---
title: "How to Track AI Mentions After a Business Rebrand: A Practical Guide"
url: "https://jayyseo.com/track-ai-mentions-after-a-rebrand/"
date: "2026-11-19T19:22:24+00:00"
updated: "2026-11-19T19:22:24+00:00"
author: "Jay"
description: "Track AI mentions after a rebrand in 7 points: set one business, service, city and date range, log real answers and sources, then act on the evidence."
category: "AEO"
---

# How to Track AI Mentions After a Business Rebrand: A Practical Guide

To track AI mentions after a rebrand, pick a single business, service, city and date window, save the real answers and sources each AI assistant gives, then choose your next move from what those answers show. The goal is simple: find out whether each of the six big assistants (ChatGPT, Perplexity, Gemini and Copilot, along with AI Overviews and AI Mode from Google) has caught up with your new name.

A new name goes live on your website in a day. AI assistants take longer. Some keep using the old name for a while, some connect the two names correctly, and some leave the business out of local answers until they find enough proof that the new name is real.

I use RadarKit for this work because it saves what a customer would actually read inside each assistant, plus the pages every reply leans on. This guide focuses on the method first. It walks through seven points, uses one illustrative business from start to finish, and shows how to turn the evidence into a clear next action.

## What Does It Mean to Track AI Mentions After a Rebrand?

Tracking AI mentions after a rebrand means checking how AI assistants name, describe and recommend your business once the name has changed. You are not only asking “does AI mention us?” You are asking which name it uses, whether the facts are right, and which sources it trusts.

Every answer you collect falls into one of five simple labels:

- **New name:** The answer uses the new name and the correct details.
- **Linked:** The answer mentions both names and explains that one replaced the other.
- **Old only:** The answer still uses the former name and never mentions the new one.
- **Mixed up:** The answer blends your business with another company, or pairs the new name with old facts.
- **Absent:** The answer does not mention the business under any name.

These five labels give you a clean way to compare one week with the next. They also stop a common trap, where any mention at all gets counted as a win even when the assistant is still sending people to an old phone number.

## Why AI Assistants Keep Using an Old Business Name

AI assistants build answers from two places. The first is what the model learned during training. The second is what it finds on the web when it searches at the moment someone asks a question.

Your own website updates right away, but the wider web does not. Directory listings, review profiles, old news stories, partner pages and social bios can carry the former name for months. When an assistant searches and finds more pages with the old name than the new one, the old name often wins.

Training data moves on a slower clock. If a model learned about your business before the change, it can keep repeating the former name until a later model update. Nobody outside the AI companies can tell you when that happens, so it pays to know which of the two clocks is behind a given answer.

## What to Fix Before You Start Tracking

Tracking works best once the basics of the name change are already in place. Otherwise, you end up measuring gaps you already know about. Before your first check, confirm these items:

- **Website:** The new name appears in the header, title tags, about page and footer, and the about page says in plain words that the business used to be called by the former name.
- **Old domain:** If the domain changed, the old one redirects page by page to the matching new pages.
- **Google Business Profile:** The profile shows the new name, and the description mentions the change. Our [Google Business Profile checks for AI search](https://jayyseo.com/google-business-profile-for-ai-search/) cover the fields that matter most.
- **Structured data:** Organization or LocalBusiness markup uses the new name, lists the former name as an alternate name, and links to your official profiles.
- **Main listings:** The top directory and review profiles in your area show the new name, phone and address.

You do not need every listing fixed before day one. The tracking will tell you which stale pages AI still leans on, and that list is far more useful than a guess.

## The Example Business Used in This Guide

This guide follows one made-up business so the steps stay concrete. The names and details below are illustrative and do not describe a real company.

| Setting | Example value |
| --- | --- |
| Former name | Fernside Home Services |
| New name | Tidewell Home |
| Business type | Home repair company with three branches |
| Service tested | Water heater repair |
| Location tested | Columbus, Ohio |
| Date range | One baseline week before launch, then four weekly checks over the 28 days after launch |
| Engines | The six assistants named in the intro |

The same setup works for a salon, a law office, a clinic, a restaurant group, a gym or a retail store. Swap in your own service and city, and keep everything else fixed for the full window.

## How to Track AI Mentions After a Business Rebrand: 7 Points

The seven points, in order:

1. State the buyer question and the check you will run.
2. List every name and lock the business, service, location and dates.
3. Build one fixed prompt set that tests both names.
4. Run the checks and record the real answers.
5. Record the sources behind every answer.
6. Compare results across weeks and engines.
7. Choose the next action and note the limits.

Each point below includes the exact checks for the example business.

