Track ChatGPT Brand Mentions With Ahrefs Brand Radar: What the June 2025 Update Actually Measured

Ahrefs launched 17 product updates in June 2025, but the one that changed the AI-search conversation was Brand Radar leaving beta with indexes for ChatGPT and Perplexity. If you want to track ChatGPT brand mentions, this guide explains exactly what that June 2025 feature measured, what a result could tell you, and how to avoid mistaking a sampled AI answer for proof of overall demand.

The practical payoff is simple: we can separate a useful visibility signal—your brand appearing for a relevant buyer question—from a vague dashboard number. We will also show how to compare competitor appearances, citations, and share of answer across AI engines without blending later Brand Radar capabilities into what was actually available at launch. (ahrefs.com)

What Ahrefs Brand Radar added in June 2025

In the June 2025 release, Ahrefs said Brand Radar was officially out of beta. The headline AI feature was an LLM index that let users search for brand mentions in ChatGPT and Perplexity; Gemini was described as coming soon. The product announcement showed an example search for “Pokémon,” returning questions related to that entity and indicating how it appeared in ChatGPT answers. (ahrefs.com)

At that moment, the dataset had only two months of LLM-index data and Ahrefs planned monthly updates. That detail matters. The June feature was not a live feed of every answer that every ChatGPT user received. It was an indexed, periodically refreshed view of a defined prompt set.

The same announcement also introduced a search-demand index for brand-related keywords, plus an author-presence filter in Web Visibility. Those are adjacent signals, not substitutes for an AI-answer measurement:

• Search demand estimates interest in a brand through related keyword volume.
• Web visibility identifies mentions on web pages.
• LLM visibility shows whether a brand appears in stored AI-generated answers for the prompts tracked.

In June 2025, Ahrefs priced AI Overviews, ChatGPT, and Perplexity as separate add-ons at $99 per month per index, while search demand was included with paid Ahrefs subscriptions. That was historical launch pricing, not a reliable statement of current pricing or packaging. (ahrefs.com)

How to track ChatGPT brand mentions in Brand Radar

The core June 2025 workflow was straightforward: search for a brand, product, or entity in the ChatGPT index, then inspect the questions and answers where it appeared. This is best understood as entity discovery across a pre-collected prompt dataset, rather than rank tracking for a fixed keyword list.

A disciplined workflow looks like this:

1. Search the brand and its common variants. Include the company name, flagship product, abbreviated name, and any names users may use conversationally.
2. Filter to the ChatGPT dataset. Do not combine ChatGPT findings with Perplexity or Google AI Overview results before reviewing them separately.
3. Read the full prompt and answer. A mention in “What are the best CRM tools for a 10-person agency?” has different value from one in a broad definition query.
4. Record the surrounding competitors. A single appearance is less informative than the brands ChatGPT named before, after, or instead of yours.
5. Group findings by buyer intent. Separate comparison, recommendation, alternative, pricing, implementation, and problem-aware prompts.
6. Repeat on the refresh cadence. Compare like with like: same platform, same country or audience setting where applicable, and the same tracked prompt set.

The output should be an evidence log, not merely a visibility score. For each important answer, we recommend retaining the prompt, answer date, brand mention status, cited sources when available, named competitors, and a human assessment of relevance.

That distinction is central to AI search measurement: measurement begins with preserving the question and answer that produced a metric. Without that evidence, it is difficult to tell whether a change reflects a meaningful market shift, a change in the prompt sample, or a different answer generation run.

What a ChatGPT mention does—and does not—mean

A ChatGPT brand mention means that your entity appeared at least once in a response to a specific prompt. It does not automatically mean that ChatGPT recommended you, cited your website, described you accurately, or would return the same response to another user.

