Ahrefs AI Search Visibility Topics and ChatGPT Mentions
Ahrefs Brand Radar topics are keyword-modeled index research. Promptwatch topics/tags sit on prompts you typed. ChatGPT mention counts are not the same job.
Ahrefs AI search visibility topics are not the same object as ChatGPT mentions. Brand Radar sits on a corpus of 260 to 405 million prompts modeled from keyword and People Also Ask data. "Topics" in that world are clusters in an index. ChatGPT mentions on a prompt you actually hear are a ledger. Buyers mix the two because both live under "Ahrefs AI search visibility." The mix is the expensive mistake, because the two objects answer different questions and cost different amounts. A topic cluster tells you a category is showing up in AI. A mention ledger tells you whether a specific buyer question named you, and those are different jobs with different owners.
Ahrefs Brand Radar is $199/mo per AI index on top of an Ahrefs plan (from $129/mo). Promptwatch is the ChatGPT mention ledger with topics and tags you attach to prompts you typed. Essential is $95/mo. Explore is free (ChatGPT, 10 prompts). Site: promptwatch.com.
What Brand Radar topics actually are
Brand Radar measures brand mentions and citations across ChatGPT, Perplexity, Gemini, Copilot, and Google's AI Overviews and AI Mode. The useful part is the pivot: a cited URL is one click from backlinks and content gaps in vanilla Ahrefs. Category-level "is this topic showing up in AI" questions fit that index. If the question is research-shaped, the index is a good fit.
The method is not a prompt you typed. Chatbot indexes refresh on roughly monthly snapshots. An independent January 2026 test found Brand Radar reporting 3 ChatGPT brand mentions where manual checking verified 123. Treat topic-level ChatGPT counts as a floor. A floor is useful for direction, and it is the wrong object for a QBR that has to match what a buyer can reproduce by asking the model themselves. A modeled count that undercounts by a factor of forty is not a number you can hand to a steering committee and defend.
Custom prompt checks add $50/mo per 2,500 checks if you want your own prompts tracked. Native indexes skip Claude. A full stack often lands between $828 and $1,148+/mo. The full-stack price is the real cost of using Brand Radar as a Claude-inclusive mention ledger, and it is a different order of magnitude from a single Promptwatch plan. The skip on Claude is the part that bites, because Claude is one of the engines a brand team already pastes into.
Promptwatch topics are labels on your list
Promptwatch topics and tags group the prompts you defined (personas, country, a campaign). That is operational taxonomy, not an Ahrefs topic cluster. Paid plans check ChatGPT daily from the real UI, plus Gemini, Claude, Perplexity, and Overviews. The taxonomy is yours, and the checks are daily, which is what turns a list of prompts into a ledger you can defend.
Citation analytics at the page, domain, Reddit, and YouTube level sit behind each mention. Visitor analytics through a script or GTM attach sessions. Essential includes those views but has no listed crawler-log allowance. Agent Analytics logs ChatGPTBot on Professional, Business, and the self-serve agency plans. Allow OAI-SearchBot. The crawler-log layer is what turns a mention into a diagnosis, because it tells you whether the miss was a content gap or a blocked fetch. A mention ledger without logs answers "were we named." The same ledger with logs answers "were we named, and if not, was it because the page was not written or because the page was not read." Those are different remediations, and the logs are what split them.
Keep Brand Radar if you already pay for Ahrefs and want quarterly topic research at index scale. Do not paste its ChatGPT mention count into a QBR as the official number. Method: how we rank. Tools list.
Professional is $245/mo. Agency Kick-off is $199/mo. 4.7/5 on G2, 1,840+ brands.
FAQ
Can Brand Radar replace a ChatGPT mention tracker?
No. Index topics are research. Prompt mentions are operations. The January 2026 undercount is why.
Does Promptwatch have "topics" like Ahrefs?
It has topics and tags on your prompt list. It does not have a 260-million-prompt keyword model. Different object, similar word.
Should I cancel Ahrefs?
No. Keep Ahrefs for links and content gaps. Add Promptwatch for ChatGPT mentions. Keep Brand Radar only if someone will open it. A paid add-on that nobody opens is a line item, not a tool, and the undercount is the reason it cannot stand in for the ledger.
What to do this week
- Write 15 ChatGPT prompts from sales, not from an Ahrefs topic export.
- Load them into Promptwatch and tag them by campaign.
- If Brand Radar is on the invoice, pull the ChatGPT topic view and label it modeled.
- Compare one prompt's mention count by hand vs Brand Radar.
- Use Ahrefs on the cited URL that beat you. Use Promptwatch on the prompt that lost.