# Rebrand AI Search: How to Get AI to Learn Your New Name

> After a rebrand or acquisition, AI answers keep using the old name because their training data, the pages they retrieve and the listings they cite still say it. Track both names, write "New, formerly Old" everywhere you control, update the listings answers cite, keep the old name in community content, and migrate tracked questions in batches.

By Othmane Khadri, founder of Earleads. Published 2026-10-19. Canonical: https://www.earleads.com/blog/rebrand-ai-search/

## Why AI keeps using your old name

AI assistants keep using an old company name because almost everything they read about you was written before the change. A rename is a single decision on your side. On the engines' side, it's thousands of pages, listings, threads and model weights that each have to catch up, and they don't catch up together.

There's a story that comes up in founder forums about an app that renamed after a big share of its downloads had started coming from ChatGPT recommendations. The team had a plan for the logo, the domain and the app store listing. They had no plan for the assistant that was sending them users, and nobody could say whether it would recommend the new name.

Rebrand checklists cover redirects, title tags and signage. Very few cover what ChatGPT, Perplexity, Gemini, Claude and Google's AI features will say after the announcement.

The old name survives in three layers, and each one moves at its own speed.

1. **Training memory.** When an answer doesn't search the web, it comes from what the model learned in training. That knowledge stops at a cutoff date, and the cutoff can be well behind the release. OpenAI lists GPT-5's knowledge cutoff as September 30, 2024 ([OpenAI](https://developers.openai.com/api/docs/models/gpt-5)). A rename after a model's cutoff simply doesn't exist in its memory.
2. **Retrieved pages.** When an answer does search the web, it reads pages that rank for the question. Your old blog posts, a two-year-old comparison article and a press release about your launch may all still say the old name.
3. **The wider web.** Directories, review sites, data providers, knowledge bases and community threads describe you in their own words. They update when someone updates them, which is often never.

Even very large companies live with this. Google's own blog announced on February 8, 2024 that "Bard is now known as Gemini" ([Google](https://blog.google/products/gemini/bard-gemini-advanced-app/)), and every article, review and thread written before that date still says Bard. If the company that runs the search engine leaves a trail of its old name behind, so will you.

![Three layers where an old company name lives on, training memory with a knowledge cutoff, pages retrieved at answer time, and the wider web of listings and community threads, all feeding into the AI answer a buyer reads](/blog/rebrand-ai-search/diagram-1.svg)
*An AI answer draws on three layers. A rename has to reach all of them, and they update at different speeds.*

## What changes in AI answers after a company name change

Not every rename is the same problem. The scenario decides what engines get wrong and where to start, so name yours before you touch anything.

| Scenario | What AI answers tend to get wrong | Where to start |
|---|---|---|
| Straight rename, same domain, same product | Old name in answers, or both names described as two companies | "Formerly" wording on owned pages, structured data, listings |
| Rename plus new domain | Old domain cited, new one treated as a stranger | Redirects and Search Console's Change of Address, then listings |
| Repositioning into a new category | You're still described in the old category, or not named in the new one | Batch migration of tracked questions, new category pages |
| Acquisition, acquired brand absorbed | Acquired brand described as independent, or its facts blended into yours | "Now part of" wording, page by page redirects, ownership on listings |
| Acquisition, both brands kept | Wrong parent, wrong products, the two brands confused | Each site states the relationship, listings name the owner |

The common thread is continuity. Every fix below exists to tell engines that the new name is the same company they already know, not a new one with no history.

## Step 1: Track both names before you announce

Start measuring before the rename goes public. Once the announcement lands, you can't recover the baseline.

Build one list of the questions buyers ask in your category, then run each one three ways.

- **Unbranded.** "Best expense software for remote teams." Are you named, under which name, and next to whom?
- **Old name.** "What is [Old]?" "Is [Old] legit?" "[Old] pricing." These are the validation questions buyers ask once they already know you.
- **New name.** The same validation questions with the new name. Before launch, the honest answer is usually nothing, or a different company that shares the name.

Read each answer for appearance, position and description. Description matters most during a rename. An engine that names you but calls you by the old name, or describes the old product, is giving buyers a stale picture even when your visibility looks fine on a dashboard.

