There are two places you can stand to measure how AI assistants treat a brand or a publisher. You can stand where the user stands, send the assistant a question and record what comes back. Or you can stand where the content lives, on your own server or CDN, and record every time an AI system arrives to read a page. Almost every AI visibility product on the market in 2026 does the first. A growing minority now also does the second. Knowing which one a tool is doing, and how, matters more than any feature list, because it decides what the number on the dashboard can and cannot tell you.
What is an AI visibility tracking tool?
An AI visibility tracking tool reports how often, how prominently and how accurately AI assistants mention or cite a brand or a website when people ask questions in its category. The core metrics are mention rate, citation rate and share of voice against competitors, usually with a sentiment or accuracy layer on top. Gartner now treats this as a distinct software category, which it calls answer engine visibility tools, and in its March 2026 Market Guide described them as "a baseline martech necessity". Gartner puts the market at $481 million in 2024, rising to a projected $729 million by 2031.
The category exists because ChatGPT, Google AI Overviews and AI Mode, Gemini, Perplexity, Microsoft Copilot and Claude return a composed answer rather than a ranked list of links. A rank tracker has nothing to rank. Visibility tools were built to look inside the answer instead.
How do AI visibility tools collect their data?
This is the question to ask first, and since August 2026 there is an industry vocabulary for it. The IAB's Measuring Visibility in the AI Era guidelines require providers to disclose which of four data collection architectures they use:
- Active query simulation. The provider writes or generates prompts, sends them to the assistants on a schedule and captures the responses. This is how most tools work.
- Passive behavioural observation. Real user activity is observed through an opted-in panel. The framework asks providers to disclose panel size, recruitment and representativeness, and tells buyers to treat undisclosed panel composition as a material gap.
- Platform-native data. First-party reporting from the AI platform itself, such as Microsoft reporting which of your pages its answers cited.
- Hybrid. Some combination of the above, which the framework says must not be blended or presented as equivalent without disclosure.
Two further disclosures change what a number means. Providers must state whether prompts were synthetic, derived from search keyword data or drawn from observed user behaviour, and whether the assistant had live web retrieval switched on when it answered. The same brand measured with retrieval on and off returns different results, because one reading reflects what the model remembers and the other reflects what it just read. Our guide to measuring ads served to AI agents covers the framework's directional and decision-grade thresholds in detail; the short version is that most of what the market sells today is directional.
The main AI visibility tools in 2026, compared
The field still sorts into tiers, but the tiers have moved since the start of the year.
Enterprise platforms: Profound
Profound is the category leader by capital and customer count. On 15 September 2026 it announced a $180 million Series D at a $1.8 billion valuation, jointly led by Sequoia Capital and Kleiner Perkins, seven months after a $96 million Series C at $1 billion. Its method is a hybrid. According to SiliconANGLE's report of the round, Profound sources prompt data from consumer panels whose members opt in to share their AI queries, then tracks how often a brand is mentioned in responses to those prompts, with accuracy and sentiment checks. It has spent the year moving from measurement into execution, with an autonomous planning feature called Aim, content-writing Agents and an Ads Studio for campaigns on AI platforms. Forbes told Digiday it uses Profound to measure its own AI visibility across thousands of prompts. Pricing is weighted towards enterprise contracts.
SEO suite modules: Semrush and Ahrefs
Adobe completed its $1.9 billion acquisition of Semrush on 28 April 2026, and the combined offer is now sold two ways: the Semrush AI Visibility Toolkit as an add-on for existing Semrush users, and Adobe Brand Visibility, which pairs Semrush data with Adobe agents that act on the findings. Ahrefs Brand Radar takes the opposite approach to Profound's panels. Ahrefs describes a database of more than 400 million search-backed prompts across Google AI Overviews, AI Mode, ChatGPT, Perplexity, Gemini, Copilot and Grok, modelled on real keyword data, and in April 2026 began distinguishing pages a model found from pages it actually cited. The strength of both suites is convenience: AI share of voice sits beside keyword and backlink data a team already pays for. The weakness is that prompts derived from search behaviour are a proxy for how people phrase questions to an assistant, not a record of it.
Specialists: Otterly and Peec
The specialists are built for smaller teams that want monitoring without an enterprise contract. Otterly's published pricing in September 2026 starts at $29 a month for 15 tracked prompts across ChatGPT, Google AI Overviews, Perplexity and Copilot, rising to $189 for 100 prompts and $489 for 400, with Gemini, AI Mode and Claude sold as add-ons. Berlin-based Peec AI raised an 18 million euro Series A in November 2025 and has become one of the fastest-growing names in the segment, selling daily prompt tracking on self-serve monthly plans. Both answer "am I showing up?" well. Neither is designed to be decision-grade on its own.
Free, platform-native: Microsoft Clarity
The most significant change for budget holders is that the platforms have started giving this away. Microsoft Clarity's Citations report reached general availability on 13 May 2026, showing grounding queries, cited pages and share of authority. On 9 July it added Topic Insights, which groups prompts into themes and shows where competitors win citations, free to every Clarity user. Microsoft is candid about the limits: the feature is in beta, capped at ten reports per project per week, and positioned for directional monitoring rather than guaranteed accuracy. It is also the only one of these tools drawing on the platform's own grounding data, which is exactly the platform-native architecture the IAB describes.
Why do different AI visibility tools give different answers?
