A redesign is always a structuring moment in the life of a site. The CTO wants to secure the go-live, the PO wants to hold the milestones, the SEO Manager wants to avoid a drop in traffic – and in the bottom, everyone defends a legitimate stake. With regard to SEO, it is often referred to as a mysterious block that "mounts" or "down". In the field, the SEO is not lost because we change a design: it gets lost when governance is blurred, no one has a hand on it after posting, and the content strategy is treated as a detail.
Above all, the redirection plan is not a deliverable that is deposited at the end: it is a brick of continuity that is built early, implemented cleanly and tested as a functionality. Avoidable errors (provided you don't control the feeling!) What we detail here: from the initial audit to the J+30 follow-up, the sequence that is applied to secure a redesign without sacrificing the SEO. The differentiator is not to be cautious: it is to measure and correct quickly.
What really breaks the SEO during a redesign
In most post-recast falls, the problem is not technology. It's a pilot shift: decisions exist, but not at the right time, or not at the right place.
Cause #1: blurred governance
When no one « owns » redirections, canonics, sitemap, analytic rocking or validation of robots.txt, we end up with blind angles. These blind spots are rarely visible from a functional point of view — but very visible from the point of view of crawl and indexing.
Cause 2 : post-production tracking absent.
Without J+1/J+7/J+30 routine, 404, soft-404, excluded pages, cropl anomalies are discovered late. And the later the patch arrives, the longer the engines take to re-evaluate and re-index the site correctly.
Cause #3: improvised content strategy.
Repeat often rhymes with temptation to « All rewrite », « merge all », « simplify everything ». Without content mapping, you can easily create cannibalization, remove pages that wore backlinks, or change target intent without deciding.
These three causes often combine. And when they combine, the most common symptom is an incomplete redirection plan.
The "perfect crime": forgotten redirections (real case of recovery)
Classical scenario : the site is put online, works, pages load... and organic traffic dives. A few days later, 404 avalanche in Search Console. Positions back on core business requests.
And the cause is often very simple: the redirection plan was forgotten, delivered but not implemented, or implemented without serious testing.
This is a typical incident: production is considered to be validated, while organic traffic continues to arrive on old undirected URLs.
Symptoms → diagnosis → patches
On one of the agency's projects, the signals were clear: sharp rise of 404, decrease of relevant indexed pages, and landing historical pages (top traffic + backlinks) replaced by close pages... but not equivalent.
From there, the diagnosis follows a simple order:
- first look at logs (Googlebot on old URLs + answers 404)
- top pages (traffic, conversions, backlinks) are extracted to check equivalence
- then reconstruct a prioritized URL mapping (money pages first) e
- We deploy 301 own in 1→1.
So we put on a clear redirection plan, a preprod recipe, and a tooled post-production monitoring. With this, we get back on track and get back on our hands.
You will understand, so the best practice to remember is to prepare all of this BEFORE the redesign.
Before the redesign: laying the foundations
Governance & RACI
A loss-free recast SEO is a recast where the « who does what when » is written black on white. Five perimeters to cover – SEO, redirections, tracking, dev/performance, release – each with a designated official.
The staking follows a sequence logic: The next step will not be taken without having validated the previous step.
A realistic staking looks like:
- & basic audit
- content + mapping strategy
- redirection plan v1
- Staging ready
- SEO preprod recipe
- go/no-go
- online
- monitoring J+1/J+7/J+30.
Finally, three pilot rituals are enough to avoid blind spots.
- A weekly ritual (30–45 min) dedicated to risks, decisions and dependencies;
- a SEO x Dev (30 min) ritual centered on anomalies and recipe fixes;
- and two pre-go-live points (J-3 then J-1) for the final check and freezing of changes.
SEO Audit & Content Strategy
What could we break?
Before changing anything, the goal is to have a clear overview.
The top landing organic pages (sessions, conversions, income/leads), page money (products, categories, forms, pricing, local), and pages with backlinks (sources, anchors, "authority" pages).
Then we segment by templates (blog, categories, sheets, institutional pages, local pages, internal engine, facets...). SEO often degrades "by template", so this segmentation allows to test intelligently in preprod.
Finally, a KPI reference point is set before redesign (by directory and template): organic traffic, organic conversions, GSC prints/clics, indexed pages, crawl errors, response time, and Core Web Vitals if available.
Content strategy (mapping, pluming, cannibalization)
Then the goal is not to have more content, but the right content, in the right place, with a clear intention.
Intentions are mapped to target pages: For each important intention, it is decided which page should win (and which should no longer fight for the same query).
Then the arbitrations come:
- keep (traffic/backlinks/conversions + own intent)
- merge (identical intentions that dilute signals)
- delete (pruning) if zero value – with redirection if URL still receives crawl or links – and rewrite if the page is strategic but off-game.
