How to Measure AI Visibility: IAB’s New Guide for Agencies

Clients are asking a new question: “Do we show up in ChatGPT, Gemini or Perplexity, and is it getting better?” A growing number of tools promise to answer it, but they measure in different ways and can give different answers for the same brand. In August 2026, IAB published a framework to bring some order to this. Here is what it says, and how agencies can use it to report on AI visibility honestly.

What IAB published

On August 3, 2026, IAB released Measuring Visibility in the AI Era, which it describes as a standardised set of measurement guidelines for tracking brand and publisher visibility in AI-powered discovery platforms. In its announcement, IAB says more than 20 companies now sell AI visibility measurement tools, each with different methods that can produce different answers. The framework provides shared vocabulary, quality criteria and disclosure requirements, without recommending specific tools or ranking providers.

The 4 P’s of AI visibility

The framework organises metrics into a hierarchy that IAB calls the 4 P’s:

• Presence: does the brand or publisher appear in an AI response? Metrics include Mention Rate, Citation Rate, Share of Voice and Visibility Momentum.
• Prominence: where and how prominently does it appear? This covers placement and ranking order, and for publishers how substantively their content is used versus superficially cited.
• Portrayal: in what context, and how accurately? Metrics include Sentiment, Framing, Hallucination Rate and Factual Inaccuracy Rate. IAB notes this adds a brand safety dimension unique to AI.
• Persuasion: does visibility drive action? Metrics include Recommendation Strength and Post-Citation Click-Through Rate, linking to an attribution framework IAB says is forthcoming.

The order matters. Being mentioned is only the first step. Being shown prominently, described accurately and recommended is what actually drives business.

Directional vs decision-grade data

The most useful part for agencies may be the two-tier quality standard:

• Directional measurement spots patterns and trends and supports early signal detection and competitive awareness. IAB says it is not sufficient for budget allocation or executive strategy decisions.
• Decision-grade measurement meets a higher bar for query volume, sample size, prompt type coverage, testing cadence, reproducibility and platform coverage. IAB calls it the required standard before making any budget or strategy decisions.

The framework also sets out disclosures that measurement providers should make, covering things like prompt libraries, how queries are sourced, collection methods, validation and how factual accuracy is classified.

Why this matters for agencies

It gives you language to set expectations

AI answers vary from one run to the next. A screenshot of one ChatGPT response is not a measurement. The directional vs decision-grade split gives you a clear, neutral way to explain to clients what a report can and can’t support.

It helps you choose tools

Instead of comparing dashboards, you can ask vendors the questions the framework raises. Which prompts do you use, and where do they come from? How many times is each prompt run? Which platforms and regions are covered? How do you judge accuracy? Can results be reproduced?

Portrayal is a real risk

AI systems sometimes state wrong prices, outdated services or features a business doesn’t offer. Tracking hallucination and factual inaccuracy rates turns that from an anecdote into something you can monitor and fix, usually by improving the clarity and consistency of the client’s own content and third-party mentions.

It connects to PR and content work

Citation Rate and Share of Voice depend heavily on where a brand is mentioned across the web. Digital PR, credible guest content and clear, well-structured site pages are some of the levers that can influence presence and portrayal. Still, no one can guarantee how an AI model will respond.

A simple AI visibility report structure

• Scope: platforms tested, regions, dates and number of runs per prompt.
• Prompt set: a fixed list of category, comparison and brand questions that real customers would ask, kept stable over time.
• Presence and prominence: mention and citation rates against two or three competitors.
• Portrayal: any inaccurate or negative statements, with examples and the likely source.
• Persuasion: referral traffic from AI platforms in analytics, where it can be measured.
• Quality label: clearly mark the report as directional or decision-grade.

What to do this week

• Read IAB’s framework summary and share the 4 P’s with your team.
• Draft a fixed prompt set of 20–30 questions for one client.
• Run them across the AI platforms that matter to that client and record mentions, citations and inaccuracies.
• Label the results as directional, and don’t move budgets based on them alone.
• Ask any AI visibility vendor you use to explain its method in terms of the IAB disclosures.

The bottom line

IAB’s framework doesn’t tell you which tool to buy, but it gives agencies a shared language and a quality bar for a measurement category that has been mostly guesswork. Use the 4 P’s to structure reports, be honest about whether data is directional or decision-grade, and focus on accuracy as much as visibility.

Want to understand how your brand shows up in AI answers, and what you can realistically do about it? Book a free strategy call and we’ll walk you through it.