Rehaus.
From Messy
To Momentum.
I took over a failed .co.uk to .co migration three weeks after the switch. Rankings in freefall, 82% of the site's redirects pointing at a dead domain, and 42,593 historical clicks draining away. Four months later: average position halved, monthly impressions up 5.9x, and ChatGPT ranks REHAUS the #1 place to buy second-hand designer furniture. Here's the full rebuild.
"Appreciate your getting stuck into this so quickly, the depth of the diagnostic is genuinely useful. The migration rescue work is real: the 530 redirects, brand consolidation, 269 invisible designers, schema fixes, CLS, /pages/sell. All of that needs to ship and ship fast."
"The methodology architecture (Reviews + Alternatives + Versus + Round-ups feeding brand pages) is exactly the right shape. Genuinely useful work."
THE
BRIEF
REHAUS sells authenticated, restored designer furniture: Ligne Roset Togos, Roche Bobois Mah Jongs, B&B Italia, Cassina, Vitra, Knoll. Every piece collected, authenticated, and stored before sale. A genuinely strong proposition. However, when I arrived in early April 2026, none of that mattered to Google. The site had just been migrated from rehaus.co.uk to rehaus.co, and the migration had been executed without any SEO management at all. Traffic wasn't dipping. It was collapsing.
- 01 A migration with the engine missing The GoDaddy DNS had never been updated, so redirects from .co.uk simply didn't fire. Worse, the old domain was connected to Shopify as a "redirect domain", which dumps every historical URL onto the homepage before path-level redirects even get a chance. Sixteen months of history, 2.9M impressions and 42,593 clicks across the old site's top pages, all draining into a dead end.
- 02 A redirect table actively working against the site 5,270 live redirect rules were in place. My audit found 82% of them pointed back at the dead .co.uk domain, nearly all resolving to the homepage. Google reads irrelevant homepage redirects as soft 404s, so the "fix" someone had installed was compounding the damage. Separately, 530 product URLs had been switched from numeric IDs to descriptive slugs before the migration, with no redirects at all.
- 03 Templates Google couldn't read 790 of 791 indexable pages had at least one metadata issue. 246 collection pages had a meta description of literally "@ REHAUS". Eight characters. 335 of 336 collection pages had no H1. 269 designer links pointed at noindexed vendor filters, an entire tier of invisible pages. The most commercially valuable templates on the site were giving Google nothing to rank.
- 04 Zero presence in AI search Buyers were already asking ChatGPT where to find pre-owned Ligne Roset before they Googled it, and the models were misrepresenting REHAUS as a generic marketplace or discount site. No llms.txt, no canonical information source, duplicate Product schema fighting itself, and an SEO app quietly overwriting metadata across the site. The AI layer wasn't underperforming. It didn't exist.
THE
STRATEGY
You can't build growth on a site that's actively leaking. My line to the client in week one: pushing content before fixing the migration is like building a sail for a ship with massive holes in the hull. So the plan ran three workstreams in deliberate sequence, with content overlapping the rescue rather than waiting politely behind it.
- DNS fixed at GoDaddy: the reason .co.uk redirects weren't firing at all
- Domain type switched from Redirect to Alias in Shopify, so path-level 301s could finally chain through instead of dumping on the homepage
- 530-URL product redirect map built by matching old numeric URLs to new slugs via Shopify product IDs
- Full audit of all 5,270 live redirects, with rollback exports saved before any change
- Google Change of Address filed once the Search Console ownership saga was untangled
- 800 priority titles and metas rewritten by hand after killing the app auto-overwrite that was flattening them
- New collection template coded: entity-rich copy, accordion FAQs and FAQPage schema under the product grid
- 113 duplicate brand_ collections consolidated so rankings stopped cannibalising
- Schema brought in-house: duplicate Product JSON-LD removed, TinySEO phased out safely, AVADA ghost code stripped
- E-E-A-T layer built: founder author page, author bios on every article
- The methodology map: Reviews × Alternatives × Versus × Round-ups, all feeding brand and collection pages
- Honest Review template built rigid on purpose, so AI assistants can lift and cite its answers
- 2 SEO + 2 AEO articles a week, 22 live and earning impressions by August
- llms.txt and a canonical AI information page telling every assistant exactly what REHAUS is
- Collection copy rollout targeting terms already in striking distance: togo sofa, tylko, roche bobois, boconcept
5,270 redirects audited row by row. 2,921 rebuilt to precise destinations, 2,349 junk rules removed, 530 numeric product URLs mapped, 1,600 more rebuilt from a legacy export. Roughly 7,000 spreadsheet checks.
