How to Measure AEO and GEO: AI Search Visibility and Conversion Tracking

What should an AEO and GEO scorecard measure?

Use five layers rather than one headline number.

| Layer | Primary question | Useful evidence | What it cannot prove alone | | --- | --- | --- | --- | | Visibility | Are AI features surfacing our pages? | AI-feature impressions, cited pages, countries and devices | That anyone visited or converted | | Acquisition | Did an AI experience send a visit? | Session source, referrer, landing page and campaign parameters | Every unclicked citation or untagged visit | | Engagement | Did the visit find useful depth? | Engaged sessions, scrolls, tool starts and content paths | Commercial intent by itself | | Conversion | Did the visitor take a meaningful action? | Tool completions, forms, purchases and booked calls | Whether AI was the only influence | | Commercial outcome | Did the action create business value? | Qualified leads, opportunities, revenue and pipeline | Incremental impact without a control |

This structure prevents a common reporting failure: presenting a large visibility number next to a small lead count and implying a conversion rate when the numerator and denominator do not describe the same audience.

What does Google Search Console report for AI search?

Google launched dedicated generative-AI performance reports in June 2026. The Search report covers impressions from AI Overviews and AI Mode and documents page, country, device and date views.

Google's report documentation says the insights were rolled out to all websites worldwide on 31 August 2026. A property may still not show the report when it has not received enough impressions or has been excluded from supported Search generative-AI features.

As verified on 16 September 2026, the dedicated report exposes:

• AI-feature impressions;
• pages receiving those impressions;
• countries;
• devices; and
• date-based trends.

Google does not list dedicated click, query or conversion fields in this generative-AI report. AI activity also remains part of the standard Search Console Web performance data. Google's counting methodology says an external link click from AI Mode counts as a click, while AI Overview impressions require the link to be scrolled or expanded into view.

Do not add the dedicated AI impressions to total Web impressions: Google says the generative-AI data is already included in the overall performance reporting.

How do you track ChatGPT referral traffic in GA4?

OpenAI's current publisher guidance says ChatGPT automatically adds `utm_source=chatgpt.com` to referral URLs. That gives publishers an observable acquisition signal when a user clicks through.

In GA4, create an exploration or report using:

• Session source;
• Session source/medium;
• Landing page plus query string;
• Engaged sessions;
• Key events; and
• Revenue or lead-value fields where implemented.

Google Analytics defines traffic-source dimensions as the evidence used to identify where visitors came from and how they behaved. Its traffic-source documentation also explains how UTM parameters populate source, medium and campaign fields.

Start with an exact filter for `chatgpt.com`, then maintain a reviewed list of other AI referrers actually observed in your property. Do not assume every AI platform uses stable referrers or consistent UTM parameters. Some visits can appear as direct traffic because the browser, app or redirect chain did not preserve referral information.

How should conversion tracking work for AEO?

The website and CRM need to preserve three types of context:

1. Acquisition context: session source, medium, referrer and campaign values.
2. Content context: landing-page slug, content type, topic and conversion asset.
3. Commercial context: lead quality, booked call, opportunity stage, pipeline value and eventual revenue.

For an agency or B2B service site, track actions such as:

• article or tool CTA click;
• tool start and completion;
• result-email request;
• qualified contact-form submission;
• strategy-call click and completed booking; and
• opportunity creation in the CRM.

Store the landing-page slug and observable traffic source with the form submission. Add a short, optional "How did you hear about us?" field for leads influenced by an AI answer but arriving without a measurable referrer. Keep self-reported influence separate from click-based attribution.

HML's AEO Readiness Checker can establish a technical and content baseline before the measurement period begins.

A worked 90-day AEO scorecard

This is a hypothetical example using round figures, not HML or client performance.

An Indian growth-stage brand records the following over 90 days:

• 2,400 Google generative-AI impressions across 18 pages;
• 120 GA4 sessions with an observed ChatGPT source;
• 58 engaged ChatGPT-referred sessions;
• six qualified forms from those observable sessions;
• two completed strategy calls; and
• one CRM opportunity.

The directly calculable referral metrics are:

• 48.3% engaged-session rate: 58 divided by 120;
• 5% observable session-to-qualified-form rate: six divided by 120; and
• 33.3% qualified-form-to-call rate: two divided by six.

Do not divide the six ChatGPT-referred forms by the 2,400 Google AI impressions. Those numbers come from different platforms and do not represent a common user cohort.

The team should report Google visibility and ChatGPT-referred outcomes in adjacent sections, then add a qualitative note if CRM interviews reveal additional AI influence. This is more defensible than combining everything into an invented "AI conversion rate."

Can manual citation checks measure share of voice?

Manual prompt checks are useful for diagnosis: they can reveal how an engine describes the brand, which sources it cites and whether factual errors appear. They are weak primary KPIs because answers can change with phrasing, location, account context, model updates and time.

If you run prompt checks:

• freeze a small, intent-led prompt set;
• record date, platform, model or product surface and location;
• capture citations and brand mentions separately;
• distinguish owned-page citations from third-party mentions; and
• treat results as a repeated sample, not the full universe of answers.

Use citation checks to generate content and entity tasks. Use platform and analytics data for the performance scorecard.

Which AEO metrics belong in the weekly report?

Leading indicators

• Generative-AI impressions by page and country.
• Number of pages receiving AI-feature visibility.
• Observable AI referral sessions by source.
• Engaged-session rate and tool engagement.
• New versus returning AI-referred users.

Conversion indicators

• Qualified forms and booked calls from observable AI referrals.
• Assisted key events where the reporting model supports them.
• Self-reported AI influence from lead forms or sales discovery.
• Landing pages creating qualified outcomes.

Business indicators

• AI-referred qualified leads.
• Opportunities and pipeline value.
• Closed revenue, reported separately from influenced pipeline.
• Cost of content and measurement work.

Avoid universal AEO conversion-rate or citation-rate benchmarks. Establish HML or brand-specific baselines by intent, page type and market.

How do you establish an AEO baseline?

1. Confirm the site is crawlable and indexable. Google says ordinary SEO requirements apply to AI Overviews and AI Mode; there is no special AI schema or machine-readable file requirement.
2. Record a fixed 28- or 90-day baseline for Search Console AI impressions where the report is available.
3. Create a GA4 exploration for observed AI sources, landing pages, engagement and key events.
4. Verify every important CTA and form fires once with its content slug and source context.
5. Preserve first-touch, latest-touch and self-reported discovery evidence in the CRM.
6. Separate brand and non-brand Search Console demand to monitor possible downstream awareness.
7. Review results weekly, but make strategic decisions on a longer window unless traffic volume is high.
8. Annotate site releases, PR activity and major content changes so trends have operational context.

The search-everywhere optimisation guide explains the broader discovery landscape. The zero-click search guide covers visibility when a search does not produce a visit, while the Shopify AEO checklist applies the foundations to ecommerce product discovery.

What should your next action be?

Run the AEO Readiness Checker, then choose one high-intent topic cluster and implement the five-layer scorecard for 90 days. If analytics, CRM and Search Console cannot be reconciled, request an AI-search measurement review before setting a visibility or lead target. HML's digital marketing service connects technical visibility with content, conversion tracking and qualified pipeline.

For a practical example of organic visibility work tied to a commercial website, review HML's SEO case study for an Indian luxury home-decor brand.