The Website Redesign Playbook: How to Rebuild Without Losing Your Rankings (2026)

Here is the strange truth about website redesigns: the businesses that most need a new website are usually the most afraid to build one. They know the site is dated. They know it is costing them enquiries. But they have heard the horror story, or lived it. A new site launches, everyone celebrates, and three weeks later the phone goes quiet. Traffic falls off a cliff, rankings that took years to earn evaporate, and nobody can say exactly why.

That fear keeps good businesses on bad websites for years. It should not, because the horror story has a boring explanation. Rankings loss after a redesign is almost never mysterious. It is caused by a small, well-understood set of preventable mistakes: URLs changed without redirects, earning content deleted, a noindex tag left on at launch. Every one of them is avoidable if you follow a process.

This is that process. It is the same playbook we ran on our own relaunch, the site that ranks #1 for “web design bristol”, and on every client migration we handle. Follow it and you can rebuild with confidence.

Key takeaways

• A redesign only loses rankings when URLs, content, or crawlability change without a plan. Baseline first, map every URL, preserve what is earning, and monitor for 90 days.
• The single most common catastrophic mistake is launching with a noindex tag still switched on. Check it at launch, then check it again.
• Keep your URLs wherever you can. The strongest redirect is the one you never need.
• Anything that ranks or converts defaults to “keep”. Redesign the wrapper, not the substance.
• Some post-launch volatility is normal for two to four weeks. A sustained decline at week six is not. Know the difference before launch day, not after.
• A redesign is also the cheapest moment you will ever have to build AEO into your site, because the pages are already on the table.

1. Why redesigns kill rankings (and why they don’t have to)

Redesigns kill rankings for one underlying reason: something Google relied on to understand and trust your site changed without a plan. That is the whole mechanism. Google does not penalise you for having a new design. It has no opinion on your colour palette. What it reacts to is broken signals: URLs that now return 404s, content that answered a query and no longer exists, a site it suddenly cannot crawl.

This should be reassuring, because it means the risk is not some unknowable force. Nearly every “the new site killed our SEO” case we have investigated traces back to the same short list of failure modes. Ranked by how often we actually see them:

1. URLs changed with no redirects. The site structure gets tidied up, old URLs die, and every ranking signal attached to them dies too.
2. Earning content deleted or rewritten. Pages that quietly ranked get culled or “freshened up” until they no longer answer the query that earned the ranking.
3. Staging site indexed, or noindex left on at launch. Either Google indexes the half-built site, or the tag that was correctly hiding staging goes live with the real thing.
4. Internal links and headings restructured. Navigation is simplified, contextual links vanish, heading hierarchies change, and pages lose the internal signals that told Google what mattered.
5. A slower, heavier build. The new site looks better but ships twice the JavaScript and half the Core Web Vitals scores.
6. Consolidated pages losing long-tail coverage. Six specific pages become one elegant overview, and forty long-tail queries lose their landing page.

Read the right-hand column again. Every single failure mode is fully preventable. Not “mostly”, not “with luck”. Fully. The rest of this guide is how.

2. The pre-redesign audit: know what you’re sitting on

The first job of a safe redesign happens before anyone opens a design tool: audit the current site so you know exactly what it earns and where. You cannot protect what you have not identified, and the most damaging losses we see are accidental. Somebody deletes a page nobody in the business cared about, without knowing it drove a fifth of the organic enquiries.

The audit is not complicated, but it is non-negotiable. It has four parts.

Crawl the current site

Run a full crawl with Screaming Frog or a similar tool before anything else. You want a complete inventory of every URL on the site: pages, posts, categories, tags, PDFs, images, the lot. For each URL, capture the status code, the page title, the meta description, the H1, and the canonical tag.

This crawl becomes the master list for your redirect map later, so export it and keep it safe. It is also your first diagnostic: if it turns up 300 URLs and you thought you had 40, you have just learned something important about your own site.