A recoded Shopify collection template adding copy, accordion FAQs and schema below the grid, without touching products, tracking or layout. 29 pages relaunched on it so far.
22 live articles: Honest Reviews of Made.com, Loaf, Soho Home, Sofology, DUSK, Swoon and more, plus definitive guides to the Togo, Mah Jong and USM Haller. Built for SERPs and AI citation simultaneously.
llms.txt, a canonical "Official Information About REHAUS" page structured for LLM retrieval, and clean single-source schema. The layer that turned ChatGPT from misrepresenting REHAUS to recommending it first.
800 priority pages given hand-written titles and descriptions, replacing eight-character metas and "Sofas @ REHAUS" titles. H1s restored to collection pages that had none. Four template patterns, one per collection type.
Every live backlink to the old domain pulled from Ahrefs and triaged: 13 editorial links worth reclaiming (House & Garden, MO Magazine, Country & Town House), 8 self-owned profiles fixed, 83 spam domains disavowed.
THE ARTICLE ENGINE:
GOOGLE + CHATGPT
The content brief wrote itself once I looked at where REHAUS buyers actually start: searching the brands they already know. "Is Made.com any good?" "Loaf alternatives." "Soho Home review." Every one of those searches is a £2,000 to £5,000 buyer at the exact moment they're open to a smarter answer.
So I built the plan around four content types that feed each other: Honest Reviews, Alternatives, Versus pieces and Best-of round-ups, all linking down into REHAUS brand and collection pages. The reviews follow a deliberately rigid template. Declarative sections, structured verdicts, Review schema, questions phrased the way buyers ask them. That rigidity isn't a style choice. It's what makes an article liftable by an AI assistant, because LLMs cite content in small declarative chunks, not flowing essays.
The result? Within three months, two of the site's top ten pages were articles. Within four, the review cluster was outranking the brands' own coverage. The Made.com review sits at position 3.5 with 50,866 impressions. And the same content is now cited by ChatGPT when buyers ask it where to shop.
| Article | Clicks | Impressions | Avg Pos | Signal |
|---|---|---|---|---|
| Honest Sofology Review | 332 | 32,320 | 6.6 | Top-performing article on the site. A four-month-old domain outranking national press for a major sofa retailer's review terms. |
| Honest Loaf Sofa Review | 290 | 14,388 | 6.9 | 2.02% CTR, the strongest of the review cluster. Loaf buyers stretching to pre-owned designer are exactly the REHAUS customer. |
| Honest Soho Home Review | 130 | 16,017 | 6.1 | Captures the Soho House lifestyle audience researching the furniture line. Feeds the /collections/soho-home page directly below it. |
| Honest Made.com Review | 88 | 50,866 | 3.5 | Highest impression volume and best average position of any article. Position 3.5 on a review of one of the UK's most-searched furniture brands. |
| The Ligne Roset Togo: Definitive Guide | 47 | 3,126 | 7.5 | The hero-piece treatment for REHAUS's centre-of-gravity product. Supports the togo-sofa collection targeting 8,700 searches a month. |
| Honest DUSK Review | 41 | 4,843 | 6.6 | Mid-market hijack working down the price ladder. Part of a review cluster that now spans 14 competitor brands. |
| Honest Tylko Review | 52 | 1,871 | 7.6 | Paired deliberately with the /collections/tylko page, the site's top non-brand collection. Article answers the research query, collection takes the buyer. |
| + the rest of the cluster | 28 published · 22 already earning impressions | Swyft, Swoon, SCS, DFS, Wayfair, Design Within Reach, USM Haller, Mah Jong, B&B Italia Sofas, Loaf vs B&B Italia and more. 1,110 clicks and 140,231 impressions from the article layer since the first piece went live in May, with 16 articles published in July alone still maturing. | ||
And here's what compounding looks like when it arrives. Comparing 1 to 29 June against 1 to 29 July, so the day counts match exactly: editorial clicks went from 167 to 613. Up 267% in a month. The Sofology review jumped 57 to 199. Loaf went 31 to 184, a 494% rise. Made.com more than quintupled from 10 to 54. And the Tylko, DUSK and DFS reviews all started ranking from a standing start, three pages going from nothing to 59 combined clicks in their first weeks.
