Case study: Gumlet turned ChatGPT mentions into 20% of inbound revenue. Read it →
Moment 06 · Post-migration
You rebranded. The launch went well. Then you checked ChatGPT.
The new name is rolled out. The new domain redirects are clean. The new positioning is on the site. Everything looks like a successful rebrand. Then you typed your new brand name into ChatGPT, and it returned the old company name. Or worse, it returned your competitor. Or worst of all, it described you as something you used to be three years ago.
★★★★★
5.0 on Clutch · Proven on Gumlet, REsimpli, Fable
You
What does [New Brand Name] do?
ChatGPT
[Old Brand Name], formerly known as something else, is a B2B SaaS company that helps teams with workflow automation. They’ve been around since 2019 and are headquartered in...
Old identity returned. Positioning is 3 years out of date. Domain reference is stale.
Why LLMs don’t see your rebrand
A rebrand is an event for you. A slow-motion data update for LLMs.
ChatGPT’s training data has a knowledge cutoff. Perplexity uses real-time retrieval but weights established entity associations heavily. Claude prefers documented sources over recent announcements. Gemini cross-references Google’s knowledge graph, which updates on its own schedule. Even after a clean technical rebrand, here’s what LLMs continue to do.
✕
Refer to you by the old brand name.
✕
Describe you using positioning from your previous identity.
✕
Cite the old domain when sources reference your content.
✕
Conflate you with competitors if the new positioning sits closer to theirs.
✕
Show inconsistent entity recognition across LLMs (ChatGPT updates, Perplexity hasn’t).
✕
Prefer consistency across many sources over recency in one source.
This isn’t a bug. It’s how LLMs handle entity updates. Until enough new sources exist that consistently describe you under the new identity, the old identity persists. The result: your rebrand is invisible to the channel that 87% of B2B software buyers are using to research vendors.
How long each engine takes
The four engines update on four different clocks.
Each LLM has a different mechanism for incorporating new entity information. Plan for the slowest one.
ChatGPT
60–120
days
Claude
2–3
months
Gemini
4–8
weeks
Perplexity
2–4
weeks
The post-rebrand recovery playbook
Four phases. Twelve weeks to a clean LLM identity.
Most of the answers you’ll find about rebrand recovery are about Google migrations: 301 redirects, sitemap updates, Search Console reconfiguration. Those are necessary but not sufficient. The GEO half is structurally different. Here’s what we’d run.
1
Weeks 1–2
Audit the current LLM understanding.
We figure out exactly how each LLM currently describes you. The damage report comes first.
01
Prompt audit across all four LLMs. 30–50 prompts about you and your category, capturing how ChatGPT, Perplexity, Claude, and Gemini currently describe your brand.
02
Old-name retention scoring. What percentage of mentions still reference the old name, the old positioning, or the old domain.
03
Conflation analysis. Are LLMs confusing you with a competitor, an acquirer, or a previous version of yourself.
04
Citation source audit. Which sources LLMs are pulling from to form their current understanding, and whether those sources have been updated.
By end of week 2: a precise map of what each LLM thinks you are vs what you want them to think.
2
Weeks 3–4
The entity reset.
The highest-leverage work. Every authoritative source LLMs use to form entity understanding gets updated.
01
Wikipedia. If you have an entry, update it. If you don’t, evaluate eligibility and submit if appropriate.
02
Crunchbase. Full profile update including company description, category tags, founders, funding history.
03
G2, Capterra, TrustRadius. Listings updated with new name, new positioning, new logo, new feature descriptions.
04
LinkedIn. Company page, founder profiles, employee profile bulk updates, About section rewritten.
05
Industry directories. Every listing where you appear (BuiltIn, Product Hunt, AngelList, etc.).
06
Press release database. Historical releases that reference the old name don’t get deleted. New release distribution under the new name happens here.
By end of week 4: every authoritative entity source LLMs use has been updated under the new identity.
3
Weeks 5–8
Content rebuild and citation generation.
Now we generate the new sources LLMs need to incorporate the rebrand. Five workstreams in parallel.
01
The “[New name] is [category]” pillar. The definitional content that anchors LLM understanding of who you are now.
02
The “[New name] vs [Old name]” page if there’s enough search volume. Captures buyers who knew you under the old name.
03
The “[New name] [primary use case]” cluster. 4–6 pages that reinforce the new positioning across multiple angles.
04
Third-party citation building under the new name. Guest posts, podcast appearances, founder bylines, all under the new identity, in publications LLMs cite.
05
Press wave. Coordinated mentions in industry publications, newsletters, and analyst coverage under the new name.
By end of week 8: the LLM citation pool starts to include new-name sources alongside old-name sources.
4
Weeks 9–12
Tracking, reinforcement, and validation.
The final 30 days are about confirming the update is taking. By day 90, most LLMs should be referring to you primarily by the new name.
01
Bi-weekly citation tracking. We run the same prompts from week 1 and measure old-name retention vs new-name adoption.