## Point 1: State the Buyer Question and the Check You Will Run

Start with one sentence that says what you want to learn. For Tidewell Home, it reads: “When someone in Columbus asks AI for water heater repair help, does the answer use our new name, show the right details, and still recommend us?”

That sentence decides the rest of the test. It names the service, the city and the three things you care about: the name, the facts and the recommendation. Anything outside that sentence can wait for a later round.

Next, write down what a good result looks like at the end of the 28 days. For example:

- Branded answers about the new name use correct details in every engine you test.
- Answers that mention the former name also connect it to the new one.
- The business stays in unbranded local answers at about the same level as the baseline week.

Writing this down before any data comes in keeps the review honest. It is easy to move the goalposts after you see the first results.

## Point 2: List Every Name and Lock the Business, Service, Location and Dates

A rebrand usually leaves more names in circulation than people expect. Write down every version a customer might know:

- The former name and its short forms, such as “Fernside” or “Fernside Home”
- The new name and its short forms, such as “Tidewell”
- Common misspellings of either name
- Any parent company or franchise name, if one exists

Mark the new name as the one every answer is scored against. Then lock the four settings from the table above (business, service, city, dates) so nothing drifts during the window.

The baseline week matters more than it looks. If you only start checking after launch, you will not know whether a gap was caused by the rebrand or was already there. A [local AI visibility baseline](https://jayyseo.com/how-to-set-a-local-ai-visibility-baseline-5-steps/) taken under the former name gives you the “before” picture to compare against.

Keep the date range fixed once it starts. Weekly checks on the same weekday make it easier to compare like with like.

## Point 3: Build One Fixed Prompt Set That Tests Both Names

Your prompt set needs to cover four kinds of questions, because each one tests something different. The split between named and unnamed questions follows the same logic as [branded versus unbranded local prompts](https://jayyseo.com/branded-vs-unbranded-local-prompts-3-differences/).

| Prompt group | Example prompt | What it tests |
| --- | --- | --- |
| New name, branded | “Is Tidewell Home good for water heater repair in Columbus?” | Whether AI knows the new name and its details |
| Former name, branded | “Fernside Home Services Columbus phone number” | Whether AI still sends people to old details |
| Transition | “Is Fernside Home Services now called Tidewell Home?” | Whether AI links the two names |
| Unbranded, local | “Best water heater repair company in Columbus” | Whether the business still gets recommended at all |

For one service in one city, two or three prompts per group is usually enough. That gives you eight to twelve prompts, which is small enough to review by hand every week.

Write the prompts the way a real customer would type them. Then freeze the list. If you edit a prompt halfway through, the week before and the week after stop being comparable.

## Point 4: Run the Checks and Record the Real Answers

Now ask every prompt in every engine on your list. If you run the checks by hand, use a private browser window, stay signed out, and ask from the city you are testing. Personal history and a different location can both change what you see.

For each answer, log the same fields:

- Date, engine and prompt
- Label: New name, Linked, Old only, Mixed up or Absent
- Recommended: yes or no
- Details correct: phone, address, service area and hours
- Competitors named in the same answer
- A short quote of how the business was described

Here is what a few rows from the first week after launch could look like for the example. These rows are illustrative and show the format, not real results.

| Engine | Prompt | What the answer said | Label |
| --- | --- | --- | --- |
| ChatGPT | “Fernside Home Services Columbus phone number” | Gave the former name and the old phone number | Old only |
| Perplexity | “Is Fernside Home Services now called Tidewell Home?” | Confirmed the name change and cited the business’s announcement page | Linked |
| Google AI Mode | “Best water heater repair company in Columbus” | Listed three competitors and the former name, using a review site profile | Old only |
| Gemini | “Is Tidewell Home good for water heater repair in Columbus?” | Described the services correctly under the new name | New name |
| Microsoft Copilot | “Best water heater repair company in Columbus” | Named four other companies and no version of the business | Absent |

Keep the full answer text, not just the label. A short note like “used old phone” is much easier to act on later than a single word.

## Point 5: Record the Sources Behind Every Answer

The sources are where the fixes live. For every answer that cites pages, write down each page and whether it shows the old name, the new name or both.

In the example, the Google AI Mode answer that used the former name leaned on a review site profile that still had the old name. That one row points to a specific task: update that profile. The Perplexity answer that linked both names cited the announcement page on the business’s own site, which tells you that page is doing its job.

Also note when an answer cites nothing at all. Some answers come straight from the model’s memory without a fresh search. That difference changes your next step, so it deserves its own column.

If you want a repeatable format for this part, a [local AI citation report](https://jayyseo.com/how-to-build-a-local-ai-citation-report-5-steps/) groups cited pages by domain, so you can see which directories or review sites show up again and again.