Ask ChatGPT for the best second hand designer furniture and it now answers with a ranked table. REHAUS sits at #1. "Best overall." Above Vinterior, Pamono, 1stDibs and eBay.
"I'd put REHAUS first if your definition of 'designer furniture' means brands such as Ligne Roset, Roche Bobois, Cassina, B&B Italia, Poltrona Frau, Vitra and Knoll. The important difference is that REHAUS operates as a specialist reseller rather than simply an open marketplace. It collects furniture across the UK, authenticates it, prepares it and stores it before sale."
ChatGPT, August 2026, with REHAUS cited as the source and a "Shop REHAUS" link in the answer. That description isn't luck. It's the canonical language from the AI information page, llms.txt and the review content, repeated back almost word for word.
Worth being precise about why this matters commercially: AI search traffic converts up to 22x higher than traditional search traffic, because the assistant has already done the qualifying. When ChatGPT tells a buyer "REHAUS, best overall" and hands them a link, that's not an impression. That's a referral. This is the same layer I build in every AI visibility audit.
The AI visibility audit maps how ChatGPT, Perplexity and Google's AI results describe your brand today, then sets out exactly what to change. Same playbook, your brand.
COLLECTION PAGES,
REBUILT IN CODE
Here's the thing about e-commerce SEO that blog-first strategies miss: a collection page ranking beats a blog post ranking, every time. The visitor lands one click from checkout, not three. And REHAUS's collection pages were sitting in striking distance of page one for serious money terms, with almost nothing on them. No H1. An eight-character meta. A product grid and silence.
The Ahrefs data made the priority list for me. "Togo sofa": 8,700 UK searches a month, REHAUS at position 20. "Tylko": 3,300 a month, position 6. "Roche bobois": 3,000 a month, position 13. "Boconcept": 4,800 a month combined, position 11. One push from real traffic, all of them. This is exactly the situation money-page entity optimisation exists for.
The fix? I went into the theme code and built a new version of the default collection template. Not a plugin. Not an app. A coded template variant that adds an entity-rich intro, a description block, accordion FAQs and FAQPage schema, all rendered below the product grid so the shopping experience is untouched. Products render in the same order. Tracking stays intact. Google finally gets something to read, and it's in the crawlable HTML, not injected by JavaScript. Alicja's review confirmed it: "the copy's there and in the crawlable HTML, Googlebot sees the full block."
| Collection | Target · UK Volume | Clicks | Impressions | Signal |
|---|---|---|---|---|
| /collections/tylko | tylko · 3,300/mo | 465 | 24,490 | The site's top non-brand page. "Tylko" now averages position 6, and "tylko second hand" sits at 2.7 with a 28.9% CTR. Buyer-ready traffic straight into inventory. |
| /collections/togo-sofa | togo sofa · 8,700/mo | 202 | 21,495 | The centre-of-gravity term. From position 20 pre-rebuild to averaging 14 and climbing, on the single biggest commercial keyword REHAUS can own. |
| /collections/soho-home | soho home second hand | 201 | 6,809 | 2.95% CTR. Paired with the Honest Soho Home Review above it in the funnel. The article wins the research query, this page takes the buyer. |
| /collections/roche-bobois | roche bobois · 3,000/mo | 112 | 7,798 | Cracked the top 10 in July from position 13, once the brand_ duplicate stopped splitting its signals. "Roche bobois sofa" ranking alongside at 9.6. |
| /collections/sofa | category head term | 110 | 6,571 | Average position 6.2 on the core category page. Category, brand and designer collections all lifted by the same template change. |
| All collection URLs | 2,520 clicks | 224,175 | 161 collection pages now earn impressions on rehaus.co. The template does the heavy lifting on every single one, which is the point of fixing templates instead of pages. | |
The month-on-month numbers show the template earning its keep. Day-matched, 1 to 29 June against 1 to 29 July: collection page clicks rose 26.4% from 673 to 851, and collection impressions rose 26.3% from 61,952 to 78,255. The best part? 13 of the 30 pages relaunched in July went live in the final fortnight and hadn't even registered in the data yet, with roughly 91,000 monthly searches of combined target volume still to come through in August.
