Why I rebuild websites in twelve weeks, not twelve months

I have been on both sides of the year-long website project. The agency side, where the scope grows every time a stakeholder joins a call. The client side, where the launch date slips a quarter at a time and the old site keeps losing rankings while everyone argues about the hero image.

This summer I rebuilt two professional-services websites in twelve weeks, from the first day on the job to both of them live. Not because the team was large. Because the constraints were tight. Here is what those constraints were, and why I would use them again.

Start with the URL map, not the mockup

The first deliverable was a spreadsheet, not a design. Every existing URL, its traffic, its rankings, its backlinks, and the page it would become. Legacy sites accumulate triple-nested paths and near-duplicate service pages, and the single biggest SEO win of a rebuild is flattening that structure while keeping every redirect honest.

Doing this first also surfaced the real problem: two sister sites had been competing for the same searches. Roughly forty service pages were reassigned between the two domains by search intent, so each query had one clear home. That decision drove the information architecture, the navigation and the content plan. A mockup made before that decision would have been wrong.

Decide what the site must do for a machine

In 2026 a site has three readers: a person, a search crawler and an answer engine. The person needs clarity and a fast path to contact. The crawler needs clean structure and internal links. The answer engine needs something to cite: a named author with credentials, a direct answer near the top of the page, and structured data that says what the page is.

So every page got a byline and a bio. Every service page got schema. Every long-form page got a short answer before the long explanation. None of this is visible in a design review, and all of it is why the rebuild was worth doing.

Build a content owner into the plan

Websites die because nobody owns the words after launch. Before the first page was built, the editorial process existed: who writes, who verifies, who approves, how fast. Attorney bylines are not a design flourish; they are an accountability system. If a name is on the page, the person behind it reads the page.

Cut the things that do not change behavior

Here is what we did not do. No custom illustration program. No animation beyond what the template did already. No brand refresh in parallel. No "phase two" wishlist carried into scope. Each of those is a fine project. None of them affects whether a prospect finds the page, trusts it and calls, and each one adds a month.

Why twelve weeks and not six

Two reasons. Migration is where sites get hurt, and a careful redirect map, a staging-to-live checklist and a week of post-launch monitoring are not optional. And the content had to be verified, not just written. On a professional-services site, an unverified credential or a stale media mention is a liability. Speed came from sequence and scope, not from skipping the checks.

What I would do differently

Start the glossary and the evergreen content cluster on day one, in parallel with the architecture work, instead of after launch. The structural rebuild gives you a clean house; the definitional content is what fills it with visitors, and it has the longest ramp.

The principle underneath

A website is a system for being found, being trusted and being contacted. Everything in a rebuild should be judged against those three jobs. When you hold to that, twelve weeks is plenty, and twelve months is a sign that someone is building something else.