Run the important questions many times per engine, not once. SparkToro had 600 volunteers run 12 prompts 2,961 times and found "a <1 in 100 chance" that ChatGPT or Google's AI returns the same list of brands in any two responses, so it suggests running a prompt "60-100X" before trusting the pattern ([SparkToro](https://sparktoro.com/blog/new-research-ais-are-highly-inconsistent-when-recommending-brands-or-products-marketers-should-take-care-when-tracking-ai-visibility/)). One run after launch proves nothing in either direction. We cover the tracking setup in [AI brand tracking](/blog/ai-brand-tracking/).

Also ask sales and support to log any AI claim a prospect or customer repeats. "ChatGPT says you're the old company" or "Perplexity says you were bought" are early, free signals that a source is out of date.

## Step 2: Write "New, formerly Old" in every source you control

The single most useful sentence in a rebrand is "New Name, formerly Old Name". It gives every reader, human or machine, the link between the two in one line. Put it everywhere you control, and keep it there for longer than feels necessary.

**On the site.**

- A dated, plain-text page that says what changed, when and why. "On 3 March 2026, Old Name became New Name. Same team, same product, same customers." Engines quote plain sentences like that.
- The About page opens with it. The homepage mentions it in text, not only in an image or a logo animation.
- A short "formerly Old Name" in the title or meta description of the homepage and the most visited pages for the first months, and in the footer.
- Docs, help center and pricing pages updated, because those are the pages answers quote for product facts.

**In structured data.** Google's Organization markup supports the exact fields a rename needs. `alternateName` is "another common name that your organization goes by", `legalName` is the registered name if it differs, and `sameAs` points to your profiles on other sites ([Google](https://developers.google.com/search/docs/appearance/structured-data/organization)). Put the old name in `alternateName` and point `sameAs` at the updated profiles.

The site name Google shows works the same way. Google says site names are generated automatically from the home page and "references to it that appear on the web", that `WebSite` structured data is "most important", and that changes take "anywhere from a few days to a few weeks after publishing" ([Google](https://developers.google.com/search/docs/appearance/site-names)).

**Everywhere else you own.** Social profiles, the press kit and press release boilerplate carry the same sentence, in the same words.

## Step 3: Move the domain without breaking the trail

If the rename comes with a new domain, this is classic company name change SEO, and it matters more for AI answers than it looks. Retrieved answers cite URLs. A broken old URL is a citation pointing at nothing.

Google's site move guide is the reference. Map every old URL to its closest new equivalent, not to the homepage. "Keep the redirects for as long as possible, generally at least 1 year." If you're changing domains or subdomains, submit a Change of Address in Search Console for the old site. And "expect temporary fluctuation in site ranking during the move", with medium-sized sites taking "a few weeks or more" before Google shows the new URLs ([Google](https://developers.google.com/search/docs/crawling-indexing/site-move-with-url-changes)).

Keep the old domain registered for years, not months. Old links in articles, threads and listings will keep pointing at it, and each one is a path from what engines already know to where you are now.

## Step 4: Update the listings AI answers already cite

You can't update every page on the web that mentions you. You don't need to. Start with the ones engines actually use.

Go back to the answers from Step 1 and open the citations. The directories, review sites, comparison articles and data profiles that appear there are the ones to fix first, in order of how often they're cited. That's the same rule we use for any correction. Find the source in the citations, fix it there, then re-run, which we describe in [AI reputation management](/blog/ai-reputation-management/).

A few records deserve special care.

- **Google Business Profile.** If you have one, check whether your change counts as a rebrand. Google allows it on the same profile for "a minor name change, in which the proper nouns and services described in the business name remain unchanged and the business category remains unchanged". Otherwise "it's considered a new business", and the old profile has to be closed and a new one created ([Google](https://support.google.com/business/answer/3038177)).
- **Wikidata.** Update the item's label and keep the old name as an alias. Aliases are "alternative names for items that are placed in the Also known as column" ([Wikidata](https://www.wikidata.org/wiki/Help:Aliases)).
- **Wikipedia.** If there's an article about you, don't edit it yourself. People with a conflict of interest are "strongly discouraged from editing affected articles directly" and may "propose changes on talk pages (by using the {{edit COI}} template)" ([Wikipedia](https://en.wikipedia.org/wiki/Wikipedia:Conflict_of_interest)).
- **Review and software directories, app stores, marketplaces, data providers.** Update the name, the description and the logo, and add the "formerly" line in the description field where it's allowed.