One detail worth stealing: the copy sits below the products deliberately. Customers rarely need it. Google expects it as a secondary element of a collection page, not the primary one. So the page ranks better without converting worse. That's the whole trick with money pages: you optimise for the algorithm without taxing the buyer.
THE REDIRECT
RESCUE
This was the big one. The site had 5,270 live redirect rules, and when I audited every single row, 82% of them pointed back at the dead rehaus.co.uk domain. Nearly all resolved to the homepage. Google treats an irrelevant homepage redirect as a soft 404, so the site wasn't just failing to pass equity. It was teaching Google that thousands of its URLs were worthless.
The upstream culprit? An SEO app's "broken link" feature, quietly auto-redirecting every 404 on the new domain back to the old one, which then bounced back again. A two-hop chain ending nowhere useful, on every broken URL, for months. Bad for Google, bad for AI crawlers, bad for anyone landing from an old link.
Then the legacy layer. Before the migration, 530 product URLs had been switched from numeric IDs like /products/1507083949 to descriptive slugs, with no redirects. Those old URLs held real equity: the top 1,000 old product pages carried 9,655 clicks and 791,264 impressions over 16 months. The only reliable bridge between old and new was the Shopify product ID, so I pulled a product export from the old store and mapped every one. 705 of the top 1,000 historical URLs, 78% of historic clicks, now land on the correct product page. A further 1,600 numeric redirects came from working through a legacy URL list the same way.
The matching logic ran in strict order: exact product match first, then collection match, then brand match, then product-type match. Where a dying URL had a category hint in the slug ("sofa", "armchair", "sideboard"), it redirects to the matching collection rather than 404ing. And before any URL was allowed to die, it went through a safety net: Ahrefs referring-domain check plus 16 months of GSC clicks. Anything with backlinks or traffic got elevated to a fallback. The rest die cleanly, because Google prefers an honest 404 to a dishonest redirect.
And the duplicates. Two collection URL patterns had been running in parallel for years: /collections/artek and /collections/brand_artek, each with different products, each splitting the other's ranking signals. Roche Bobois was literally ranking through its brand_ duplicate. After confirming product parity with the client team, I consolidated all 113 brand_ URLs into their clean equivalents. One page per brand, all signals in one place.
The client asked a fair question before sign-off: won't thousands of removed redirects spike our 404s? Yes, briefly, and that's the correct outcome. I documented the whole call with sources from Google Search Central before touching anything, saved rollback exports of both stores' tables, and tested the chains manually before the bulk work. Cautious and slower beats fast and wrong when you're moving 5,000 URLs' worth of equity. This is the discipline a proper technical SEO audit buys you.
A MIGRATION PUT
BACK ON RAILS
A domain migration isn't a DNS switch. It's an equity transfer, and every hop in the chain has to resolve or the transfer fails. When I arrived, three separate hops were broken at once.
First, the DNS itself. The GoDaddy records had never been updated for the move, so .co.uk requests weren't even reaching Shopify's redirect layer. Fixed in week one, once we'd untangled account access.
Second, the Shopify domain configuration. rehaus.co.uk was connected as a "redirect domain", which sends every old URL to the new homepage before the redirect table ever fires. The fix is one setting most people never touch: switch the domain type to Alias, and suddenly path-level redirects chain through. Old product URL, to new product URL, like a migration is supposed to work. My crawl of the GSC export confirmed it: search site:rehaus.co.uk and every result now resolves to its equivalent page on rehaus.co.
