AEO Tracking: How to Measure Your Brand's Visibility in AI Search Answers

Every AEO conversation with a marketing team eventually hits the same wall: "how do we prove this is working?" And every default answer — check GA4, look at referral traffic from chat.openai.com, count sessions — is fundamentally wrong for the channel.

There are three reasons your existing analytics stack can't measure AEO.

Citations are not clicks. A ChatGPT user asks a question, ChatGPT answers with a paragraph that quotes your data and links your site as a source. The user reads the paragraph, has their answer, and never clicks. You were cited — potentially in front of thousands of high-intent buyers — and Google Analytics recorded nothing. ChatGPT's estimated click-through rate on cited sources is about 1.3%, versus Google's ~29.2%. That 22× gap means most of your AEO value is happening invisibly.

Referral attribution is broken across AI products. Some ChatGPT clicks arrive with `chat.openai.com` in the referrer, but many arrive with no referrer at all (direct traffic), and many arrive from a proxy domain the user's browser or extension inserted. Perplexity, Claude, Gemini, and Copilot referrers similarly fragment. GA4's default channel grouping puts these across Direct, Referral, and Unassigned buckets — so even the traffic that does convert is misattributed.

Volume and value are decoupled. Even where you can measure AI referral traffic, the numbers look tiny compared to Google organic. AI referral volume remains 47-190× smaller than Google organic for most sites. That leads teams to conclude the channel doesn't matter — missing that the same traffic converts up to 31% better than non-branded organic and up to 16% vs 1.8% in the strongest documented case study.

AEO needs a measurement layer that answers a different question: are we in the room? Are AI products citing us when our prospects ask the questions we sell against? Everything else is downstream of that.

1. Citation rate (also called mention rate or answer share). Of the target prompts your prospects would run, what percentage of answers cite your site or mention your brand by name? This is the core AEO metric — the AI-search equivalent of "do we rank for our keywords?" Track it by platform (ChatGPT vs Perplexity vs AI Overviews) because platform variance is significant.

2. Share of voice (SOV) within cited sources. When your category is being discussed, what percentage of citations across all AI products go to you versus each competitor? This is your competitive positioning metric. A 5% SOV against three competitors at 15% each tells you you're being discussed but underrepresented — a different problem than being absent entirely.

3. Sentiment of mentions. Are you being cited positively ("one of the leading agencies in India"), neutrally ("agencies like Digitaso, Webchutney, and iProspect"), or negatively ("users report inconsistent delivery")? Peec AI, one of the AEO-native tools, scores sentiment on a 0-100 scale for exactly this reason. Negative mentions at high citation rate are worse than no mentions at all.

4. Referral traffic and conversion from AI sources. This is the traditional-analytics companion metric — not the primary one. Filter GA4 for the domains you can identify (`chatgpt.com`, `chat.openai.com`, `perplexity.ai`, `copilot.microsoft.com`, `gemini.google.com`) and monitor session count, engagement rate, and conversion rate versus organic search. Expect small numbers with high quality: full-year 2025 ecommerce data showed AI referral traffic converting 31% better than non-branded organic; one Seer Interactive case study found ChatGPT traffic converting at ~16% vs 1.8% for Google organic; other 2026 studies clock AI-referred visitors converting 42% better than non-AI.

The cardinal error is skipping metrics 1-3 and looking only at metric 4. Referral traffic lags citations by weeks to months. If you measure only what shows up in GA4, you'll conclude AEO isn't working long after it has quietly become half your pipeline.

Most AEO tracking tools fall into three groups: free checkers, AEO-native platforms, and SEO suites that have added AI modules. The table compares the tools we looked at, using each vendor's own website (checked 28 September 2026). Pricing and engine coverage change often, so treat this as a starting point and confirm on the vendor's site before you buy. Where we could not confirm a detail, the cell says "Check site".

Pricing verified 28 September 2026 against each vendor's own website. Plans and prices change often, so confirm on the vendor's site before you buy.

AEO-native platforms. These are built from the ground up for measuring AI search visibility: Profound, AthenaHQ, Peec AI, Otterly.AI, and a handful of newer entrants (RevvGrowth, Airank Lab, Toffu, Aiclicks). A well-designed 30-day head-to-head test of the three most-referenced platforms found AthenaHQ delivered a 45% net gain in answer share, Peec AI 8%, and Profound a 1% decline over the test period — but tool efficacy varies significantly by category, query set, and how much you commit to executing the recommended actions. Peec AI monitors ChatGPT, Perplexity, and Gemini; AthenaHQ covers 8+ LLMs including Claude and Copilot; Profound is broadest but with reported data-consistency complaints and difficulty translating insights into actions.

All-in-one SEO suites. Ahrefs Brand Radar tracks AI visibility across ChatGPT, Perplexity, Google AI Overviews, Google AI Mode, Gemini, and Copilot. Its methodology uses a static prompt library and timed snapshots (not live queries), which one review noted led to only 3 mentions reported vs 123 found through manual verification for a test brand — meaning the tool is directionally useful but likely under-counts. Semrush's AI Visibility Index runs a similar model, with the advantage of integrating into an existing Semrush subscription for teams already using their broader SEO stack.