Find what’s actually earning

Traffic is not the measure. Business is the measure. Open Google Search Console and pull the Performance report for the last 12 months: which pages get impressions and clicks, and for which queries. Then open GA4 and pull the landing-page report, but do not stop at sessions. Look at which landing pages start journeys that end in enquiries, calls, or sales.

You are building a simple picture: this page earns rankings, this page earns leads, this page earns both. The pages that earn both are your crown jewels, and they get handled with gloves through the entire rebuild. A page with modest traffic that converts at 8% is worth more than a blog post with ten times the sessions and no enquiries.

Backlink audit

Check which of your pages have earned links from other websites, using Ahrefs, GSC’s Links report, or both. This matters because backlinks are attached to URLs, not to your brand. If a trade publication linked to your guide five years ago, that link passes authority to that exact URL. Delete or move the page without a redirect and the equity is gone.

Think of these URLs as load-bearing walls. You can redecorate around them, you can even move them carefully with a 301, but you cannot knock them out and hope the house stands.

Identify the “quiet earners”

Quiet earners are pages that nobody inside the business rates but that quietly rank and bring in traffic. An old FAQ answer. A blog post from 2021 about a niche problem. A location page someone built in an afternoon. These are the #1 category of accidental deletion, precisely because nobody advocates for them in redesign meetings. Marketing wants them gone because they are off-brand; the designer because they are ugly.

Your audit data settles the argument. Sort your GSC pages report by clicks, go past the obvious winners, and flag everything in the long middle that earns real impressions. Every one of those pages needs a decision made about it on purpose, not by omission. We cover the decision framework in section 5.

3. Baseline everything before you touch anything

The rule here is blunt: if you cannot measure the before, you cannot diagnose the after. Before a single URL changes, capture a complete snapshot of the site’s current performance. This is the evidence pack you will thank yourself for in month two, when someone asks “is this dip normal?” and you can answer with data instead of dread.

Without a baseline, every post-launch conversation is anxiety and guesswork. With one, it is a comparison exercise that takes twenty minutes. Here is exactly what to capture, all of it before the build starts:

A few practical notes. Date everything. Store it all in one folder that the whole project team can find. And take item six seriously: search results in 2026 include AI Overviews and AI assistants citing pages directly, and knowing whether you appeared in them before the rebuild is part of the baseline now. Section 5 explains why a redesign is exactly the right moment to act on that.

This whole pack takes an hour or two to assemble. That is the entire insurance premium on years of accumulated rankings. Pay it.

4. URL and redirect mapping: the heart of the whole thing

If you only do one thing from this playbook properly, do this one: account for every single URL on the old site, and make sure each one either survives unchanged or 301-redirects to its closest equivalent on the new site. Redirect mapping is where redesigns are saved or lost, and it is the discipline that separates agencies who say “we’ll look after your SEO” from agencies who can show you the spreadsheet.

Keep URLs where you can

The strongest redirect is the one you never need. Before mapping anything, challenge every proposed URL change. Does /services/web-design/ really need to become /what-we-do/websites/? Usually the honest answer is no. The new structure feels tidier to the people who built the old one, but Google has years of history with the existing URL, and history is exactly what you are trying to keep.

A 301 redirect preserves most link equity, but “most” is doing some work in that sentence, and every redirect adds a hop and a chance for something to go wrong. So the default is: URLs stay put unless there is a genuine reason to move them. Fixing a chaotic legacy structure, removing dates from blog URLs, or consolidating duplicate paths qualifies. Aesthetic preference does not.

Building the redirect map

For every URL that does change, you build a redirect map: a spreadsheet with one row per old URL. The method is simple and it scales from a 30-page brochure site to a 3,000-URL catalogue.

Four columns:

• Old URL. Every URL from your crawl in section 2. Every one. Not just the pages you like.
• New URL. The closest equivalent destination on the new site.
• Type. 301 for permanent moves, which is almost everything. “Gone” (410) for the rare page with no equivalent and no value.
• Priority. Flag the crown jewels: pages with rankings, conversions, or backlinks. These get tested individually on launch day.