Ahrefs’ current metric definition makes this explicit: multiple repetitions of the same brand within one AI response count as one mention. For example, if an answer names Salesforce three times, it is still one mention for that response. (help.ahrefs.com)

That is a sensible counting rule, but it creates a reporting obligation. We should pair mention counts with qualitative checks:

• Was the brand described as a leading option, a niche alternative, or a warning?
• Did the answer answer a commercial buyer question or a generic informational question?
• Was the brand one of two options or one of 15?
• Did the answer name a competitor more prominently?
• Was the response supported by a citation, and if so, whose page was cited?

Consider a B2B analytics company. It may show up in 20 answers about “best dashboard tools,” yet be absent from 10 high-intent prompts such as “best analytics platform for multi-location retail.” The second gap is often more actionable than the first total.

This is why we treat visibility as a distribution across prompts and intent clusters. A raw mention total is an entry point. It is not the conclusion.

Mentions, citations, impressions, and AI Share of Voice are different metrics

The June 2025 announcement focused on finding brand mentions, but subsequent Brand Radar documentation separates AI visibility into four measures: mentions, citations, impressions, and AI Share of Voice (SOV). These should not be used interchangeably. (help.ahrefs.com)

Mentions

A mention answers: “In how many AI responses did our brand appear?” It is useful for identifying whether the model recognizes and includes your brand in relevant conversations.

Citations

A citation answers: “Did the AI answer visibly cite a page from our domain or another domain?” Ahrefs distinguishes cited pages from pages that may have been retrieved in the background but not displayed as an in-line source. A page can be “found in” an answer-generation process without being cited to the user. (help.ahrefs.com)

This difference is especially important for ChatGPT. A brand can be named without its own website being cited; conversely, a page can be cited while the answer does not frame the company as a recommended brand.

Impressions

Ahrefs calculates estimated impressions from the Google search volume associated with prompts where a brand appears. That makes impressions a modeled estimate of potential visibility, not a count of people who actually saw an answer. Ahrefs’ own methodology says these metrics indicate potential visibility rather than actual audience reach. (help.ahrefs.com)

AI Share of Voice

AI Share of Voice compares your presence with specified competitors or a wider category. It is most useful when the competitor set is deliberate. Comparing a regional accounting platform only against global enterprise suites can make a percentage look precise while hiding the competitors buyers actually evaluate.

For a fuller metric framework, see our guide to the AI Visibility Index. The key principle is that share of voice needs a documented denominator: which prompts, which AI engine, which geography, which time window, and which competitors were included.

ChatGPT, Perplexity, Gemini, and Google AI Overviews should be reported separately

The June release gave marketers ChatGPT and Perplexity tracking, while Gemini was still future coverage. By July 2025, Ahrefs announced Gemini and Microsoft Copilot indexes and said it had expanded ChatGPT and Perplexity datasets by 10 times; most ChatGPT results then also included citations. Those are important later developments, but they should not be retroactively attributed to the June 2025 launch. (ahrefs.com)

As of Ahrefs documentation published in July 2026, Brand Radar describes coverage across Google AI Overviews, Google AI Mode, ChatGPT, Perplexity, Gemini, Copilot, Grok, and Claude for certain custom-prompt use cases. The documentation notes that new Grok data collection was temporarily unavailable because of policy changes. (help.ahrefs.com)

We should also use current names carefully. Google SGE was Google’s earlier Search Generative Experience label; current reporting should generally distinguish Google AI Overviews and Google AI Mode rather than treating “SGE” as a current product name.

A cross-engine dashboard should therefore have separate columns, not one blended “AI visibility” total:

Ahrefs’ June 2025 analysis found only 7 of the top 50 mentioned sources in common across ChatGPT, Perplexity, and AI Overviews—14% overlap—using its June datasets. That is strong evidence against assuming that Google rankings or one AI engine’s citations predict visibility everywhere else. (ahrefs.com)

How to find competitor gaps in AI-generated answers

The most useful outcome from tracking brand mentions is not “we were mentioned 43 times.” It is discovering the prompts where a competitor is present and your brand is absent, then determining whether the gap is real, relevant, and addressable.

