scope a narrow pilot, measure it against how the work is done today, and drop it if the numbers do not hold up. Roughly one in five ideas we assess does not survive that test.

Not on the configurations we deploy. We use enterprise API tiers where inputs are excluded from training, and we document exactly what is sent to which provider. Where data is too sensitive to leave your environment, we can run open models on infrastructure you control.

Retrieval grounding so answers come from your documents, citations so any claim can be checked, an evaluation set that catches regressions, refusal behaviour when confidence is low, and human review on consequential outputs. Not eliminated — measured, bounded and monitored.

We model cost per request during the pilot and give you a projected monthly figure at expected volume, with a hard cap and alerting. For most business workloads the running cost is far smaller than clients expect; the build is the significant number.

Yes — clients across 18 countries so far. We agree a shared overlap window at kickoff, keep everything in one channel, and record any meeting you cannot attend. Asynchronous by default, live when it genuinely helps.

Retail and e-commerce, healthcare, professional services, finance, logistics, education and hospitality. What matters more than sector is whether you have a real customer problem and someone internally who can make decisions.

Yes, routinely. We can send ours or sign yours before the first call if you would prefer to discuss specifics straight away.

We turn down roughly one project a month. Usually it is a budget that cannot deliver the outcome being asked for, a timeline that would require cutting the parts that make the work good, or a brief where we are simply not the strongest team available. We would rather say so than take the money.