Third, the formal handover. Google's Change of Address tool is how you tell Google to move the old domain's ranking signals to the new one, and it needs verified ownership of the old property. Getting there took a small saga of failed DNS verifications across the team before the GoDaddy TXT record finally landed and ownership auto-verified. Then the Change of Address went through, locking the signal transfer in before peak season. Boring, fiddly, and worth more than any single piece of content on the site.
One more habit that mattered: after every structural fix, sitemaps were resubmitted and the top organic landing pages force-reindexed, pulling Google's recrawl forward by days each time. Recovery timing isn't luck. You schedule it, ideally ahead of the next algorithm update rather than behind it.
THE UNGLAMOROUS WINS
THAT ADD UP
My note to the client midway through says it plainly: every issue I dived into had five more behind it. Half-removed SEO apps, leftover scripts, duplicate schema, contradictory settings. None of these fixes makes a headline on its own. Together, they're the difference between a site Google trusts and a site Google tolerates.
The app behind the redirect mess and the metadata overwrites. However, it was also generating nearly all the site's schema, so ripping it out on day one would have caused an AEO regression. The play: full audit, kill the 404 auto-redirect and the meta overwrite immediately, rebuild the schema natively in the theme, then uninstall. Phased, documented, zero regression.
A previously uninstalled SEO app had left script tags, metafields and theme snippets behind, including live Universal Analytics tags buried in the theme. Dead weight on every page load. Found it, confirmed the app was long gone, stripped it all out.
"Official Information About REHAUS": a canonical reference page built for AI assistants, not customers. Declarative 100 to 300 token chunks, structured tables, definitive answers, every claim fact-checked with the client before publishing. This is the page ChatGPT now echoes back word for word.
The AI-crawler equivalent of a sitemap: a curated file telling LLMs what REHAUS is, what it sells, and which pages to trust. Five facts confirmed with the client before deployment, live within the week.
The old site's second-biggest page: 3,554 clicks and 331,737 impressions in its .co.uk life. I rebuilt the copy only: H1 fix, new meta, entity-optimised sections and a 26-question FAQ with FAQPage schema. The result? Sell page clicks up 70% month on month in July (180 to 306), impressions up 45.9%. Also found and redirected a forgotten duplicate sell page with a broken form that was still live.
Visitors from Brazil and most of the world were getting a blank page: no Rest of World market existed in Shopify, and "International" secretly meant US-only. Diagnosed it, presented the options, got sign-off, and now the store renders worldwide.
Every live link still pointing at .co.uk, pulled from Ahrefs and triaged by quality: 13 genuine editorial targets (MO Magazine, House & Garden, Country & Town House), 8 self-owned profiles fixed directly, and 83 spam domains queued for disavow rather than chased.
Out-of-stock product pages were injecting JavaScript that returned 403s to AI crawlers, invisible to a standard Google check. Diagnosed the pattern, flagged it for dev support, and it's on the fix list with a workaround documented. AI search checks differently to Google. You have to test both.
A full author page for the founder, built, verified and linked from every article. Why would Google trust a B&B Italia guide from an unnamed source? It wouldn't. Now every piece carries a named, credentialed author with a dedicated page behind it.
THE
RESULTS
GSC data for rehaus.co: 3 April (migration day) to 9 August 2026. This is a brand new domain inheriting a broken migration, measured from day zero. Read the curve, not just the totals. It's a textbook recovery: visibility first, positions second, clicks third.
