Give every redesign SEO step an owner before design is signed off. The SEO lead lists the old URLs and writes the redirect map, the client decides what each old page becomes, the agency agrees the new structure, and the developer builds and launches it. After launch, the client keeps the redirects for at least a year. A step without an owner is a step that does not happen.
The short answer
The steps are well known. The failure is ownership. Agencies own the design, developers own the build, clients own the business. Search history sits between all three, and on most projects it belongs to nobody until something breaks. Download the checklist below and fill in a name next to every row.
What happened to old agency URLs
We took the 89 design agency websites from our agency study and looked them up in the Internet Archive. For each site we listed the pages the Archive captured between 2021 and 2023 that loaded normally at the time, took up to 15 at random, and requested each one on the live site in October 2026. 82 agencies had archived pages we could test, giving 1,204 old addresses.
| What the old address does today | Share of old URLs with a clear outcome |
|---|---|
| Still works at the same address | 503 (47%) |
| Redirects to a specific new page | 196 (18%) |
| Gone: returns 404 or 410 | 311 (29%) |
| Redirects to the homepage | 28 (3%) |
| Redirects to a catch-all page shared by many old URLs | 38 (4%) |
Looked at agency by agency, among the 73 with at least five old URLs we could classify, 54 (74%) had at least one old page that is now gone, 20 (27%) sent some old pages to the homepage or a catch-all page, and 25 (34%) had half or more of their old URLs broken in one of those ways. Only 14 (19%) kept every old address working or redirected to a specific page.
Agency websites, not their clients' sites, and not only sites that were redesigned: some of these pages were simply deleted. A 404 is the right answer for a page with nothing to replace it, so not every 404 here is a mistake. What the numbers show is how much of a site's history quietly stops working over three years when nobody is responsible for it. 128 of the 1,204 old URLs refused our request or timed out, and are left out of the shares.
Why nobody owns it
Each person on a redesign has a reasonable view of where their job ends. The designer delivers the new pages. The developer builds what was designed. The client approves what they see. The old URLs are not in anyone's brief, because nobody can see them in the new design. On We Digital Hub’s rebuild, the migration plan was its own deliverable for exactly this reason: around 185 project pages already known to Google needed an owner and a new home.
Google's guidance on site moves is clear about what should happen. Map each old URL to its new one, don't redirect many old URLs "to one irrelevant single URL destination, such as the home page", update internal links to the new addresses, and keep the redirects "for as long as possible, generally at least 1 year". None of that is hard. It just needs a name next to it. We describe the full register we use in the redesign handoff.
The checklist, with owners
| Stage | Step | Owner | Done when |
|---|---|---|---|
| Before design sign-off | List every old URL with visits, links or search impressions | SEO lead | One sheet of old URLs, each with its evidence |
| Before design sign-off | Decide what each old page becomes: keep, merge or remove | Client | Every row has a decision and a reason |
| Before design sign-off | Agree the new URL structure, keeping old URLs that do not need to change | Agency | A new URL for every keep or merge row |
| Design and build | Write the redirect map, closest real replacement for each old URL | SEO lead | Every old URL maps to one live page, or is marked to 404 on purpose |
| Design and build | Carry over titles, descriptions and headings for pages that already rank | SEO lead | Any change is deliberate and written down |
| Design and build | Build canonicals, structured data and alt text fields into the templates | Developer | Templates output all three |
| Design and build | Keep staging out of search, with a plan to undo it at launch | Developer | Staging is password protected or noindexed |
| Pre-launch | Crawl staging and test every redirect in the map | SEO lead | One direct redirect per old URL, no chains |
| Pre-launch | Check analytics, Search Console and forms on the new build | Agency | A test enquiry arrives |
| Launch day | Remove staging blocks, publish redirects, submit the sitemap | Developer | No noindex left on the live site |
| Launch day | Test 20 old URLs from outside the network | SEO lead | All 20 land on their mapped page |
| After launch | Watch Search Console for 404s and redirect errors | SEO lead | New 404s mapped or confirmed deliberate |
| After launch | Keep the redirects for at least a year, ideally for good | Client | Rules survive any later hosting change |
Download the checklist as a CSV, with a supporting owner and a pass check for every step. If your agency does not have an SEO lead, that column is the one to fill first, whether with someone in-house or a partner who works under your name.
Three rules that do most of the work
- Don't change a URL you don't have to. The cheapest redirect is the one you never need. A new design rarely requires new addresses.
- Never send old pages to the homepage by default. Map each one to its closest real replacement, or let it return a 404 on purpose. Google's own guidance names the homepage as the wrong destination for many old URLs.
- Make the redirects someone's job after launch. Hosting moves, platform changes and plugin clean-ups delete redirect rules. The client owns the domain, so the client should own the rules, in writing.
Limits
- These are design agencies' own websites, found through search in 12 UK, US and Australian cities. Their clients' sites may be better or worse.
- The Internet Archive does not capture every page, and captures more of some sites than others. We tested a random sample of up to 15 pages per site.
- We cannot tell which old pages were removed on purpose. A 404 counts as "gone" whether or not it was a good decision.
- "Catch-all" means three or more different old pages on one site landing on the same page that is not the homepage, such as a blog index. Some of those may be reasonable.
- Pages that refused our requests are counted as unknown and left out of the shares.
Google: site moves with URL changes
Internet Archive: Wayback CDX server, used to list archived pages
Our own test of archived URLs from 89 design agency websites, 12 and 13 October 2026