Two rules govern the “new URL” column. First, map one-to-one wherever possible: the old service page goes to the new service page. Relevance is what preserves equity. Second, never blanket-redirect everything to the homepage. It is the lazy option and Google treats masses of irrelevant redirects as soft 404s, which means the equity evaporates anyway. A redirect to the homepage is only correct when the homepage genuinely is the closest remaining equivalent.

We have turned our own working spreadsheet into a template you can download above, with the columns, the type dropdown, and a testing tab pre-built. It is the same structure we use on client migrations.

301 vs 302 (and why “temporary” redirects quietly bleed equity)

Use 301s. A 301 tells Google the move is permanent: transfer the history, the links, and the rankings to the new URL. A 302 says the move is temporary, keep the old URL in the index, this will probably revert. Google has said it treats long-standing 302s like 301s eventually, but “eventually” is the problem. In the weeks after launch, exactly when your site is being re-evaluated, a 302 leaves Google hedging while your rankings sit in limbo.

302s creep in by accident: some plugins default to them, some server configurations issue them. So do not just set your redirects, verify them: crawl your old URL list after launch and check the actual status code returned, not the one you intended.

Edge cases: where tidy maps go wrong

The main map is usually the easy part. These are the edges that catch people:

• Parameters. Decide how URLs like /products?colour=green are handled on the new stack; redirects should preserve or gracefully drop parameters rather than 404ing.
• Trailing slashes. If the old site served /about/ and the new one serves /about, that is technically a different URL. Pick one convention, enforce it site-wide, and redirect the other form in a single hop.
• HTTP/HTTPS and www variants. Every combination should resolve to your one canonical form in one redirect, not a chain of three.
• Pagination. Old paginated archives (/blog/page/7/) need a sensible destination, usually the main archive, not a 404.
• Media files with backlinks. PDFs, brochures, and even images earn links. Your backlink audit from section 2 will show which ones. If a linked PDF moves, redirect it like any other URL.

One more rule for the whole section: avoid redirect chains. Old URL to new URL in one hop. If a URL already redirected once in a previous rebuild, update the original rule to point at the final destination. Chains slow crawling, leak equity, and eventually break.

5. Preserving the content that’s earning

The safest content strategy in a redesign is also the simplest: anything that is ranking or converting defaults to “keep”. A redesign is a visual and structural project, not a mandate to rewrite everything, and the moment the project brief drifts from “rebuild the site” to “rethink all the content” is the moment risk multiplies.

That does not mean nothing changes. It means changes are made deliberately, page by page, using a framework instead of taste.

The keep/improve/consolidate/kill framework

Take your full page inventory from section 2 and give every page one of four labels:

• Keep. The page earns rankings, traffic, links, or conversions. It moves to the new site substantially intact. New design, same substance. This is the default for anything with measurable value, including the quiet earners.
• Improve. The page has a job and does it half-well: ranking on page two, converting below par, or visibly dated. Rebuild it on the new site with better content, but preserve what earned its current position (see below).
• Consolidate. Several pages compete for the same intent, or cover one topic in fragments. Merge them into one stronger page, and 301 every retired URL to it. Consolidate cautiously: check the GSC query data first, because six “thin” pages sometimes cover forty long-tail queries that one overview page will not.
• Kill. The page has no rankings, no links, no traffic, and no business purpose. Redirect it to the nearest relevant page, or 410 it if nothing relevant exists.

The framework’s real value is that it forces a decision on every page. Nothing gets deleted by omission, which is how quiet earners die.

Rewriting without wrecking

When you do rewrite an “improve” page, keep the skeleton that earned the ranking. In practice:

• Keep the H1 intent. You can sharpen the wording, but the H1 should still promise the same answer to the same query. If the page ranks for “how much does rewiring cost”, the new H1 had better still be about rewiring costs.
• Keep the headings that match queries. Check GSC for the queries the page earns, and make sure the subheadings that answer them survive the rewrite. Those headings are often the exact reason the page ranks.
• Keep the answer blocks. If a paragraph directly and cleanly answers a question, that paragraph is an asset. It may be why the page gets featured snippets or AI citations. Polish it, do not bury it under brand storytelling.
• Redesign the wrapper, not the substance. New layout, new imagery, new components, better calls to action. The information architecture of the page, the thing search engines actually parsed and rewarded, changes as little as possible.