Start with a constrained comparison set of three to eight competitors. Include direct competitors, high-authority category leaders, and alternatives that prospects repeatedly mention in sales calls or reviews. Then classify prompts into intent clusters.

For example, a project-management software company might create these clusters:

• “Best project management software for agencies” — category recommendation
• “Asana alternatives for creative teams” — competitor displacement
• “How to manage client approvals in one place” — problem-aware
• “Project management tool with time tracking and budgets” — feature-led evaluation

For each cluster, calculate three practical measures:

1. Brand presence rate: the proportion of reviewed answers that mention your brand.
2. Competitor gap rate: the proportion that mention a named competitor but exclude you.
3. Answer position or prominence: whether the brand is the first option, an unranked example, or mentioned with a negative qualifier.

This approach gets closer to share of answer, meaning the portion of relevant answers where the brand earns a meaningful place. It is more decision-useful than treating every appearance as equal.

We should not assume that a competitor’s higher AI Share of Voice proves superior product quality. It may reflect category language, brand recognition, public documentation, third-party coverage, retrieval behavior, or the exact prompt mix. The correct next action is to inspect the answer-level evidence.

Can you track custom prompts rather than rely on a sampled index?

Yes. This is the missing layer in many AI visibility reports. A broad prompt index is valuable for discovery, but it cannot guarantee coverage of the 50 questions your sales team hears every month.

Ahrefs’ current custom-prompt feature lets users choose prompts, AI assistants, location, and refresh frequency. Its supported list includes Google AI Overviews, Google AI Mode, ChatGPT, Perplexity, Gemini, Copilot, and Grok, with checks available daily, weekly, or monthly; availability constraints can change by platform. (help.ahrefs.com)

Custom AI prompt tracking is where we recommend building a controlled benchmark. Use prompts that represent actual customer language, preserve the exact wording, and rerun them on a documented schedule. Examples include:

• “What is the best payroll provider for a 75-person construction company?”
• “Which CRM integrates with [specific tool] for a small legal team?”
• “What are the alternatives to [competitor] for European data residency?”

A local-first workflow gives us another option: run those exact prompts from a desktop app using the customer’s own API key, retain each response locally, and compare the answers directly over time. Our AI Visibility Tracker is designed around that reproducibility: prompt-level checks, competitor gaps, and share-of-answer analysis across ChatGPT, Claude, Gemini, Perplexity, and Grok.

That does not make a local-first tracker inherently “better” for broad market discovery. It solves a different problem: direct control over the prompts, runs, data retention, and API usage. For teams weighing those models, we compare the trade-offs in AI Visibility Tracker vs cloud AI search visibility tools.

The limits of sampled prompts and changing AI answers

Every AI visibility measurement system has limits. AI answers are not fixed rankings: the same prompt can produce different wording, sources, and brands across time, locations, accounts, modes, or retrieval conditions.

Ahrefs says its broad dataset is anchored in real search behavior through its keyword database and Google People Also Ask data, then expanded with semantic fanout. It runs questions across platforms, stores responses, updates question sets, and tests chatbots monthly with a 90-day reporting window. That methodology provides breadth, but it still represents a modeled sample rather than a census of all AI conversations. (ahrefs.com)

We should account for five limitations in every report:

• Prompt selection bias: a search-backed dataset may underrepresent questions people ask only in private chat.
• Temporal variation: an answer retrieved in August may differ from the same prompt in September.
• Personalization and geography: regional context and user settings can affect AI output.
• Entity ambiguity: a product name can overlap with another company, person, or common word.
• Citation ambiguity: a cited source is not always the cause of a brand mention, and a mention is not always an endorsement.

The fix is not to abandon measurement. It is to attach methodology to every number and use a two-layer system: broad discovery from an indexed dataset plus controlled testing on the exact prompts that matter to the business.

