Most companies buy a website the way they buy office furniture: as a capital project with a launch date and a completion state. Then they are surprised when conversion flatlines eighteen months later.
A website is not a deliverable. It is the highest-intent surface your brand owns, and it should be operated like a product—with a roadmap, a backlog, and a measurement loop.
The launch is the starting line
At launch you have hypotheses, not answers. The first ninety days of real traffic are the most valuable research the site will ever produce. Teams that stop investing at launch throw that research away.
The productive post-launch rhythm looks like this:
- Instrument first. Every page has one primary action and one event that proves it happened.
- Establish a baseline. Thirty days of clean data before you change anything.
- Fix friction before optimising copy. Technical and structural problems dominate early results.
- Test one hypothesis at a time. Sequential tests with clear decision rules beat a backlog of opinions.
Where conversion actually leaks
After auditing hundreds of pages, the same five leaks show up:
- The first screen does not state the offer. Visitors cannot tell what is being sold or to whom.
- The primary action is ambiguous. Two competing calls to action split intent and reduce both.
- Proof is buried. Logos, numbers, and testimonials live three scrolls below the claim they support.
- Forms ask for too much, too early. Every additional field has a measurable cost.
- Mobile is a scaled-down desktop. Thumb reach, load time, and input types are afterthoughts.
Run this before your next redesign
List the single most valuable action on each page. If it is not visible without scrolling on a 390px-wide viewport, the page has a structural problem, not a copy problem.
Technical performance is a conversion feature
Core Web Vitals are not a developer vanity metric. They map directly to the experience that decides whether a visitor stays:
| Metric | What it feels like | Practical target |
|---|---|---|
| LCP | Time until the main content appears | under 2.5s on 4G |
| INP | How quickly the page responds to a tap | under 200ms |
| CLS | Whether the layout jumps while reading | under 0.1 |
If your site ships a large hero image without explicit dimensions, a render-blocking font, and third-party tags loaded synchronously, you are paying for traffic that leaves before it renders.
1 action
One primary conversion goal per page. Secondary actions should be visually quieter.
30 days
Minimum clean baseline before a meaningful A/B test can be trusted.
90 days
The window where post-launch traffic teaches you more than any pre-launch research.
Build a backlog, not a redesign
The alternative to the every-three-years redesign is a continuous backlog ranked by expected impact against effort. It keeps the site improving in small, measurable increments and prevents the political theatre that surrounds big relaunches.
A healthy backlog has four lanes: technical performance, conversion experiments, content expansion, and structural improvements. Each quarter, ship from all four.
Ownership is the deciding factor
The single strongest predictor of website performance is whether one person owns it. Not a committee, not an agency ticket queue—one owner with a roadmap and a scoreboard. Without that, the site becomes everyone's side project and no one's priority.