02
Conflation tracking. Has LLM confusion between you and competitors decreased.
03
Pipeline attribution under the new name. GA4 segments configured to track AI-sourced sessions under the new domain.
04
The day-90 recovery report. Where each LLM stands on recognizing the new identity, and what the next 90 days should look like.
By day 90: most LLMs refer to you primarily by the new name, with the old name appearing as historical context, not current identity.
Why most rebrands fail at AI search recovery
Three patterns we see in unrecovered rebrands.
If you rebranded six months ago and ChatGPT still uses the old name, you’re probably here.
01
Technical migration without entity work.
The team handles 301 redirects, sitemap updates, GSC reconfiguration. They forget Wikipedia, Crunchbase, G2, and the dozen directories that LLMs treat as authoritative entity sources. Six months later, ChatGPT still uses the old name because none of the sources it pulls from have been updated.
02
New content under new name, no third-party citations.
The team publishes new blog posts under the new identity. They don’t pursue guest posts, podcast appearances, or industry citations under the new name. LLMs see the new content as one source (the company itself), which they weight less than third-party validation.
03
No measurement.
The team assumes the rebrand worked because Google is showing the new name. They don’t run the LLMs themselves. They find out 9 months later that ChatGPT still returns the old identity when buyers ask about them.
What this looks like for you
Three common rebrand profiles. Three different recoveries.
The complexity (and cost) depends on how much identity changed. Name only is easy. Identity-plus-category is medium. Post-merger is hard.
Profile 01
Name change, same category, same positioning
Easiest case · 90-day recovery
The simplest version. You changed names but the company description, category, and positioning are unchanged. Entity layer updates and new-name citation generation get you there.
Recommended package: Rank & Get Found $3,500/mo for a 90-day recovery sprint.
Profile 02
Rebrand with category repositioning
Medium difficulty · 6-month engagement
You changed names and shifted positioning. Moved upmarket, narrowed ICP, changed your category. LLMs not only need to learn the new name, they need to update the category association.
Recommended package: Own Your Category $7K/mo for a 6-month engagement.
Profile 03
Domain change post-acquisition or merger
Hardest case · 6–9 month engagement
You’re now part of a larger company or acquired a smaller one. LLMs are confused about which entity is which. The conflation problem is structural.
Recommended package: Market Leader $12K/mo or Enterprise, 6–9 months.
The proof point for this moment
One migration. Zero rankings lost.
The SEO half of a rebrand is preserving rankings. The GEO half is preserving and rebuilding the LLM entity understanding. We do both. Fable’s migration is what the SEO half looks like done right.
Fable · Accessibility testing platform
0
rankings of significance lost across the full domain migration
The SEO half done right. The GEO half is what we add on top.
Fable migrated to a new domain without losing a single ranking of significance. That’s the SEO half of a rebrand done right. The GEO half (LLM entity understanding) is the part most rebrands skip, and it’s exactly what we add.
The post-rebrand recovery work isn’t about preserving rankings; it’s about preserving and rebuilding the LLM entity understanding that drives 14.2% AI conversion vs 2.8% standard organic.
Read the Fable migration case study
Questions we get
Frequently asked.
The five questions every post-rebrand marketing leader asks us. Honest answers, not pitch answers.
Ask your own question60–120 days for ChatGPT (training data updates), 2–4 weeks for Perplexity (real-time retrieval), 4–8 weeks for Gemini (Google knowledge graph cycles), 2–3 months for Claude (slower source incorporation).
Full transition usually completes around day 90–120 if the entity work is done thoroughly.
Yes, in historical context. LLMs maintain the association between the old name and the new name as a “formerly known as” relationship if both names are well-sourced. This is actually useful. Buyers who knew you under the old name still find you.
The goal isn’t to delete the old name; it’s to make the new name the primary identifier.
301s handle the SEO half. They don’t fully handle GEO because LLM citation pools include third-party sources that reference the old domain directly.
We update those sources during the entity reset phase (the Crunchbase listing, the G2 page, the Wikipedia entry, the industry directories), so LLMs stop being trained on the old domain.
This is a recoverable but slower case. The old-name retention in LLM citation pools is now embedded in their training data and cross-referenced across many sources. We can move the needle, but expect 4–6 months for full transition instead of 3.
The first 60 days of work usually produces the most visible movement. The remaining 4 months are reinforcement.
The recovery sprint is 90 days. Most clients continue into an ongoing retainer because the entity layer needs maintenance: directories drift, Wikipedia gets edited, G2 reviews accumulate, third-party citations decay.
The first 90 days is the heavy lift. Ongoing work is maintenance and continued citation building.
What happens next
Send us your old and new brand names. We’ll show you what each LLM still returns.
We’ll run your old and new brand names through all four major LLMs live during the call, show you exactly which sources are still returning the old identity, and outline what a 90-day recovery sprint would produce for your specific situation.