A redesign is finished when the old site’s promises are accounted for. That is a different state from the new design being live.
The closeout, in order
1. every old url has a disposition: kept, moved, merged or retired
2. every moved url redirects in one hop
3. the sitemap lists the new urls only
4. no staging host is publicly resolvable
5. the search property still points at the same host
6. forms post to an endpoint that still stores responses
7. the analytics tag fires once per page
8. a new baseline reading is taken and dated
A filled closeout
Meridian Landscaping, October 2026, replacing a 2019 site.
old urls: 96
kept 61
moved 22 one hop, all fetched
merged 9 three pairs of thin service pages
retired 4 campaign pages, redirected to the services hub
staging host not resolvable from outside
search property unchanged, domain property
forms one endpoint, tested, response stored
analytics tag fires once, consent banner present
new baseline, 2026-10-06 to 2026-11-02, 28 days
impressions 2,180
clicks 54
click rate 2.48%
avg position 17.9
the year-earlier site, same 28 days, for context only
impressions 2,040
clicks 58
click rate 2.84%
The last block is context, not evidence. A year apart, the site, the market and the query mix all changed.
The month that decides the next quarter
Read pages first and queries second. Expect impressions to move before clicks, and expect a dip while the new URLs settle.
Do not conclude the redesign failed because traffic dipped in week two. Do not conclude it worked because a page looks better.
Acceptance: the redesign is closed when
- every old URL has one disposition, and the count is recorded
- the redirects are fetched, with pass and fail counts written down
- a new baseline exists, with its own dates
- the old numbers are labelled as context, not as a comparison
- the first month’s reading is booked, with a date
- a content owner is named for the pages that lost impressions
Troubleshooting
| Symptom | Likely cause | What to do |
|---|---|---|
| Traffic dips and stays down | Redirects missing for pages with impressions | Compare the map with the impressions list |
| The client says the old site was faster | New images, new fonts | Measure both, then decide together |
| Staging pages appear in search | The preview host is reachable | Block it, then request removal |
| Every service page lost impressions | Titles changed with the template | Reinstate the old title and re-measure |
| Nobody knows what changed | No inventory was frozen | Build one now, from the crawl and the property |
The first month, week by week
week 1 confirm every redirect resolves, and that forms still store
week 2 read pages with impressions, and compare them to the old map
week 3 read the queries behind the pages that lost impressions
week 4 write the first post-redesign report, with the new baseline
The first week is plumbing, the second is content, and only the third is search data. Reading search data in week one produces a decision about a site that has not settled yet.
Limits
Parity is about content and URLs, not the design that replaced them.
If the redesign was also a strategy change, two effects arrive at once. Say that in the report rather than attributing one to the other.