Most website projects are sold like a finish line. Six months of design, build, content, launch — congratulations, here's the new site. Then the team disappears, the agency sends an invoice, and the client realizes they don't quite know what to do next. The truth is the launch is a milestone, not a finish. The 90 days after a site goes live are where it earns or wastes the entire investment.
Week one: catch the things QA missed
No matter how careful pre-launch testing is, real traffic finds bugs that staging never will. The first week is about catching them fast.
- Form submissions. Every form, every confirmation email, every CRM entry. Test from real devices, not just dashboards.
- Analytics. Confirm events fire correctly and the data is landing where reports expect it.
- Search Console. Watch for crawl errors, redirect chains, and indexation gaps from the migration.
- Speed and Core Web Vitals. Real-user metrics often differ from lab tests. Watch the first week of field data carefully.
- Edge cases. Mobile Safari, low-end Android, slow networks. Real audiences have all three.
The first week's job is to fix anything that's actively losing the business money — broken checkouts, broken contact forms, broken redirects from old URLs. Save the polish for week two.
Weeks two to four: the migration tail
If the site replaced an existing one, the migration tail is the most important thing happening that no one's watching. Search rankings shift, redirects either hold or don't, and old backlinks land where they should — or 404. We watch:
- 301 redirect coverage — every old URL with traffic mapped to a new destination, no chains, no loops.
- Indexation — pages indexed vs. pages submitted, and the gap closing on the right pages.
- Ranking shifts — top organic terms tracked daily for 30 days post-launch.
- Backlink health — the most-linked old pages confirmed redirecting cleanly.
Most ranking dips after a launch are recoverable in two to four weeks if the migration was clean. The ones that don't recover usually trace back to a redirect mistake nobody caught.
The 90-day roadmap that should have shipped with the site
Every site we launch ships with a 90-day plan attached. Not a wishlist — a prioritized list of the next changes the data will tell us to make. It typically covers:
- Conversion-focused iteration. The pages that drive 80% of the value get rewritten and re-tested first. Everything else waits.
- Content gaps. Pages we knew were missing at launch but didn't block the ship date.
- Performance fixes. Anything Core Web Vitals exposed in real-user data.
- Tracking refinements. Events that need cleaning up, attribution rules to tighten, dashboards to consolidate.
- SEO depth. Internal linking, schema markup, and content depth on the pages closest to commercial intent.
Where AI fits in the post-launch loop
We've started baking small AI helpers into the post-launch phase because they collapse the time between what's happening on the site and what should we do about it:
- Session-recording summarization. A model that watches the highest-friction sessions and surfaces patterns the team would otherwise miss.
- Search-Console anomaly detection. Flag rankings that move outside normal variance before a human notices.
- Form-failure clustering. A small agent reads form-error logs and groups them by likely cause.
- Content-gap surfacing. Comparing the queries you rank for vs. the queries you should — automated and updated weekly.
None of this replaces judgment. It just makes sure the team is spending judgment on the right problems instead of digging through dashboards.
What ongoing actually means
Ongoing support is one of the vaguest line items in any agency contract. Pin it down before you sign. A real ongoing engagement covers, at minimum:
- Monthly performance review against the metrics we care about — not a screenshot of GA.
- A backlog of prioritized improvements with clear owners.
- Security and platform updates handled without drama.
- A response SLA when something breaks.
- An honest, recurring check-in on the what's working / what's not of the program.
A site that gets a small, smart edit every week beats a site that gets a heroic redesign every three years.
What success looks like at 90 days
If the launch went well and the post-launch work was real, by day 90 you should have:
- All migration losses recovered or explained.
- Conversion on top pages flat-or-up vs. the old site, with a clear test backlog underway.
- Real-user performance inside Google's good thresholds.
- A monthly report you'd actually defend to leadership.
- A roadmap for the next 90 days that's based on what the data has said since launch — not on what was guessed before it.
Anything less and the launch was the project. The work that follows is what makes a site actually pay back.