## Point 6: Compare Results Across Weeks and Engines

At the end of each week, count the labels for each engine. A simple grid works well:

| Week | New name | Linked | Old only | Mixed up | Absent |
| --- | --- | --- | --- | --- | --- |
| Baseline (former name) |  |  |  |  |  |
| Week 1 |  |  |  |  |  |
| Week 2 |  |  |  |  |  |
| Week 3 |  |  |  |  |  |
| Week 4 |  |  |  |  |  |

Fill one grid per engine, plus one combined grid. Then work through it in this order:

- **Branded rows first:** Is “Old only” shrinking and “New name” or “Linked” growing?
- **Transition prompts next:** Do more engines link the two names each week?
- **Unbranded rows last:** Is the business still recommended at roughly the baseline rate?

Look for patterns, not single answers. One week where ChatGPT uses the former name is a data point. Four weeks in a row, across several prompts, is a trend worth acting on. The same reading habits apply when you [track local AI visibility changes](https://jayyseo.com/how-to-track-local-ai-visibility-changes-5-step/) for any other reason.

Also check who took your place. If a competitor fills the slot your business used to hold in unbranded answers, note which pages back that competitor. There is a simple way to log rivals in our [local AI competitor walkthrough](https://jayyseo.com/how-to-find-competitors-in-local-ai-answers-5-steps-2026/).

## Point 7: Choose the Next Action and Note the Limits

Each pattern in your grid points to a different next step. Use this table to match the evidence to the action:

| What the evidence shows | Likely reason | Next action |
| --- | --- | --- |
| Old only, and the answer cites a stale listing | A third-party page still shows the former name | Update that listing or ask the site owner to change it |
| Old only, and no search ran before the answer | The model is answering from older training data | Keep every source consistent and keep checking, since there is no page to fix |
| Linked in most engines | The name change is getting picked up | Keep weekly checks and shift attention to unbranded prompts |
| Mixed up with another company | Weak or conflicting signals about who you are | Make the about page, structured data and profiles state the same facts |
| Wrong details under the new name | An old profile still holds the former phone or address | Fix the profile and the top directories first |
| Absent from unbranded answers | The new name has little proof of quality yet | Build reviews, service pages and local mentions under the new name |

Choose a couple of actions per week, not ten. Then run the same prompts again the next week and see whether the answers moved.

Here is one clear limit to keep in mind: every check is a snapshot. AI answers change with time, location and wording, and no tool or process can force an assistant to use your new name. Tracking shows you what is happening and where to work. It does not promise a result or a date.

## Common Mistakes When Tracking AI Mentions After a Rebrand

A few habits make the results harder to trust. Watch out for these:

- **Checking only the new name:** You miss every answer that still sends customers to old details.
- **Testing one engine:** Assistants pick up the change at different speeds, so one engine tells only part of the story.
- **Skipping the baseline:** Without a “before” week, a drop in unbranded answers cannot be tied to the rebrand.
- **Changing prompts midway:** Even small edits break the week to week comparison.
- **Reading one answer as a trend:** Single answers vary. Look at several prompts across several weeks.
- **Letting the old domain lapse:** A working redirect helps every crawler connect the former name to the new one.

## How RadarKit Helps Track AI Mentions After a Rebrand

[RadarKit](https://radarkit.ai/) is an AI search tracking tool that records how brands and local businesses show up in AI answers, which pages those answers cite, and which competitors appear beside them. It enters each prompt in the public, consumer-facing apps of six assistants, using residential IPs, so the saved replies match what a shopper would read.

![RadarKit homepage dashboard used to track AI mentions after a rebrand](https://jayyseo.com/wp-content/uploads/2026/11/ai-mentions-after-rebrand-radarkit-homepage.jpg)

For a rebrand, a few parts of RadarKit line up directly with the seven points above:

- **Brand aliases:** Project settings hold brand and competitor aliases, so the former name and the new name can sit in one project.
- **Run a prompt now:** On launch day, you can run the prompt set right away instead of waiting for the next scheduled refresh.
- **Citation Data:** Lists the pages behind every reply, which makes stale listings with the former name easy to spot.
- **Query Fanout:** Shows the searches a model ran before it answered, which helps separate memory-based answers from search-based ones.
- **Page Radar:** Lets you watch your name change announcement page and see which models cite it and for which prompts.
- **Full history:** Keeps every run, so you can set the weeks after launch against the period before it.

