Web Application Development Where Performance Is an Acceptance Criterion

A focused internal tool is six to ten weeks. A customer-facing product is three to five months. A multi-tenant SaaS platform with billing and onboarding runs five to eight. You get a range at the end of discovery, not a guess on the first call.

We put specific targets in the acceptance criteria: LCP under 2.5s, INP under 200ms, CLS under 0.1, measured on a mid-range Android over 4G. They are enforced in CI so a regression fails the build. What we cannot control is third-party scripts your marketing team adds after launch, and we will be explicit about that.

Yes, that is the performance rescue engagement at $6,500 or about £5,200. We use field data rather than only Lighthouse, because lab scores routinely disagree with what real users experience, and we leave a CI budget behind so the fix holds.

WCAG 2.2 AA is our baseline, built in from the first component rather than retrofitted. If you sell to US public sector, UK public bodies or large enterprises, you will eventually be asked for a VPAT or an accessibility statement, and retrofitting costs several times more.

Usually React with Next.js, because the ecosystem and the hiring pool are deepest, which matters when you eventually maintain it without us. If your team already runs Vue or Angular well, we use that instead. The framework is rarely the thing that decides a project.

Yes, if it should be indexed. Anything public gets server rendering or static generation with structured data and clean URLs. Client-only rendering for public pages is one of the more expensive mistakes in this space and it is entirely avoidable.

Yes, and it is one of our more common requests. Semantic search and in-context drafting are usually two-sprint additions. We instrument cost per request from day one so the bill does not surprise you when usage grows.

We can, or we work to your existing design system, or we build you one. If you have a designer, we would rather collaborate with them than replace them. Discovery week establishes which of those is happening.

Your accounts, your infrastructure, in your name. Typical hosting for a mid-sized application runs $150 to $600 a month on AWS or Vercel. We set it up with infrastructure as code so it is reproducible and portable.

A support arrangement sized to what you need, from a small monthly block of hours to a full SLA. Or a clean handover to your team, which we plan for from day one either way.

Yes. Our US office is in Oakland, California, and delivery hours span Pacific through GMT. For UK clients that means most of your working day overlaps with the team, not just the first hour.