| Month | Impressions | Clicks | Avg Pos | Note |
|---|---|---|---|---|
| April 2026 (from the 3rd) | 37,123 | 1,695 | 22.2 | Migration month. DNS fixed, domain switched to Alias, redirect audit begins. Clicks mostly brand searches finding the new domain. |
| May 2026 | 57,571 | 1,572 | 16.4 | The full redirect rebuild goes live on 13 May. Impressions +55% as Google recrawls. Clicks flat while rankings reset, expected and flagged in advance. |
| June 2026 | 170,720 | 2,293 | 11.4 | The visibility inflection. Impressions triple in a month as the redirect rescue and metadata rebuild get recrawled. Peak day 21,592 impressions on 19 June. Collection relaunches begin shipping. |
| July 2026 | 218,042 | 3,229 | 10.6 | The click inflection. First 3,000-click month, +41% on June, and the month non-brand crossed 50% of clicks for the first time. Made.com review at position 3, Tylko collection at 6, Roche Bobois in the top 10, 30 more collections relaunched. |
| August 2026 (to 9 Aug) | 63,066 | 1,044 | 11.0 | Tracking toward the best month yet: 116 clicks a day average, record 152-click day on 5 August, and every one of the six biggest click days of the engagement inside the last three weeks. |
One number in the July data looks like bad news and isn't: homepage impressions fell 43.6%. However, homepage clicks were near flat and its CTR nearly doubled, 3.94% to 6.71%. What actually happened is that Google moved broad queries off the homepage and onto the pages now built to answer them, mostly the rebuilt collections. It reads like a drop. It's a move. That redistribution, away from a homepage catch-all and onto intent-matched money pages, is exactly what the template work was designed to cause.
KEY
INSIGHTS
Migration rescues are where SEO theory meets consequences. Four things the REHAUS data settles.
IT'S AN EQUITY TRANSFER.
REHAUS had "done the migration": new domain live, old domain redirecting, job apparently finished. However, DNS, domain configuration, path-level redirects and Change of Address are four separate hops, and equity only moves when all four resolve. One wrong Shopify setting (Redirect instead of Alias) was silently converting sixteen months of history into homepage bounces. The lesson: audit the chain end to end, URL by URL. Assume nothing transferred until you've watched it transfer.
BEFORE CONTENT DOES.
The fastest wins on REHAUS didn't come from new pages. They came from giving Google something to read on pages it already wanted to rank: H1s on 335 collection pages that had none, real metas replacing "@ REHAUS", copy and FAQs coded into the collection template itself. Fix a page, you improve a page. Fix a template, you improve 246 at once. On e-commerce sites, the template layer is almost always the highest-leverage work available, and almost always ignored.
STRUCTURE, NOT LUCK.
ChatGPT describes REHAUS as "a specialist reseller rather than simply an open marketplace" that "collects, authenticates, prepares and stores" furniture. That is the canonical language I wrote into the AI information page and llms.txt, repeated back almost verbatim, with REHAUS cited and ranked #1. LLMs cite declarative chunks from sources they can parse and trust. Give them a clean canonical source, rigidly structured reviews and single-source schema, and the answer engine builds your brand's answer from your own words.
COMPOUND IN PARALLEL.
The client pushed hard for commercial content to ship alongside the technical rescue rather than behind it, and the data proved the sequencing right. Redirects and templates rebuilt trust while articles built reach, and they converged in July: editorial clicks +267%, collection clicks +26%, non-brand past 50% of all clicks for the first time, six record click days in three weeks. Foundations without content recover slowly. Content without foundations leaks. Together, on a schedule, they compound.
WHERE THIS
GOES NEXT
Four months in, the foundations hold and the curve is still steepening. Here's what's already in motion, because a recovery isn't the goal. It's the starting line.
This case study updates as the results roll in. Figures last refreshed 11 August 2026, from Google Search Console. Check back.
ABOUT
GRAEME
Migration rescues are the deep end of technical SEO. There's no template to follow, no clean starting point, and every day of delay costs the client real money. REHAUS is what my methodology looks like under those conditions: forensic auditing first, surgical fixes second, and a content engine built on top once the foundations hold.
I'm an independent SEO and AEO strategist based in Horsforth, Leeds, with over a decade in content marketing. I work with a small number of clients at a time, senior-only delivery, no juniors and no account managers. The person who audits the site is the person who fixes it and the person who writes for it.
My background is in journalism, via Worcester University's Journalism School. That editorial instinct is why the REHAUS review content reads like it was written by someone with opinions rather than a keyword tool, and why ChatGPT treats it as citable.
FAQ
Everything you might want to know about the REHAUS rescue and what it demonstrates about the GWContent methodology.
STOPPED. THEN IT
COMPOUNDED.
Mid-migration, post-migration, or invisible to AI search: the REHAUS playbook applies. I work with a small number of clients at a time, so every engagement gets this level of attention. Get in touch and I'll tell you what's recoverable.