A practical reporting template for AI visibility tracking

A monthly report should let a marketer move from observation to action in under an hour. We recommend one dashboard for leadership and one answer-level evidence table for the people who will investigate gaps.

Leadership view

Track the following by engine and intent cluster:

• Mention rate
• AI Share of Voice against a fixed competitor set
• Estimated impressions, clearly labeled as modeled potential visibility
• Number of high-intent competitor gaps
• Number of owned-domain citations
• Month-over-month changes with notes on prompt-set or platform changes

Evidence view

For every material change, retain:

This format protects the team from a common reporting error: reacting to an aggregate metric without knowing which specific answers changed. It also creates a cleaner handoff between SEO, content, PR, product marketing, and sales enablement.

If citations are a central KPI, read our comparison of AI citations vs AI visibility tracking. Citation reporting is valuable, but it should sit beside brand inclusion and competitor context—not replace them.

What the June 2025 update taught marketers

Ahrefs’ June 2025 Brand Radar release made a valuable capability more accessible: marketers could search a brand against stored ChatGPT and Perplexity answers rather than manually testing a handful of prompts. It established an important category of measurement—AI-generated answer visibility—and made the question of brand mentions operational. (ahrefs.com)

But the correct interpretation was never “this tells us everything ChatGPT says about our brand.” It told users how a brand appeared in Ahrefs’ indexed prompt sample at the recorded time. That can reveal category opportunities and competitor gaps, especially when reviewed at the answer level.

Our working standard is therefore simple: use broad indexes to find the unknown unknowns, use custom prompts to test known buyer questions, and retain the exact evidence behind every conclusion. That is how AI visibility tracking becomes a repeatable measurement practice rather than a monthly screenshot of a changing chatbot.

FAQ

How do I find my brand mentions in ChatGPT with Ahrefs Brand Radar?

Search your brand, product, or entity in Brand Radar and select the ChatGPT data source. Review the prompts and stored AI responses where the brand appears, then inspect named competitors and any citations. Treat the output as a sampled, time-bound dataset; save high-value prompts and compare results consistently over future refreshes. (ahrefs.com)

What AI platforms does Brand Radar track?

At launch in June 2025, Brand Radar added ChatGPT and Perplexity indexes, with Gemini described as coming soon. Ahrefs’ July 2026 documentation lists AI Overviews, AI Mode, ChatGPT, Perplexity, Gemini, Copilot, Grok, and Claude for specified use cases, though platform availability can change. (ahrefs.com)

How does Brand Radar measure AI visibility, mentions, citations, and Share of Voice?

A mention is counted when a brand appears in an AI-generated response, while a citation is counted when a page is visibly cited in that response. Ahrefs also models impressions from associated search volume and calculates AI Share of Voice relative to competitors or a comparison set. These are potential-visibility metrics, not audited audience reach. (help.ahrefs.com)

Can I track custom prompts and competitor mentions in AI-generated answers?

Yes. Ahrefs’ custom prompt tracking lets users choose questions, AI assistants, locations, and daily, weekly, or monthly refresh schedules. A local-first tracker can also run a fixed buyer-question set through your own API key, which gives you direct control over the prompts, outputs, and stored evidence for competitor comparisons. (help.ahrefs.com)

How much do ChatGPT answers overlap with Google search results?

They should not be assumed to overlap closely. In Ahrefs’ June 2025 analysis, only 7 of the top 50 mentioned sources were shared across ChatGPT, Perplexity, and Google AI Overviews. Search behavior, retrieval systems, citations, licensing, and platform-specific biases can produce materially different source and brand visibility patterns. (ahrefs.com)

What are the alternatives to Ahrefs Brand Radar for tracking AI visibility?

The main alternatives divide into broad cloud platforms that supply their own prompt datasets and controlled trackers that rerun a brand’s selected prompts. We recommend choosing based on the question: use broad datasets for market discovery, then use custom or local-first prompt monitoring when privacy, reproducibility, direct API control, and answer-level evidence matter most.