Because they are measuring different things in different ways, and the thing being measured will not hold still. Paul Dyer of /prompt told Digiday in May that giving three tools the same prompts produces three different answers. Heather Physioc of VML noted that most tools return point-in-time results rather than continuous measurement. Semrush's own research, reported by PPC Land, found that citation sets across the major assistants change by roughly half month on month, with only 11 per cent overlap between platforms.
The publishers who now sell AI visibility to advertisers are blunt about it. "They all have different methodologies. Their numbers are all different," Time's chief operating officer Mark Howard told Digiday in July, comparing the market to ad viewability a decade ago, before verification vendors were forced onto a common standard. The practical rule has not changed: pick one tool, hold its prompt set and settings constant, read the trend rather than the absolute figure, and never compare one vendor's score with another's.
The second category: tools that watch the request, not the answer
The June version of this guide said no visibility tool could see an AI system actually reading your page. That is no longer true, and the change is the most important development in the category this year.
A second group of products now observes the request side. Otterly's Agent Analytics, included from its Standard tier, runs as an integration at the server, edge or CDN and classifies incoming requests from named agents such as OAI-SearchBot, ChatGPT-User and PerplexityBot, alongside the AI-referred humans who follow. Microsoft Clarity added Bot Activity tracking in January 2026 and, in June, flagged AI crawlers ignoring robots.txt. TollBit's network reports bot activity across thousands of publisher sites.
Publishers are already using this data commercially. Howard told Digiday that Time uses bot traffic as its main proxy metric in GEO conversations with clients, citing its position in the 98th percentile for AI bot activity among the nearly 7,000 publisher sites in TollBit's network. The logic is sound: an answer-side tool can tell you that you were cited, but only the request log can tell you how often your content was actually read to produce answers, including answers that never named you.
Note what this means for the IAB framework. Its four architectures all measure the answer. A publisher's own request log is not one of them, which is why the most defensible first-party signal a publisher holds sits outside the vocabulary buyers are now being taught to ask for.
What neither category does
Answer-side tools measure the output. Request-side tools measure the input. Both are reporting layers. Neither changes what happens at the moment an AI agent reads the page, and neither attaches any value to that moment. A publisher can now see, in detail, that an assistant fetched a buying guide 300 times this week to answer shopping questions, and still earn nothing from any of those reads.
That gap is where blankspace operates. It detects Live Search Agent requests at the CDN edge and serves a paid, contextual brand fact into the content the agent retrieves, so the read that visibility tools count becomes a transaction on the publisher's own domain. It is not a visibility tool and does not replace one. A brand still needs an answer-side tool to know whether it is winning the answer, and a publisher still benefits from request-side analytics to know what is being read. Monetising the read is a separate job.
How should you choose an AI visibility tool?
Start with the question you need answered, then work back to the architecture.
- A brand defending category share needs depth, auditability and a consistent prompt set. An enterprise platform or a suite module, held to the IAB disclosures, is the right shape. Ask whether the vendor's numbers are directional or decision-grade before they reach a board slide.
- A growth team running its first GEO tests can start with Microsoft Clarity for free, then add a specialist monitor if it needs coverage beyond Microsoft's grounding data.
- A publisher should treat the request log as the primary signal and answer-side citation data as the supporting one. It is the only dataset in this whole category the publisher owns outright, and it is the one advertisers cannot get anywhere else.
Frequently asked questions
What is the best AI visibility tracking tool?
There is no single best tool, only the best fit. Profound is the most heavily funded enterprise platform and combines panel-sourced prompts with monitoring and content agents. Semrush, now owned by Adobe, and Ahrefs Brand Radar suit teams already using those suites. Otterly and Peec are the accessible specialists. Microsoft Clarity offers citation reporting free, within the limits of Microsoft's own grounding data. Choose by collection method and budget, not by brand name.
How much do AI visibility tools cost?
Prices range from free to enterprise contracts. Microsoft Clarity's Citations and Topic Insights cost nothing. Otterly's published plans start at $29 a month for 15 prompts and reach $489 for 400. SEO suite modules are sold as add-ons to existing subscriptions, typically in the low hundreds of dollars a month. Enterprise platforms such as Profound are weighted towards custom contracts. The number of prompts and engines tracked drives the price more than anything else.
Can AI visibility tools see when an AI bot reads my site?
Some now can. Most visibility tools only prompt the assistants and read the answers, so they never see the fetch. Request-side products such as Otterly's Agent Analytics and Microsoft Clarity's Bot Activity report sit at the server, edge or analytics layer and log named AI agents requesting your pages. Check which kind of data a tool provides before buying, because answer-side and request-side figures measure different events.
Why do AI visibility tools report different numbers for the same brand?
They sample different prompts, on different engines, at different times, with live retrieval switched on or off, and assistants change their sources constantly. Semrush research found only 11 per cent citation overlap between major platforms and roughly half of citations changing month to month. The IAB's August 2026 framework asks vendors to disclose these choices so buyers can see why figures diverge. Compare trends within one tool, never scores across tools.
What is the IAB framework for AI visibility measurement?
Measuring Visibility in the AI Era, published by the IAB in August 2026, is a set of guidelines defining shared metrics for brand and publisher visibility in AI answers and the disclosures a measurement provider should make. It names four data collection architectures, requires vendors to state whether live web retrieval was enabled, and separates directional data from decision-grade data. It does not rate or recommend any vendor.