Building AEO in while the patient is on the table

Here is the opportunity hiding inside all this caution: a redesign is the cheapest AEO project you’ll ever run. Every page on the site is already being rebuilt in a template. The developer is already in the code. The content is already open in a document. The marginal cost of doing the AEO work now, while the patient is on the table, is a fraction of doing it as a separate project next year.

Concretely, that means three things get built into the new templates rather than bolted on later. Liftable answer blocks: a clear, self-contained answer near the top of every page that an AI assistant can quote. Schema markup: Organisation, Service, and FAQ structured data baked into the templates. And entity cleanup: consistent naming and descriptions of who you are and what you do, so machines stop guessing.

We cover the full methodology in the AEO guide linked above. For this playbook, the takeaway is simple: if you are rebuilding anyway, do both jobs at once.

6. The launch checklist

Launch day should be boring. If launch day is exciting, something has gone wrong. The way you make it boring is a written checklist, worked through in order, with a named person ticking each line. This is ours. Print it, or download the PDF version below.

ty)

Thirty items, roughly. It looks like a lot because it is a lot, and that is rather the point. This is what “we’ll look after your SEO during the rebuild” actually means when someone does it properly.

7. The first 90 days: what to monitor and what’s normal

The honest answer to “what happens after launch?” is this: expect some turbulence for two to four weeks, watch it weekly against your baseline, and only act if the data crosses the red-flag lines below. Most post-launch panic is caused by normal volatility being misread as disaster, usually by someone without a baseline. You have a baseline. Use it.

What a normal post-launch wobble looks like

When a redesigned site launches, Google has work to do. It recrawls every URL, follows every redirect, re-evaluates every page against the new templates, and reshuffles accordingly. While that happens, it is completely normal to see impressions fluctuate, individual keywords bounce around by a few positions, indexed page counts move as old URLs drop out, and the odd day that looks alarming in isolation.

The typical shape, assuming the playbook was followed: a dip or wobble in weeks one to three, stabilisation around weeks three to six, and a return to baseline (often above it, if the new site is faster and better structured) by weeks six to twelve. The bigger the structural change, the longer the settle. A like-for-like rebuild with preserved URLs may barely flicker.

The weekly monitoring routine

Fifteen minutes, once a week, same four checks, every week for 90 days:

1. GSC coverage and performance vs baseline. Are indexed pages trending the right way? Are clicks and impressions tracking towards the 12-month baseline for the equivalent period?
2. 404 report. New 404s in GSC mean a missed redirect or a broken internal link. Fix, redirect, and move on. This list should shrink week on week.
3. Rankings snapshot. Your priority keywords against the pre-launch positions. Look at the trend across weeks, not the daily noise.
4. GA4 vs baseline. Organic sessions and, more importantly, organic enquiries against the same period pre-launch, adjusted for seasonality.

This is also where hosting quality shows up. Uptime blips, slow server responses, and botched plugin updates all masquerade as SEO problems in this window, which is one reason we keep monitoring and hosting and maintenance under the same roof for the sites we manage.

Green flags vs red flags

If everything is green, do nothing but keep watching. If anything is red, go straight to section 8.

8. When traffic does drop: the recovery process

If traffic genuinely drops after a redesign, the fix is diagnostic, not creative: work through the failure modes from section 1 in order of likelihood, find the specific breakage, and repair it. Do not redesign the redesign, and do not start rewriting content in a panic. Almost every “the redesign killed our SEO” case we have ever investigated traces back to one of those failure modes, and they are all findable.

