You put a new site live with a team, the homepage is a stunner, and then three weeks in you see organic traffic has fallen 30%. Don’t blame the redesign for that. The fault lies with calls made months prior to launch.
Too often SEO and a website overhaul are put in different silos even though they are wired to the same things. A redesign will alter your rendering behavior, on-page copy, navigation, templates and URLs in one go. Search engines pick up on all of it as signal, as do the AI systems churning out answers above the blue links these days. If you handle a redesign like a migration with some brand work on top, you can make headway. Handle it as a brand exercise and check off SEO at the end and you will likely cede ground.
We have put together this guide for the person in the company who has to make the call on what to protect, rebuild or put to rest before the design files are in their final form.
## Why Redesigns Quietly Damage SEO
It is rarely one large error that sends traffic down. More often it is a host of small ones happening in concert. You might have a redirect that overlooks a long-tail URL, or a new template that relegates H2s to H4s. Maybe a JS menu is hiding links from the crawler, or you have consolidated a page with 200 backlinks into a catch-all “solutions” hub. The accordion is neater, but now Google has to click through to find the words it was ranking you for.
The numbers don’t lie. According to our 2026 SEO statistics roundup, organic search is responsible for some 53% of website traffic, yet 96.55% of pages get nothing from Google. The first organic spot will take about 27.6% of the clicks. Your working pages are the exception. A redesign either preserves them or erases their context.
And the playing field is more constricted than it used to be. Core Web Vitals are a given as ranking factors and INP took over from FID in 2024. Mobile-first is the only way of indexing. Ahrefs looked at 146 million results and found AI Overviews on 20.5% of queries; Pew Research says when an AI summary is there, your CTR can fall from 15% to 8%. In fact, if you strip out brand language for the sake of clean copy, you will pay for it – Ahrefs’ data on 75,000 brands shows a 0.664 correlation between brand mentions and AI Overview visibility, compared to 0.218 for backlinks.
That doesn’t mean a redesign is inherently risky, just that it is a migration and should be planned as such.
## Phase 1: Audit Before You Touch the Design
A moodboard is not the first thing you need. What you need is a spreadsheet of what is already working.
Run a full crawl of the existing site and cross reference it with your analytics (for entry pages and conversions), backlink data and Google Search Console – look at impressions and queries, not clicks. Server logs are good to have if you can get them. Do not do a path-only crawl or you will miss the long tail where unmanaged losses tend to hide.
### What to collect before any design work
* **All live URLs** and not merely those in the main nav. * **Your top landing pages** by way of both impressions and clicks. Impressions will show you equity in pages that rank but don’t get many clicks. * **A metadata baseline** covering titles, canonicals, schema, hreflang, descriptions and heading hierarchy. * **Internal linking:** see which pages are propping up others via anchor text. * **Backlinked pages**, regardless of current traffic. * **Technical problems** like slow templates, duplicate URLs or render-blocked content.
It is tedious work but it is what will determine the outcome of the redesign. You want to come away with a decision for each URL – protect, improve, merge or retire – not a report. Any page with backlinks or impressions is protected. If two are vying for the same intent, you merge them with a 301. Those with no value get a redirect to the parent.
Prune with care. Some low-traffic pages are the reason your money pages rank well. To delete on the basis of last-click traffic is a surefire way to see a quiet decline. For a structured approach, an SEO audit and optimization is a sound place to start.
## Phase 2: Map URLs and Content Before Design Locks
This is where you decide if you are keeping your rankings.
Think of a 301 as a permanent change of address. Google will let most of the original equity pass through. But if you have soft 404s, loops or chains, you are going to leak it or lose it altogether. You don’t have to be an old hand in the business to know what Brightspot’s practical advice on protecting SEO in a redesign is telling you. The redirect map has to be one-to-one as a rule of thumb, reserving many-to-one for when you are truly consolidating.
### The shortcut that will cost you traffic
There is no worse habit than the bulk redirect of old URLs to the homepage. It is a tidy solution on paper, but it signals to Google that the original page is gone and the homepage has nothing to do with the user’s query. Your equity and your rankings will drop accordingly.
A good map is literal:
| Old URL | New URL | Decision |
|---|---|---|
| Old service page | Matching service page | 301 one-to-one |
| Old blog post | Updated equivalent article | 301 one-to-one |
| Two overlapping articles | One consolidated guide | Merge, 301 both |
| Outdated thin page | Closest topical parent | Retire with 301 |
| Page with backlinks but low traffic | Equivalent or enhanced page | Preserve URL where possible |
### Where things get more difficult
The trouble is, most of the guidance out there presumes you are merely trading one visual layer for another. In reality, projects are seldom so simple. You could be making the jump from a legacy CMS to WordPress, or folding microsites into a single domain, or putting to rest a homegrown system after ten years of content.
Take our work with Teton Gravity Research. The redesign was the easy part. We had 10,000 articles in a legacy CMS and some site sections that were no longer of any use to the business. We didn’t set out to preserve every URL, only those with link equity and search intent, and we let the rest go. We saw the same thing at the Keck School of Medicine with subdomains that had put down roots organically over the years, housing everything from faculty pages to newsroom pieces.
It puts a different spin on the question: not how do we keep it all, but what is worthy of its own URL? If you are cleaning up content as you design, preparing for a CMS migration is something you do before the build starts, not in the middle of it.
