October 1, 2026

AI visibility tools compared: 11 trackers, three questions that tell them apart

Every AI visibility tool says it tracks ChatGPT. That sentence tells you almost nothing. Three questions tell you a lot more, and most vendors answer them on their own website, if you know where to look.

Disclosure: we make Miraqo, one of the eleven tools below. Every fact here comes from each vendor's own pages as they read on October 1, 2026, with the sources listed at the end. Where a vendor does not say something, we write "not stated", which is not the same as "no".

Why "we track ChatGPT" is not enough

In August we asked ChatGPT the same question three times, on the same day, through three different collection methods. The three answers cited thirteen domains with only one in common. All three methods could honestly be sold as "ChatGPT tracking". So before comparing prices, it is worth knowing what each tool is actually measuring.

Question 1: which surface do you read the answer from?

A model answers differently in the public chat, where a person types, and through its API, where a developer calls it with settings of their own. If you want to know what your customers see, you want the first.

• Say they read the public interface: Miraqo, SE Ranking, Otterly.AI, Peec AI (logged out; its add-on models run through APIs) and Profound. Semrush says its answers are "captured from real requests and not via any APIs of LLMs", without saying more.
• Say they use APIs: GeoSuite ("the providers' API endpoints").
• Do not say: Ahrefs (except for Claude, tracked via API), Wellows, LLMrefs, Foglift.

Neither choice is fraudulent. APIs are cheaper and more stable; the public interface is closer to what people see. What matters is that you know which one your numbers come from, because switching method breaks the history just like switching model.

Question 2: how many times do you run the same prompt?

AI answers vary from one run to the next. A brand cited once may be missing on the next try, with nothing changed on its site. A single run is a sample, not a measurement.

• Several runs per check, on your prompts: Miraqo runs each prompt three times per engine at every check and reports share of voice as the percentage of runs that mention the brand.
• Repeated sampling on generated prompts: LLMrefs ("continuously checked to ensure statistical significance") and GeoSuite (scores that "stabilise after roughly 100–200 queries").
• One run per day or per cadence: Peec AI and Wellows say so explicitly. The others do not state it, and most track daily.

Daily single runs and fewer repeated runs are two ways of fighting the same noise: one spreads the samples over time, the other takes several at once. Over a month both add up. Over a week, a single daily run can swing for no reason, and if the swing lands on the day a client opens the report, you end up explaining a drop that never happened to someone who has just been told their brand vanished from ChatGPT.

Question 3: which engines, and can you choose the prompts?

Google's AI Overviews and AI Mode are two different products: one is a summary on top of the results page, the other is a separate conversational search. A tool that tracks "Google AI" may cover only one.

• All five (AI Overviews, AI Mode, ChatGPT, Gemini, Perplexity) on the entry plan: Miraqo, Wellows, LLMrefs, SE Ranking, Ahrefs.
• With conditions: Otterly.AI sells AI Mode and Gemini as add-ons; Peec AI lets you pick three engines per plan; Semrush covers Perplexity in a separate, weekly report; Profound offers AI Mode and Perplexity on Enterprise.
• Without AI Mode: Foglift, GeoSuite (which also skips AI Overviews).

On prompts: LLMrefs builds them from your keywords, GeoSuite generates them from your brand and market, and Profound's trial uses a fixed set. Everyone else lets you type the questions your customers actually ask.

The full table

Where Miraqo fits, including where it doesn't

Miraqo is the cheapest way in this list to cover all five engines (€9 a month, VAT excluded), and the only tool here that states it repeats the user's own prompts several times at every check. It reads the public versions of the assistants, which four other tools also do.

It is not the right pick for everyone:

• It does not track Claude or Copilot. Ahrefs, Otterly.AI, Peec AI, Profound, Foglift and LLMrefs cover at least one of them.
• The €9 plan updates every three days; daily data costs €11. Most of the others are daily by default.
• On volume it is small: about fifteen prompts on all five engines in the entry plan. If you need a hundred prompts tracked daily, SE Ranking's €109 plan gives you more per euro.

How to choose, in short

• You already pay for Ahrefs or Semrush: try their AI module first; neither offers a trial on it.
• You need many prompts, daily: SE Ranking or LLMrefs.
• You want a few important prompts measured reliably on all five engines, on a small budget: Miraqo or Wellows.
• You are an enterprise brand with a dedicated team: Profound or Peec AI.

Whatever you pick, ask the three questions before you sign. And ask a fourth one, which we did not check for this table: when the AI cited no one, does the report say zero, or "not available"? An answer written from memory, without searching the web, is not the same as an answer that searched and left you out. Our methodology explains how we handle it.

Sources

All read on October 1, 2026. Ahrefs: Brand Radar help, pricing. Semrush: AI visibility data, AI Visibility Toolkit. SE Ranking: AI visibility tracker, pricing. Otterly.AI: pricing, help. Peec AI: docs, product summary. Profound: pricing, comparison. LLMrefs: pricing. Wellows: pricing. Foglift: pricing. GeoSuite: FAQ, pricing. Miraqo: pricing, AI Visibility guide.