Finally, we secure the internal mesh : the redesign changes the navigation and can create orphan pages. Thematic hubs, contextual links, consistent breadcrumbs, and a control crawl avoid "losing" pages without reporting them.
Once the strategy has been framed, one can attack the most sensitive brick: redirections.
What it involves in marketing: Content decisions – which pages keep, merge, delete – have a direct SEO impact. These are not just editorial decisions. They deserve to be treated as such in the project.
The redirection plan: the life insurance of the project
Re-iriger « all to home » is often a shortcut... to a fall.
Golden Rule: 1→1
An old URL must point to a new URL as equivalent as possible – same intention, same content. Home can be an equivalent in rare cases, but it is rather the exception than the rule.
The plan is built early, is prioritized by business issue (money pages, top traffic, backlink pages first), and is tested in preprod as a feature – not to feel or hand over ten URLs.
Expected spreadsheet format
| Old URL | New URL | Type | Reason | Priority | Source | Status |
|---|---|---|---|---|---|---|
| /blog/guide-migration-seo | /resources/migration-seo-checklist | 301 | consolidated article | P1 | traffic + backlinks | tested in preprod |
| /solutions/wordpress-vip | / expertise/wordpress-vip | 301 | arbo recast | P1 | Conversions | to be implemented |
| /products/old offer | /offers/new offer | 301 | rename offer | P2 | traffic | Implemented |
SEO recipe in preprod: what is valid before switching
A redirection plan is tested in preprod – without being indexable. We prefer a protected staging (auth), coherent noindex guidelines (meta robots) and a robot.txt that avoids accidental indexing. We also avoid public links to preprod: it happens more often than we think.
Validate template by template : status codes, canonicals (not broken, not to staging), sitemap clean, robots.txt, performance, schema.org.
We test the plan of mass redirections – loops, chains, equivalence of intent.
And we make sure that observability is in place Googlebot logs, monitoring 404/5xx, Search Console ready.
Once these safeguards are laid, D-Day becomes much simpler.
D-Day: an online posting without shaking
GO if:
- robots.txt and meta robots are correct (prod indexable / static blocked)
- sitemap prod online and contains only indexable URLs
- P1 redirects in place and tested (top traffic + backlinks + money pages)
- analytics + tag manager published and validated key events
- no 5xx and acceptable performance on critical pages.
NO-GO if:
- indexable preprod (even "a little")
- 301 missing on pages at stake
- canonical incoherent or to staging
- Unvalidated tracking (blind driving)
- server incidents not included.
Once online: publish the sitemap, check robots, launch a few URL inspections on strategic pages, validate tracking.
Then you switch to the real differentiator: after putting it online.
After placing online: 30 days to secure winnings
This is where the difference between a successful recast and a recast is often played out for six months.
J+1 — Emergency & Hygiene
KPIs: 404/soft-404, 5xx, crawled pages, robot/canonical anomalies, tracked conversions.
Actions: correct missing redirects P1, repair 404 internal (menus, links), patch robots/canonics.
J+7 — Indexation & stabilization
KPIs: GSC prints/clics, pages « Excluded », error of exploration, depth of crawl, response time, CWV key pages.
Actions: adjust sitemaps, correct unwarranted excluded pages, process redirect chains, strengthen mesh to strategic pages.
J+30 — Business reading & optimization
KPIs: organic traffic by directory/template, positions on heart requests, organic conversions/revenues, performance of money pages, crawl budget (logs).
Actions: iterate on titles/H1, consolidate cannibalization, optimize templates that pull performance down.
The post-recast SEO is piloted as a product. The rest is mainly to understand Why These KPIs are moving.
Errors that are expensive (and that you see coming back)
- Redirections forgotten on the long trail — or tested « by hand » on ten URLs
- Incoherent Canonicals after navigating redesign
- Orphan pages after changing menu
- Sitemaps too wide including non-indexable URLs
- Pagination and filters that create a crawl trap
- Preprod left indexable « just during the tests »
- Change everything at the same time (URL + content + arbo + tracking + perf) without safeguards
In summary: the sequence that protects a redesign
On the recasts that we accompany, success often comes from a simple sequence:
Clear governance → structured content → solid redirects → recipe in preprod → monitoring post-prod.
Three practical positions to keep in mind:
- The redirection plan is a brick of continuity, not a deliverable of end of project
- The SEO recipe is made in preprod, not after posting online
- Post-prod is part of the project – J+1/J+7/J+30 are not optional
If you don't, we're flying in the air.
If you are preparing a redesign, it can take 30 minutes to put the SEO risks flat, look at your context and estimate the points of vigilance together. The objective is to move more calmly, with a clear framework.