Which to pick. For a bootstrapped brand or a first AEO investment: start with the free checkers plus manual monthly tracking. For a mid-market brand with 3-4 named competitors and a budget: Peec AI is, in our view, the strongest cost-to-value option in the category. For an enterprise brand needing multi-platform coverage and revenue attribution: AthenaHQ or Ahrefs Brand Radar. Trial at least two before committing — AthenaHQ, Otterly.AI and Profound all offer a free plan or trial, and tool efficacy varies more by industry and prompt set than any vendor's marketing suggests.

A free AEO visibility checker runs a fixed set of prompts about your brand through AI engines and tells you whether you show up. Two are worth trying today:

• Ahrefs AI Visibility Checker — no signup, no credit card, and no limit on how many brands you check. It checks ChatGPT, Gemini, Perplexity, Copilot, Google AI Overviews and AI Mode, but shows only a limited preview (the top 5 results per section).
• HubSpot AI Search Grader — free to run, and currently analyses ChatGPT (OpenAI), Perplexity and Gemini.

Use a free checker to get a first look, to sense-check your baseline, or to show a sceptical stakeholder that the question is worth answering. Its limits are built in: the prompt set is fixed, so it cannot tell you what happens on the prompts your own buyers use; each run is a single snapshot with no trend; and the detail is thin. Run it for your brand and two or three named competitors, screenshot the results with the date, and move to the manual monthly report below if the gap looks real.

You do not need a paid tool to start. A defensible monthly AEO report takes about 90 minutes and can be built with a spreadsheet.

Step 1 — Build your prompt library (one-time, 60 min). Write 25-40 prompts that mirror what a real prospect would ask an AI search product about your category, service, or geography. Split them across four types:

• Category prompts: "what is [service]", "how does [service] work", "best practices for [service]"
• Vendor-comparison prompts: "best [service] agencies in [city]", "top [service] providers for [industry]", "how to choose a [service] partner"
• Brand prompts: "reviews of [your brand]", "is [your brand] any good", "[your brand] vs [competitor]"
• Problem prompts: "my [service] isn't working, what should I do", "why is my [metric] declining", "how do I fix [problem]"

Step 2 — Baseline run (30 min, month 1 only). Run every prompt in ChatGPT with web search enabled, in Perplexity, and in Google AI Mode. Log for each: (a) was your brand mentioned? (b) was your site cited as a source? (c) which competitors were mentioned or cited? (d) what was the sentiment of any mention? This gives you a baseline citation rate (%), share of voice by competitor, and sentiment distribution.

Step 3 — Monthly re-run (30 min). Repeat the exact same prompt set every 30 days. Log the same fields. Track the deltas — citation rate up/down by prompt type, SOV vs each named competitor, any negative mentions to address. A three-column spreadsheet — Prompt | Result Month N | Result Month N+1 — is enough to show meaningful trend after 90 days.

Step 4 — Diagnose the losses. For any prompt where a competitor is cited and you aren't, click into the cited source and read it. What structural, entity, or authority signal did they have that you didn't? Feed those diagnoses back into your content and technical roadmap. This closes the loop from measurement to action — the step most tool-only workflows skip.

This manual method underweights coverage — you're testing 25-40 prompts, not the 405M+ Ahrefs Brand Radar tests — but it overweights signal quality because you're seeing the actual answers, not a scored aggregate.

Monthly is the right cadence for most brands. Weekly is overkill — AI answer shares fluctuate day to day for reasons unrelated to your work, so weekly noise hides the trend. Quarterly is too slow — you would miss a citation drop until it had cost you a quarter of pipeline.

What makes monthly tracking work is consistency:

• Keep the prompt set fixed. Re-run the exact same 25-40 prompts every 30 days. Changing prompts mid-stream makes month-to-month movement meaningless.
• Log the conditions. Record the date, the platform and the mode (for example ChatGPT with web search on, or Google AI Mode), because answers differ by mode.
• Add an event-driven checkpoint. If a major AI platform ships a model update or ranking change, run an extra check 2-3 weeks after it lands.
• Watch your top prompts mid-month if you are changing content. If you are actively restructuring pages, spot-check your 5-10 highest-intent prompts halfway through the month to see early direction. Treat it as a signal and let the monthly run decide.

Referral traffic lags citation rate by 4-8 weeks, so review citation rate first each month and use GA4 as confirmation, not the other way round.

The measurement problem doesn't stop at citations. Even when AI drives real traffic to your site, attribution is genuinely harder than it is for Google organic.

How to see ChatGPT and Perplexity referrals in GA4

Start with what GA4 can see. This takes about 15 minutes and gives you the floor of your AI-referred traffic and conversions.