Run the triage in this order:

1. Confirm the drop is real. Check against seasonality (compare to the same period last year in your baseline export) and rule out tracking breakage. GSC clicks are your independent witness: if GSC says clicks are steady and GA4 says sessions collapsed, your problem is measurement, not rankings.
2. Check indexation, noindex, and robots first. Look at the live source code and HTTP headers for noindex, check robots.txt, and review GSC’s indexing report. This is the most common catastrophic cause and the fastest to fix. If it is this, fix it, request indexing on key pages, and expect recovery within days to a few weeks.
3. Audit redirects against the map. Crawl the old URL list again. Are the 301s live? Any 302s, chains, or redirects that quietly point to the homepage? Compare actual behaviour against the spreadsheet, not against memory.
4. Compare page-level losses to find the pattern. In GSC, compare pages before and after launch, sorted by click loss, and look for what the losers have in common. Same template (a technical issue or slower page)? Same section (lost internal links)? Consolidated pages (lost long-tail coverage)? Rewritten pages (the answer that earned the ranking was edited out)?
5. Fix, request recrawl, re-baseline. Repair the specific breakage, request indexing on the affected URLs, and take a fresh snapshot so you can measure the recovery the same way you measured the damage.

Recovery timelines depend on the cause. Indexation and redirect fixes tend to recover fast. Content and internal-link losses recover as Google re-evaluates the repaired pages, which takes weeks rather than days. If the pattern will not yield, or the losses are tangled across several causes, that is the point to bring in a specialist: this diagnostic is the bread and butter of our SEO team, and the earlier it starts, the shorter the recovery.

9. We ran this playbook on our own website

The strongest proof we can offer for this playbook is what we staked on it: when we rebuilt our own website, we ran every phase of this exact process on the site that ranks #1 for “web design bristol” and generates our own enquiries. Publishing a process is easy. Betting your own lead engine on it is the endorsement that actually costs something.

The Grizzly relaunch, phase by phase

Our situation going in was exactly the one this guide describes. Ten-plus years of trading and 100+ five-star Google reviews sit behind our site, and the #1 ranking it holds is where our enquiries come from. A botched relaunch would have hit the pipeline of the whole business within weeks. Which made it the honest test. If we had hesitated to run our own playbook on our own site, you would be right not to trust it.

So we ran it without shortcuts, in the order this guide presents it:

• Audit and baseline. Full crawl of the old site, GSC and GA4 exports, backlink audit, dated screenshots of the SERPs that matter to us. The quiet earners surfaced exactly as described in section 2: pages nobody internally rated that turned out to be starting real enquiries.
• URL and redirect map. We challenged every proposed URL change and kept URLs wherever we could. Everything that did move got a one-to-one 301 in the spreadsheet, with the earning pages flagged as priorities for launch-day testing.
• Content pass. Keep/improve/consolidate/kill on every page, with anything ranking or converting defaulting to keep. The wrappers changed completely. The substance that earned the rankings did not.
• AEO while the templates were open. Answer blocks, schema, and entity cleanup built into the new templates rather than bolted on later. The guide you are reading is part of that same system.
• Launch by checklist. The list in section 6 is our actual list, worked through line by line. Yes, we checked noindex twice.
• The 90-day watch. Weekly checks against the baseline, using the same four-step routine from section 7.

The ranking held. The enquiries kept coming. That is why we are comfortable telling you to bet your rankings on this process: we have already bet ours. You can see the finished rebuild on our web design page.

Patterns we’ve seen on rescue jobs

We also get called in after redesigns that went wrong elsewhere, and the same stories repeat. A few, anonymised:

• The homepage blanket redirect. A business came to us after a previous rebuild had redirected every old URL to the homepage. The redirects “worked” in the browser, but search engines treated them as soft 404s and the old pages’ rankings drained away. The fix was the unglamorous one from section 4: a rebuilt map pointing every old URL at its genuine equivalent.
• The noindex that went live. A site launched with the staging noindex still in place, and nobody noticed until pages started falling out of the index. Two lines of code, weeks of damage. It is the reason that check appears twice on our launch checklist.
• The tidy-up that erased the long tail. A redesign merged a set of specific service pages into one elegant overview. The new page looked better and answered less, and the long-tail queries those pages owned quietly went elsewhere. Reinstating the specific pages, with the redirects repaired, brought the coverage back.