For third-party articles that rank for your buyer questions, ask the author for a one-line update. "Old Name (now New Name)" is a small edit most editors accept, and it keeps the page's citation working for you.

## Step 5: Keep the old name in community content

This is the step most rebrand plans get backwards. The instinct is to scrub the old name everywhere. In community content, that throws away what you've earned.

Discussion forums and online communities are where many buyer questions get answered in the first person, and engines cite those threads. Those threads say the old name. So do the answers engines have already built from them. When we worked through a rename mid-engagement, the old name stayed in community mentions while the new one built up, because engines and threads knew the old name and dropping it overnight would have thrown away citations that had already been earned.

In practice that means four rules.

1. **Don't edit or delete old threads to remove the old name.** They're part of the trail. Removing them breaks the link between the question and you.
2. **In new posts and replies, write both.** "New Name (formerly Old Name)" in a genuinely useful answer connects the two names in the place engines look for first-person experience.
3. **Answer the migration questions.** "What happened to Old Name?" and "Is New Name the same as Old Name?" are real questions buyers ask. A plain answer in the thread where they're asked is worth more than a press release.
4. **Keep it up until engines describe the new name unprompted.** That's the exit criterion, covered in Step 7.

This matters because mentions appear to carry more weight than links. Ahrefs studied 75,000 brands and found branded web mentions correlated with AI Overview visibility at 0.664, against 0.218 for backlinks ([Ahrefs](https://ahrefs.com/blog/ai-overview-brand-correlation/)). Every mention that ties the two names together is part of the bridge. Our guide on [how to get cited by AI](/blog/how-to-get-cited-by-ai/) covers what makes a thread worth citing in the first place.

## Step 6: Migrate tracked questions in batches

A rename often comes with a repositioning. A tool for agencies becomes a platform for enterprises, or a single product becomes a suite. The buyer questions you track have to change too, and how you change them decides whether you can read your results.

Swap the whole question set on launch day and you reset the baseline. You can't tell whether a drop came from the rename or from harder questions.

So migrate in batches over 30 to 60 days. Keep the questions that still describe your buyers. Retire old category questions a batch at a time, and add new category questions in matching batches, each measured for a few cycles before the next swap. At every point, part of the set is comparable to the month before.

![Timeline of a rename over about 60 days, with old category questions retired in three batches while new category questions are added in matching batches, so part of the tracked set stays comparable at every point and the baseline is never reset](/blog/rebrand-ai-search/diagram-2.svg)
*Swap questions in batches. Part of the set stays comparable at every point, so you can tell the rename's effect from the new questions'.*

Keep the old name's validation questions in the set even after the batches are done. "Is [Old] legit?" will keep being asked for a long time, and you want to know what buyers hear.

## Step 7: Re-run until engines use the new name unprompted

The rebrand is done in AI answers when engines describe the new name without being prompted. That means an unbranded question names you as New Name, and a validation question about the old name explains that it's now New Name, on most runs across the engines you track.

Until then, keep the loop going. Re-run the question set on a schedule, read the citations on answers that still say the old name, and fix the source they lean on. Answers that search the web move once the cited pages change and get recrawled. Answers from training memory move only as new, consistent sources outnumber the old ones and newer models ship, which is why Steps 2 to 5 keep paying off after the obvious fixes are done.

Only then start retiring the "formerly" line from titles. Keep it on the About page, in structured data and on the rename page for good. They cost nothing and keep answering the question for anyone who still remembers the old name.

## Acquisition SEO: absorb the brand or keep both

Acquisitions add a decision that a straight rename doesn't have. Does the acquired brand disappear into yours, or do both stay?

![Decision tree for an acquisition. If the acquired brand is absorbed, write Old now part of New everywhere and redirect its pages page by page. If both brands are kept, each site states the relationship and listings name the owner. Both paths end with tracking both names](/blog/rebrand-ai-search/diagram-3.svg)
*Decide the brand architecture first. Both paths end in the same place, tracking both names until engines describe the relationship correctly.*

**If the acquired brand is absorbed.** Treat it as a rename with an extra fact. "Old, now part of New" goes on every source you control and every listing you can update. Redirect its product pages to their closest equivalents on your site, page by page, so a cited page about a feature lands on that feature. Keep its help content live under your domain for as long as customers search it, because those pages answer the questions existing users ask engines.