![RadarKit asks prompts in real chat interfaces to track AI mentions after a rebrand](https://jayyseo.com/wp-content/uploads/2026/11/ai-mentions-after-rebrand-radarkit-real-chat-check.jpg)

One note on location: the three self-serve tiers reach 50+ countries, while targeting a single state or city is an Enterprise option. For a city test like the Columbus example, pair country level tracking with your own manual checks from the city, or talk to the RadarKit team about Enterprise.

### RadarKit Key Features

Every feature below is part of the product today. I have framed each one around what it tells you during a name change.

- **Real assistant interfaces:** Your questions go into Perplexity, ChatGPT, Microsoft Copilot, Gemini, Google AI Overviews plus AI Mode inside the very screens a shopper uses, from residential IPs in 50+ countries, so the saved reply is the one a customer would see after your rename.
- **Topics and prompt scheduling:** Group the old-name, new-name, transition and unbranded questions into topics. They refresh on your plan’s cycle, and a single prompt can be fired off right away when you want a launch-day read.
- **Visibility Score:** The share of tracked prompt and model pairs that mention you, set beside the previous period, which shows whether the new name is gaining ground week by week.
- **Brand Reputation:** A running score for how AI speaks about the brand, useful for seeing whether the tone attached to the former name carries over.
- **Average Position:** Your usual slot in a recommendation list whenever you appear, next to the slots held by rival domains.
- **Sentiment:** Labels every mention positive, mixed or negative, project-wide and model by model, and pulls out the phrases assistants keep repeating about you.
- **Local Radar:** Shows if AI recommends the business, how often and in what position, plus each competing business it names, with that business’s category, star rating and review count. A listing platform filter (Google Business Profile, Yelp and others) and a per-prompt view show which prompts each business wins.
- **Shopping Radar:** For product sellers, it saves the carousels AI shows, including product, store, price, rating and placement, how often your items show up and which stores take the top spots.
- **Ads Radar:** Logs sponsored placements beside AI replies: who placed the ad, the headline and description, where it links, the creative, and how many answers carried it.
- **Citation Data:** Splits linked citations from pages an engine only read, marks which sources carry a link to your domain, and ranks the formats (guides, listicles, reviews and so on) that earn citations. During a rebrand, this is where outdated listings surface.
- **Page Radar:** Keep an eye on one URL, such as your “we have a new name” page or a guest article, to learn which assistants pull it in and on what questions.
- **Query Fanout:** Reveals the behind-the-scenes searches an engine made before replying, and how often each engine searches in the first place. When no search shows up, the reply most likely came from memory.
- **Competitors and Share of Voice:** Puts every brand and domain that appears in your answers on one leaderboard with share of voice, average placement and rank, so you can see who filled any gap the rename left.
- **Create Content:** The Content Agent studies what assistants cite on a given prompt, the pages Google ranks, the fanout queries and the People Also Ask list, then hands you a draft that sounds like your brand. It can write how-to posts and listicles, comparison or alternatives pieces, FAQ pages and first-person reviews, which helps fill the new name’s content gaps.
- **Optimize Content:** Send an older page, such as an about page still written around the former name, so it gets rewritten around the topics and searches assistants favor right now. You approve the edit and publish to WordPress when it reads right.
- **AI Traffic Analytics:** With read-only access to Google Analytics 4, visits from AI assistants sit next to the visibility numbers.
- **Agent Analytics:** Tells you which AI bots are crawling your website, a quick way to see whether new pages are being fetched.
- **Plan Agent and Search Console:** All tiers can hook up Google Search Console. From there the Plan Agent reads your queries next to the citation records, finds the gaps, and writes a to-do sheet the other agents work from.
- **Outreach Agent:** Spots websites citing rival businesses, prepares pitch emails in your voice, and after your sign-off sends them from your own Gmail and follows up. After a rename, the same flow can ask a site to swap the former name for the new one.
- **Reddit Agent:** Surfaces the Reddit threads assistants cite in your category and flags conversations where the business is not mentioned.
- **Chat Agent:** Ask something like “which engines still use our old name?” and get a plain answer from live project data.
- **API and MCP server:** Pull RadarKit data into ChatGPT, Claude, Cursor or Codex through a REST API or an MCP connection.
- **Exports, PDF reports and Client Portal:** Every dataset (answers, cited pages, rivals, ranks, fanout queries and radar results) can leave the app as a file. Lite includes 2 branded PDF reports a month, Growth includes 10 and Pro includes 20, and a view-only Client Portal or read-only link lets owners or clients follow the rebrand without logging in.
- **Complete history:** Old runs are never dropped, so the weeks after the rename can sit right next to the weeks before it, which is exactly the before and after view a rename needs.
- **Multiple projects:** Handy when several branches or brands change names together. Projects per account: Lite 2, Growth 5, Pro 10, with no cap at all for Enterprise customers.
- **Settings, aliases and seats:** Keep the brand kit, the aliases (old and new names for you, plus rival names), the timezone and team roles in one settings area. Extra teammates: 2 on Growth, 5 on Pro, and as many as you need on Enterprise, where SSO sign-in is also included.
- **Free GEO tools:** Without signing up, you can check an existing LLMs.txt file or build a fresh one for the new domain and add the free AI SEO extension to Chrome.