> Keep a page if it serves intent, not simply because it was in the old sitemap.
## Phase 3: Technical SEO Checks During Build
While the new site is being built on staging, you must block it from search engines with authentication and noindex; a robots.txt file won’t cut it. And before anything goes live the team should confirm the new build has the same signals as the old.
**What we verify in staging:**
* **Indexing controls:** Production is ready to open, staging is locked down. * **Rendered HTML diff:** On top pages we compare the HTML and structured data from before and after, not just how it looks. That is how you spot a heading downgrade or schema that has been dropped. * **Internal links:** Menus and footers should point straight to the new URL, not via a redirect. * **Navigation in initial HTML:** Crawlers need to follow real anchor tags without having to render JavaScript. * **Mobile rendering:** Make sure key text and links are there on a small screen and not hidden behind an interaction. * **Performance budgets:** LCP, INP and CLS are measured against the targets you set prior to the design phase. * **Canonicals and hreflang:** These can be incorrectly regenerated by a change in CMS theme, so they have to be checked.
And a word on JavaScript frameworks: while Google can render JS, you can’t always count on them to do it in a timely or reliable fashion. When you are redesigning and making the switch to a Next.js or Nuxt stack (or something of that ilk), you need to have server-side rendering or static generation in place for any page with ranking ambitions. Leave it to client-side rendering for your main content and you will see those pages in Search Console as “Crawled – currently not indexed” on a regular basis; putting right what is wrong is often a matter of rebuilding the page altogether.
Don’t make the mistake of using staging as a private design review and thinking you can put an SEO patch on after launch. Once you go live, Google is only going to see the rendered HTML.
Phase 4: A Launch That Feels Boring
You want a good launch to be uneventful. Any drama on the day of is a sign your runbook was wanting. A runbook is simply a written checklist with clear order and ownership. There is no room for heroism here, just timestamps and names. One person deploys, another has the job of verifying redirects to the map, someone else is on tracking and indexing controls, and so on.
Here is a sequence that works:
1. Put the new site into production behind a maintenance flag if you can. 2. Turn on the redirect map and put some old URLs to the test against it, long-tail and backlinked ones included. 3. Strip out the noindex and authentication from the live build that were there for staging. 4. In Google Search Console, put in the new XML sitemap and ask for the high-value pages to be indexed. 5. Do a full crawl of the live site to turn up any broken links, orphan pages, redirect chains or canonicals you didn’t expect. 6. Make sure analytics and conversion tracking are firing on the new templates, not just the home page.
We have seen an agency case study where a redesign yielded an 85% bump in organic traffic over three months, with total ranking keywords nearly doubling from 2,841 to 5,964 and page-one spots going from 424 to 578. Those are not numbers you should hold as a benchmark, but they do illustrate what you can achieve when you plan your templates, content and redirects in concert instead of trying to fix them later.
If the redesign is accompanied by a platform move, get a website migration launch day checklist in writing before anyone nears production. The Refact website migration service page goes into the wider process.
> Launches go right when each check has an owner and a timestamp. They go wrong when teams put their faith in memory.
Phase 5: The First 90 Days After Launch
Going live is the start of the next phase, not the end of the project.
In the first eight weeks or so you can expect to see keyword volatility of two or five positions; it is normal. If the migration was done properly, you will see things stabilize around week eight and a full recovery in ninety days. A botched one can drag on for six to twelve months with some pages never coming back. The first fortnight will tell you which road you are on.
What to watch closely
Here is what you should be looking at:
- Search Console index coverage: keep an eye on any spikes in soft 404s or the “Crawled” and “Discovered” but not indexed categories. They are often the first sign of a thin content or rendering issue.
- Impressions for top landing pages: do not just look at the aggregate. Compare your old top fifty with their mapped counterparts on a week to week basis.
- Redirect logs: check for 404s on URLs that should have been mapped, and for any chains put in place by CMS or CDN rules you may have overlooked.
- Core Web Vitals from the field: what you see in the lab during staging will not always tell you what the actual user is getting.
- Conversion paths: things like member sign-up, form submissions and checkout. If you only review traffic totals you can miss tracking errors here for weeks.
When something does break, don’t go about rewriting the whole thing. Be diagnostic first. See if the page is indexed, if the redirect is direct, if internal links are still pointing to it and if the HTML renders as it was meant to. You will find most short term declines are due to some small structural fault rather than the content itself.
That is the case for post-launch ownership too. Refreshing is part of SEO performance now, not a side program. Ahrefs looked at 17 million AI citations and put the number at 25.7% fresher than your average organic piece. Pages you update quarterly with new examples and stats will hold their own better than ones left to stagnate after launch.
The One Document That Decides Most of the Outcome
A redesign done right begins with a spreadsheet. Nothing more exciting than a list of every URL, status and owner.
It is the structural decisions that will damage rankings, not the visual layer. Make your calls on the URL inventory and content before the designers fire up Figma and the project is much simpler. Do it once development has started and you are simply buying back ground you ceded.
For a way to frame those calls ahead of time, our website redesign services guide shows how scope, audit and migration come together. The Refact discovery process is made for this point in the game when a misstep on templates or URLs is still inexpensive to put right. We even have a money-back guarantee on the strategy phase should it fail to provide the clarity you need.