**If both brands are kept.** The risk is confusion, not loss. Engines may give the acquired brand the wrong parent, merge the two products or invent features from one onto the other. Each site should state the relationship in one plain sentence. Organization structured data can point to the parent, and listings and data profiles should say who owns whom. Then track both brands' validation questions, since that's where a blended answer shows up first.
## What changes for B2B and B2C companies

**B2B.** Buying committees check a vendor more than once. One person shortlists you under the old name, and the CFO asks about the new one weeks later. Both questions have to lead to the same company. The records that matter most are software directories, review sites, comparison articles and analyst or data profiles. Security and compliance pages need the new legal name, because procurement checks them. Brief sales on the "formerly" sentence, so the story on calls matches the one engines tell. If you track where pipeline comes from, add the old name to your "how did you hear about us" options for a while, as covered in [AI search analytics](/blog/ai-search-analytics/).

**B2C.** Shoppers move faster and check less. If an assistant recommends the old name and the shopper finds nothing, they pick the next option. App store listings, marketplace product titles, product feeds and review sites are usually where the old name survives longest, so update those first. Keep the old name searchable in the store listing description where it's allowed. Packaging and social profiles carry the "formerly" line through the first season.

## Rebranding checklist for AI answers

Before the announcement:

- Run the baseline on unbranded, old name and new name questions, several times per engine
- Brief sales and support to log any AI claim they hear
- Write the dated "Old is now New" page and the "formerly" sentence

At launch:

- Update the homepage, About, titles, footer, docs, pricing and help center
- Add `alternateName`, `legalName` and `sameAs` in Organization markup, and update `WebSite` markup
- If the domain changes, redirect page by page and submit a Change of Address
- Update Google Business Profile, Wikidata, directories, app stores and data providers

In the months after:

- Ask authors of cited articles for a one-line update
- Write "New (formerly Old)" in community answers, and never scrub old threads
- Migrate tracked questions in batches over 30 to 60 days
- Re-run until engines use the new name unprompted, then retire the "formerly" title tags

Every [Earleads](/) audit includes an accuracy check that lists each claim engines make about you, what's wrong or stale, and the source each answer leans on, which is the rename baseline in one document. You can [request an audit](/#audit) or read [what an AI visibility audit should show you](/blog/what-an-ai-visibility-audit-should-show-you/) first.

## What not to expect

No one controls what an assistant says, so be wary of anyone who guarantees ChatGPT will use your new name by a given week. What you control is the trail. Every source you own says the same sentence, the cited sources are updated, the old name stays where it still earns citations, and the tracking tells you honestly when the job is done.

## Frequently asked questions

### Why does ChatGPT still use our old company name?

Because most of what it knows about you says the old name. Answers from training memory reflect the web before the model's knowledge cutoff, which can sit a year or more before the model you're using. Answers that search the web retrieve pages, listings and threads that may still carry the old name. Both layers change only as the sources they read change.

### How long does it take AI to learn a new company name?

There's no fixed timeline, and anyone who promises one is guessing. Answers that search the web can switch once the cited pages are updated and recrawled. Answers from training memory switch only after enough new, consistent sources exist and a newer model ships. Judge it by re-running the same questions until engines use the new name without being prompted.

### Should we remove the old name from our website after a rebrand?

No. Keep it on the pages that explain the change, in your About page, in Organization structured data as an alternateName, and in a short "formerly" line in key titles and the footer. Buyers and engines still search for the old name. Removing it overnight breaks the link between what engines already know and what you're called now.

### What happens to AI answers when one company acquires another?

Engines often keep describing the acquired brand as independent, or blend the two companies' facts. Decide first whether the acquired brand is absorbed or kept. If absorbed, write "Old, now part of New" on every source you control and redirect its pages to their closest match. If kept, both sites state the relationship plainly and the listings say who owns whom.

### Do we need a new Google Business Profile after a name change?

Sometimes. Google's guidelines allow a rebrand on the same profile for a minor name change, where the proper nouns, services and business category stay the same. If the name changes beyond that, Google treats it as a new business, and the existing profile has to be marked closed and a new one created under the new name.

### Does this apply to consumer brands and apps too?

Yes. Shoppers ask assistants which app, store or product to pick, and a renamed consumer brand can lose those recommendations the same way a B2B company loses shortlist mentions. For B2C, the old name tends to live on in app store listings, marketplace pages, review sites and product feeds, so those are the first records to update.