### RadarKit Pricing

There are three self-serve RadarKit plans and one Enterprise plan quoted per customer. Each self-serve plan comes with a 7-day free trial that asks for no payment upfront. You can cancel whenever you like. If the first charge goes through and the fit is wrong, a message to support within 48 hours gets the whole amount back. Plan details live on the [RadarKit pricing page](https://radarkit.ai/pricing).

![RadarKit pricing plans for teams that track AI mentions after a rebrand](https://jayyseo.com/wp-content/uploads/2026/11/ai-mentions-after-rebrand-radarkit-pricing-plans.jpg)

| Plan and monthly price | Unique prompts | How often answers refresh | Answers analyzed each month | Projects | Fits a rebrand that |
| --- | --- | --- | --- | --- | --- |
| Lite, $29 | 15 | Every three days (72 hours) | 900 | 2 | Covers one service in one market |
| Growth, $79 | 50 | Once a day | 9,000 | 5 | Needs daily reads in the first weeks |
| Pro, $139 | 100 | Once a day | 18,000 | 10 | Spans several branches or services |
| Enterprise, priced on request (annual) | Set per contract | Daily or weekly, your choice | Set per contract | No limit | Needs city or state targeting |

What else changes between the tiers:

- **Lite** gives you 6 content articles or optimizations, 10 shared action credits for the Outreach and Reddit agents, 15 Chat Agent messages and 2 branded PDFs every month, backed by human support.
- **Growth** moves up to 19 articles, 20 shared action credits and 25 Chat Agent messages, with 10 branded PDFs, room for 2 more teammates and human support.
- **Pro** brings 32 articles, 40 shared action credits and 50 Chat Agent messages, with 20 branded PDFs, room for 5 more teammates and faster priority support.
- **Enterprise** unlocks 70+ locations, tracking at the state or city level, SSO, seats without a limit, custom contract terms, early feature access and a dedicated manager.

![RadarKit plan table with prompt frequency and locations to track AI mentions after a rebrand](https://jayyseo.com/wp-content/uploads/2026/11/ai-mentions-after-rebrand-radarkit-plan-coverage-table.jpg)

For the Tidewell Home example, 12 prompts checked weekly fit inside Lite. If you want daily answers during the first month after launch, Growth gives you room for more prompts and a fresher read each day.

### Best For

RadarKit is a good fit for:

- Local businesses that changed their name and want to see which AI engines use it
- Multi-location brands rolling out a new name across several branches
- Agencies managing a rebrand for a client and reporting progress each week
- Teams that want the cited pages behind each answer so they know which listings to update next

## Final Thoughts

Tracking AI mentions after a rebrand is mostly a matter of discipline. Hold the business, service, city and dates steady for the whole window. Ask the same prompts each week, record the real answers and the pages behind them, and label each answer the same way every time.

The evidence then tells you where to spend your effort. Stale listings get updated, mixed-up answers get clearer entity signals, and gaps in unbranded answers get new reviews and content under the new name. Give the process a few weeks, compare against your baseline, and let the trend guide you rather than any single answer.

## FAQs

### How long does it take AI assistants to use a new business name?

There is no fixed timeline. Answers based on live web searches can start using the new name within weeks once your main listings are updated. Answers based on training data can lag until a later model update, which nobody outside the AI companies can schedule.

### Should I still track the old business name after a rebrand?

Yes. Prompts that use the former name show whether AI still sends customers to old details, such as an old phone number. Keep those prompts in your set until most engines link the old name to the new one.

### What should I do if AI mixes my business up with another company?

Make every source you control say the same thing. Your about page, structured data, Google Business Profile and main listings should all state the new name, the former name and the same address and phone number.

### Can RadarKit track both my old and new business names?

Yes. RadarKit project settings include brand aliases, so you can keep the former name and the new name in one project and track prompts that use either one across six AI engines.

### Does RadarKit show which pages still use my old name?

RadarKit’s Citation Data shows the pages each AI answer cited or consulted. You can open those pages to see which ones still carry the former name, then add them to your update list.

---

Source: https://jayyseo.com/track-ai-mentions-after-a-rebrand/
