Week one: discovery and structure
We do a single working session to lock positioning, target audience, and primary conversion event. From there, we draft the site map, page-level goals, and content outlines before any design begins. Most over-runs come from skipping this step.
Deliverables in this phase are concrete: messaging hierarchy, required pages, conversion paths, and technical constraints (integrations, forms, compliance, and analytics). Ambiguity removed here prevents expensive churn in weeks two and three.
We also define decision ownership early. Who approves copy, who approves visuals, and what turnaround time applies. Shipping fast is rarely blocked by design skill; it is usually blocked by unclear approval mechanics.
Week two: design system and core pages
We build the design system in code, not in a design tool. Tokens, typography, components — all in the actual stack. Then we ship the homepage and one inner template end to end so the client can see real, scrollable pages instead of static mockups.
Building in production-adjacent code eliminates translation loss between design and engineering. Spacing, breakpoints, interactions, and accessibility behavior are validated as they are created, not deferred to a later handoff.
By the end of week two, stakeholders are reviewing functional artifacts with realistic content and navigation. Feedback quality improves because comments target user flow and clarity rather than abstract styling opinions.
This phase also sets reuse economics. A robust section/component library means later pages are assembly, not invention. That is the core reason three-week timelines are feasible.
Week three: remaining pages, polish, launch
With the system in place, the remaining pages compose quickly from existing components. We spend the back half of week three on accessibility, performance, SEO basics, and a structured QA pass on real devices before launch.
QA is run as a checklist, not a vibe check: semantic heading order, keyboard navigation, contrast, mobile viewport validation, form state handling, metadata coverage, and analytics event smoke tests.
Launch prep includes rollback confidence. We confirm deployment settings, environment variables, domain mappings, and monitoring hooks so post-launch issues are detectable and reversible within minutes.
After go-live, we monitor first-week behavior closely: key path drop-offs, form completion rates, and page speed deltas. Fast shipping matters, but stabilization discipline is what protects outcomes.