Step 1 — Check what is already arriving. In GA4 open Reports → Acquisition → Traffic acquisition, switch the primary dimension to Session source / medium, and search for the AI domains: chatgpt.com, chat.openai.com, perplexity.ai, gemini.google.com, copilot.microsoft.com and claude.ai. Some ChatGPT links carry a utm_source=chatgpt.com parameter, which makes them easy to spot.

Step 2 — Group them into one channel. In Admin → Data display → Channel groups, create a custom channel group with a channel named "AI assistants". Set the condition to Source matches regex chatgpt\.com|chat\.openai\.com|perplexity|gemini\.google\.com|copilot\.microsoft\.com|claude\.ai, and drag it above Referral so the traffic is not swallowed by the default grouping.

Step 3 — Track the conversions. Make sure your lead form, booking or purchase is marked as a key event, then compare the AI assistants channel with Organic Search on sessions, engagement rate and conversion rate. Treat the result as a floor: much AI traffic still arrives as Direct.

Step 4 — Ask people. Add a "How did you hear about us?" field to your lead form with an option such as "An AI assistant (ChatGPT, Perplexity, etc.)". It catches the visits GA4 cannot attribute.

Three things break traditional attribution:

Multi-touch journeys. A prospect might read a ChatGPT citation of your blog on Monday, Google your brand on Wednesday, click the branded ad on Thursday, and convert on the LinkedIn ad on Friday. Last-touch attribution credits LinkedIn. First-touch credits ChatGPT. Neither is right. Data-driven attribution in GA4 helps, but it can only work with the data it sees — and it doesn't see the ChatGPT citation the user never clicked.

Direct-brand-search lift. The most reliable revenue signal from AEO is the indirect one: your branded search volume in Google (and Google Ads) rises weeks after AI citation rates rise. Track branded search impressions in Google Search Console monthly. If your AEO tracking shows a rising citation rate and branded search is climbing on a similar curve, that is credible correlational evidence of AEO revenue impact — even without direct-click data.

Cookieless environments. A growing share of AI traffic arrives from browser or extension proxies that strip referrers and block third-party cookies. This is worse for AI channels than for Google. Server-side tracking (GA4 Measurement Protocol, first-party tracking via a proxy subdomain) recovers some of this but not all.

The most reliable approach we've seen work in client engagements is a three-signal model: (1) direct AI referral revenue in GA4 as a floor, (2) branded search lift as a lagging correlational signal, (3) qualitative sales-team feedback on how new leads describe finding you — "I asked ChatGPT for the best agency in Mumbai" is now a real inbound response pattern. Together the three build a defensible AEO revenue story even without perfect click-level attribution.

There is no universal "good" citation rate — it depends heavily on category, geography, and site authority. But the patterns we've observed across client work and public case studies give useful anchors.

New site, competitive category (0-12 months old). Expect 0-5% citation rate in month 1 on your baseline prompt set. If the technical foundation is in place (Bing indexation, schema, answer-first content), month 3-6 typically climbs to 10-20% on category prompts and 30-50% on brand prompts. Reference: our own site went from ~0% category-prompt citations in early 2026 to measurable citation share on "answer engine optimisation" queries by month 4, ranking position ~42 in Google but appearing in ChatGPT responses for the same query.

Established site, moderate category (2-5 years old). Baseline citation rate 5-15% is typical without any AEO-specific work. With 90-120 days of focused work, published AEO case studies show sites reaching 25-40% citation rate on target prompts, with proportional lifts in AI-referred qualified leads. The HubSpot 2026 State of Marketing report noted 58% of marketers say AI-referred visitors convert at higher rates than traditional organic.

Established authority site, established category. These sites often start with 30-50% citation rate on category prompts and can push to 60-80% share of voice with content restructuring. The gains here are smaller in percentage terms but larger in absolute business impact.

Don't chase a percentage target in the abstract. Track the delta against yourself month-over-month, and against your named competitors on the specific prompts your buyers use. That's a defensible AEO story regardless of whether your citation rate is 8% or 68%.

AEO tracking for retail and ecommerce brands

Retail brands track the same four metrics, but the prompts and the sources that matter are different.

• Use shopping-style prompts. Build the prompt library around what buyers ask when they are choosing what to buy: "best [product] under [budget]", "[product] vs [competitor product]", "where to buy [product] in [city]", and "alternatives to [competitor brand]".
• Track the pages that get cited, not just whether you are. For product and comparison questions, AI answers often cite review sites, marketplace listings and comparison articles rather than a brand's own pages. Log which third-party pages are cited for your category, and whether your products appear on them.
• Measure by product category. Report citation rate and share of voice per category (for example running shoes versus apparel) instead of one blended brand number, so you can see where you win and where a competitor owns the answer.
• Tie it to purchases. In GA4, compare the AI assistants channel's purchase conversion rate and revenue per session with Organic Search, using the channel group set up above.

For the strategic overview of AEO as a discipline, see the AEO strategic overview. For the platform-specific tactics see ChatGPT SEO, and for the underlying mechanics see how LLMs pick sources.