A website redesign is the only project where you can do beautiful work and still end up worse off — because redesigns don't usually fail at the design. They fail at the seams: the rankings that didn't survive the URL changes, the converting page that got "improved" into silence, the analytics that quietly stopped recording on launch night. This website redesign checklist is the one we run on our own redesign projects: 20 steps in six phases, with the two sections most guides skip — whether you should redesign at all, and exactly how the SEO survives.
First: do you actually need a redesign?
Most redesign guides assume the decision is made. Don't. Ask four questions first:
- Does the site still convert? If leads and sales are healthy and the complaint is "it feels dated," you may need a refresh — new imagery, tightened copy, updated sections — at a fraction of the cost.
- Is the structure the problem? When you're bolting pages where they don't belong because the architecture has no room, that's a redesign signal — structure is the thing refreshes can't fix.
- Is the technology holding you back? Failing Core Web Vitals you can't fix, plugins stacked on plugins, mobile experiences beyond patching — genuine rebuild territory.
- Have the business or its customers changed? New services, new market, new positioning — the site should tell today's story.
We wrote a fuller diagnostic in six signs it's time to redesign. If you land on "refresh," congratulations — you just saved most of a redesign budget. If it's a real redesign, continue.
Phase 1 — Audit and benchmark (steps 1–4)
1. Inventory everything. Every page, every form, every download — a spreadsheet of what exists. You can't migrate what you never listed, and orphaned pages are how rankings vanish silently.
2. Benchmark your current numbers. Traffic by page, conversion rates, keyword rankings, Core Web Vitals, top landing pages from search. A redesign without a baseline can never prove it worked — and you can't protect rankings you never measured.
3. Identify the bread-and-butter pages. The handful of pages that actually win customers get special treatment: preserved URLs, careful copy changes, extra QA. Breaking what converts is the classic self-inflicted redesign wound.
4. Set goals you can grade. "Look more modern" isn't a goal. "Increase quote requests 30%, cut load time under 2 seconds, rank for X" — those are. Add the honest budget and timeline conversation here (realistic timelines run 3–6 weeks for redesigns).
Phase 2 — Plan the architecture and protect the SEO (steps 5–8)
5. Design the new sitemap deliberately. What merges, what splits, what dies. Every decision here creates a URL consequence you handle in step 7.
6. Keep URLs where you can. The cheapest redirect is the one you never need — if a URL works and ranks, keep it. Change URLs for a reason, not for tidiness.
7. Build the redirect map — the whole redesign hides in this step. One row per old URL: does it resolve identically, or 301 somewhere specific? No wildcards to the homepage, no "we'll catch them later." When we rebuilt this site — over two hundred URLs, 124 location pages, a 63-post blog — every single live URL either kept resolving or 301'd to its successor, and an automated check enforced it before launch was allowed. That discipline is why the traffic survived. Do the same at whatever scale you have.
8. Carry the metadata. Title tags, meta descriptions, canonical logic, structured data — cataloged now, reinstated page by page. Rankings live in these details as much as in content.
Phase 3 — Design (steps 9–11)
9. Start from a design system, not from pages. Colors, type, spacing, and components defined once — so page twenty looks like page one and future additions don't drift.
10. Wireframe, then prototype, then commit. Cheap disagreements happen on wireframes; expensive ones happen after development. Test the prototype on someone who's never seen it.
11. Build accessibility in. Contrast, focus states, alt text, keyboard navigation — designed in now, or retrofitted expensively later.
Phase 4 — Content (steps 12–14)
12. Map every piece of content to its new home. The inventory from step 1 gets a destination column: migrate as-is, rewrite, merge, or retire (with a redirect).
13. Rewrite what underperforms; leave what converts. Copy changes on your best pages should be surgical. Everywhere else, the redesign is your once-in-years chance to fix the message.
14. Rebuild internal links on purpose. New structure means new pathways — point them at the pages that matter, and check nothing links into a void.
Phase 5 — Build and QA (steps 15–18)
15. Build on staging, never live. Password-protected, crawl-blocked, and disposable.
16. Reinstall and verify the measurement. Analytics, conversion events, pixels, Search Console — firing and verified before launch. Losing your data on launch night means flying blind exactly when you need eyes.
17. Test like a customer, on real devices. Every form submitted, every flow walked, on phones as well as laptops, across browsers. Then validate the redirect map — every row, not a sample.
18. Hit the performance bar. Core Web Vitals green before launch, not after — a redesign that ships slower than the old site imported a problem.
Phase 6 — Launch and the 30 days after (steps 19–20)
19. Launch with a rollback plan. Full backup, a DNS cutover plan, and a launch-day checklist: redirects live, SSL valid, forms sending, analytics recording, sitemap resubmitted. Launch when someone's around to fix things — never Friday at 5.
20. Watch the baseline for 30 days. Compare rankings, traffic, and conversions against the step-2 benchmark weekly; watch Search Console for coverage errors and 404s. Catching a missed redirect in week one costs nothing; discovering it in month three costs the ranking.
Four things we'd never do in a redesign
The checklist says what to do; experience says what to refuse:
- Wildcard-redirect everything to the homepage. Google treats it as soft-404s and your rankings evaporate anyway — one row per URL or it doesn't count.
- Launch on a Friday. Problems surface in the first 48 hours; schedule them for when people are at their desks, not at dinner.
- "Improve" the best-converting page by instinct. Your top page earned its position with the copy it has. Change it with data and a rollback plan, or barely at all.
- Skip staging to save a week. Every hour saved building live gets repaid with interest the first time a visitor watches the site change under their cursor.
The checklist, in one place
Audit: 1 inventory · 2 benchmark KPIs · 3 flag converting pages · 4 measurable goals — Plan: 5 new sitemap · 6 keep good URLs · 7 redirect map · 8 metadata carryover — Design: 9 design system · 10 wireframe → prototype · 11 accessibility — Content: 12 migration map · 13 surgical rewrites · 14 internal links — Build: 15 staging · 16 verify analytics · 17 full QA + redirect validation · 18 Core Web Vitals — Launch: 19 rollback + launch checks · 20 30-day monitoring.
If you'd rather have the team that runs this list run yours, book a free strategy call — we'll audit your current site against it and tell you honestly whether you need the full twenty steps or just a refresh.
Frequently asked questions
How long does a website redesign take? Typically 3–6 weeks for a small-business site — about a third faster than building new, because content and brand exist. The timeline stretches when content decisions stall, not when developers work.
How much does a website redesign cost? Usually a bit less than a comparable new build — the ranges in our website design cost guide apply, with savings concentrated in discovery and content. Beware quotes that don't mention redirects: they've priced a rebuild, not a redesign.
Will a redesign hurt my SEO? Only if the migration is careless — which, unfortunately, is common. Steps 2, 7, 8, and 20 exist precisely so it isn't: benchmark, map every URL, carry the metadata, and monitor. Done properly, redesigns routinely improve rankings because speed and structure improve.
How often should you redesign? On triggers, not a calendar: structure that can't hold the business, technology you can't fix, or a message that no longer matches. A well-built site with ongoing care commonly runs five years or more.
What's the single most skipped step? The redirect map, by a wide margin — it's invisible, unglamorous, and the only step whose absence can erase years of SEO in a weekend. If you delegate everything else, keep eyes on step 7 yourself.
Can we reuse our existing content? Usually most of it — that's what the step-12 mapping decides page by page. The converting pages migrate nearly untouched; the rest gets rewritten with intent rather than migrated out of habit.