None of these businesses did anything reckless. They rebuilt without a playbook. That is the entire difference.

FAQ

Will a new website hurt my SEO?

Not if it is done properly. Rankings loss after a redesign is caused by preventable mistakes, mainly changed URLs without redirects, deleted content, and indexing errors, not by the redesign itself. If you baseline your current performance, map every URL, preserve the content that earns rankings, and follow a launch checklist, a new website should hold your rankings and often improve them.

How long does it take rankings to recover after a redesign?

Expect two to four weeks of normal volatility while Google recrawls and re-evaluates the site, with full stabilisation typically inside six to twelve weeks. Like-for-like rebuilds with preserved URLs often barely wobble. If clicks are still down significantly at week six, that is not normal settling, it is a sign of a specific problem worth diagnosing immediately.

Should I keep my old URLs?

Yes, wherever you reasonably can. Your existing URLs carry years of ranking history and link equity, and the strongest redirect is the one you never need. Only change URLs when there is a genuine structural reason, and when you do, 301-redirect each old URL to its closest one-to-one equivalent on the new site.

Do I need redirects if the design changes but the URLs stay the same?

No. Redirects exist to connect old URLs to new ones, so if every URL stays identical, there is nothing to redirect. You still need the rest of the playbook, though: preserved content and headings, carried-over metadata, the noindex check, performance testing, and post-launch monitoring all still apply to a same-URL redesign.

Can I redesign and change domain at the same time?

You can, but do not if you can help it. Each project is a manageable, measurable risk on its own. Combined, they change every signal at once, and if traffic drops you cannot tell which change caused it. If both are unavoidable, do them in stages: migrate the domain like-for-like first, let it settle, then redesign, or vice versa.

Why did my traffic drop after launching a new site?

Almost always one of six causes: missing redirects, deleted or rewritten earning content, a noindex tag left on, restructured internal links, a slower build, or consolidated pages losing long-tail coverage. Check indexation and noindex first, then audit your redirects, then compare page-level losses in Search Console to find the pattern. The cause is nearly always findable.

How many redirects is too many?

There is no meaningful limit on the number of individual redirects; large sites run thousands without issue. What hurts is redirect chains, where one URL bounces through several hops before landing. Keep every redirect to a single hop from old URL to final destination, and update any legacy redirects so they point directly at the new page.

Should the old site stay live during the build?

Yes. Keep the old site live and untouched while the new one is built on a staging environment that is noindexed and ideally password-protected. That way you keep earning traffic and enquiries throughout the build, Google never sees a half-finished site, and launch becomes a controlled switchover rather than a gap in service.

Is a redesign a good time to start AEO?

It is the best time. The templates are open, the content is being handled anyway, so answer blocks, schema markup, and entity cleanup can be built in at a fraction of the cost of retrofitting them later. If AI assistants and AI Overviews matter to your market, do both jobs at once. Our Answer Engine Optimisation guide covers the full method.

Final thoughts: rebuild with confidence

The fear of losing rankings has kept more businesses on outdated websites than any budget ever has. But the fear is out of proportion to the risk, because the risk is a short list of known, fully preventable mistakes, and you now have the complete process for preventing every one of them: audit, baseline, map, preserve, launch by checklist, and watch for 90 days.

If you follow this playbook yourself, it works. If it looks like a lot of work, that is because it is, and it is exactly the work we do on every site we rebuild, our own included.

Planning a redesign? We will run the section 2 audit on your current site free of charge: what is earning, what is load-bearing, and what your rebuild needs to protect. Get your free pre-redesign audit and start the project knowing exactly what you are sitting on. We are Grizzly, a Bristol web design and SEO agency, ten-plus years in, with 100+ five-star Google reviews and a website that practises what this guide preaches.