How to Measure AEO: Tracking AI Citations and Mentions

Start by building a prompt set, not a keyword list. Write 30-50 real questions your customers ask before buying: "best inventory app for Shopify," "how to reduce cart abandonment," "CIA vs Klaviyo pricing." These are your test queries. Run the same set every month so your data is comparable. A frozen prompt set is the single most important thing you'll build, because AI answers shift daily and you need a stable measuring stick.

Track four core metrics. Citation frequency: the percentage of your prompt set where you're named or linked (aim to move from, say, 8% to 25% over a quarter). Share of voice: your mentions divided by total brand mentions in those answers. Citation position: are you the primary source or a footnote. AI referral traffic: sessions arriving from chatgpt.com, perplexity.ai, gemini.google.com, and Google AI Overviews.

For traffic, isolate AI sources in GA4. Create a segment or exploration filtering session source to hostnames like chatgpt.com, perplexity.ai, copilot.microsoft.com, and gemini.google.com. Google AI Overviews is harder because clicks often report as normal google / organic; use Search Console's performance report and watch for high-impression, lower-CTR queries, which is a fingerprint of an answer being shown above your link.

Use tools instead of checking by hand, because manual spot-checks don't scale to 50 prompts across 5 engines. Purpose-built AEO trackers include Profound, Peec AI, Otterly.ai, and Scrunch, which run your prompts on a schedule and log citations and sentiment. Budget roughly $99-$500/month depending on prompt volume. If you're bootstrapping, a spreadsheet plus a weekly manual run of 20 prompts gets you 80% of the signal for free.

Watch sentiment and accuracy, not just presence. Being mentioned isn't enough if the AI describes your pricing wrong or frames a competitor as the better fit. Log how you're characterized in each answer. When an engine repeats a wrong claim, it's usually pulling from an outdated page, a stale directory listing, or a competitor's comparison post, so fix the source it's citing rather than arguing with the model.

Set a realistic cadence and baseline. AEO moves slower than paid ads and faster than traditional SEO: expect 6-12 weeks before content changes show up in citations, because engines re-crawl and re-index on their own clock. Capture a baseline in month one, review monthly, and tie each metric to an action, for example "we published a comparison page, did our citation frequency for that query go up."

Close the loop back to content. The pages that get cited tend to answer one question directly in the first 40-60 words, use clear headings, include specific numbers, and carry schema markup. When a prompt shows you're absent, that's a content brief: write the answer, structure it for extraction, and re-test it next month. Measurement only matters if it feeds the next thing you build.