New KPIs for GEO, AEO, and AI SEO
How to measure visibility, demand capture, and revenue when AI answers sit between the user and your site.
By Bradi Slovak April 2, 2026 9 min read
Classic clicks and rankings no longer tell the full story. When AI answers sit between the user and your site, you need a new measurement stack built around answer visibility, demand capture, brand accuracy, and execution speed.
AI Overviews cut CTR on informational queries. At the same time, referrals from AI assistants are rising and they convert differently. The shift is from SEO rank → clicks to AI answer visibility → qualified actions → durable demand.
What Is Changing in Search Measurement
AI features are now folded into core reporting. Google includes AI feature traffic in Search Console Web totals, which means the old clean segmentation between organic and AI is gone.
Answers are showing earlier in the funnel. AI Overviews and AI Mode handle longer questions and follow-ups that used to drive clicks to informational pages. Clicks can fall while visibility actually rises — according to Google's Search Central documentation, CTR declines are a documented pattern when AI Overviews appear on informational queries.
AI assistants have become an acquisition channel in their own right. GA4 now recommends building a custom channel group specifically for AI assistant traffic. And with AI Mode, impressions, clicks, and position all roll into Web totals, while follow-up questions count as new queries.
The measurement environment is genuinely different, and legacy dashboards aren't built for it.
Why the Old Playbook No Longer Works
Legacy dashboards overweight rankings, sessions, and CTR. In AI-mediated results, you can occupy space inside an answer and still lose the click. CTR-only reporting marks that as failure, even when your brand is steering the decision.
Track four things instead:
- Presence in answers (GEO/AEO visibility)
- Quality of the clicks you still earn (conversion economics)
- Accuracy and brand control (risk and trust)
- Operating tempo (how fast you learn and ship)
Each maps to a category in the KPI stack below.
Category 1: Answer Visibility KPIs (GEO + AEO)
Do you show up in AI answers, and how visible are you when you do? These six KPIs measure your presence, prominence, and coverage across AI-generated results.
AI Answer Presence Rate
The percentage of a tracked query set where your brand or domain appears in an AI answer, either as a mention or a citation.
How to measure: Track 50–200 buyer questions. Sample weekly across the engines your buyers actually use — Google AI features, major assistants. Log present/not present, citation Y/N, and landing page.
Why it matters: This is your share of shelf in answer-first discovery.
Source Citation Share
When citations or links appear, what share of all cited sources belongs to you.
How to measure: Count citations per answer on the same sample set.
Why it matters: In answer-first results, citations act like top-of-page real estate.
Source Prominence Score
A weighted measure of how visible your source is inside an answer — because "present" is binary, and prominence predicts impact.
How to measure: Use a 1–5 score (top, mid, bottom, behind expansion). For more rigor, borrow from published GEO research (Aggarwal et al., 2023): use word count tied to the citation and position-weighted word count, where earlier citations count more.
Answer Coverage by Funnel Stage
The percentage of priority questions that have a clearly mapped "best page" for the intent.
How to measure: Build a question-to-URL map across awareness, consideration, and decision. Assistants reward clear intent match; catch-all pages consistently underperform.
Entity Association Score
Whether AI systems consistently pair your brand with the right entities — category terms, problems solved, industries, integrations, and standards.
How to measure: In your sampling, log the top entities mentioned alongside your brand. Compare to your target entity map.
Why it matters: Entity associations set your category position in AI-generated answers.
Citation Target Quality
The percentage of citations pointing to convertible pages (product, solution, demo, pricing, integration, or customer proof) versus blog content.
Why it matters: Visibility that routes to low-intent pages leaks revenue.
Category 2: Demand Capture KPIs (AI SEO + Classic SEO)
Does visibility turn into qualified action — leads, pipeline, revenue? These six KPIs connect AI presence to business outcomes.
AI Referral Sessions
Sessions arriving from AI assistants and AI referral sources.
How to measure: In GA4, use the Traffic acquisition report with session-scoped source dimensions. Build a custom channel group for AI assistants, starting with the source domains you already see (chatgpt.com, perplexity.ai, copilot.microsoft.com, gemini.google.com) and expanding from your referral reports.
Why it matters: Across Phasewheel client accounts, AI referral sessions have grown consistently quarter-over-quarter since Q3 2024, and they arrive with meaningfully different intent signals than organic search sessions.
AI Referral Conversion Rate (by intent)
Conversion rate for AI-referred sessions, segmented by landing page intent.
How to measure: In GA4, compare key events and conversion rates for the AI channel versus organic search versus direct.
Why it matters: Across Phasewheel client accounts, AI-referred sessions that land on decision-stage pages (solution, pricing, demo) convert at a higher rate than the same page's organic search traffic — likely because AI assistants pre-qualify intent before the click.
Qualified Visit Rate
The percentage of sessions that reach a meaningful qualification threshold. Pick one that fits your business:
- 2+ key page views (solution + proof)
- 60+ seconds of engaged time
- Demo or pricing page view
- Form start combined with scroll depth
Why it matters: When CTR drops, raw session counts mislead. Quality becomes the real lever.
Search Console Branded vs. Non-Branded Growth
Track demand creation separately from demand capture by segmenting branded versus non-branded queries in Search Console.
Why it matters: Non-branded growth drives discovery. Branded growth compounds existing demand. Conflating them hides what's actually working.
AI Feature Exposure Trend (Proxy)
An approximation of how often priority queries trigger AI features, compared against performance shifts over the same period.
How to measure: Use a SERP feature tracker or a steady manual sample. Compare GSC clicks and CTR shifts for AI-feature-likely cohorts versus unlikely ones.
Why it matters: Google blends AI feature traffic into Web totals, which limits native segmentation. This proxy fills the gap.
Pipeline per 1,000 Impressions
Revenue impact normalized to impression-heavy, click-light SERPs.
How to measure: Join GSC impressions to CRM pipeline attribution by landing page group.
Why it matters: This re-anchors SEO reporting to actual business output, not vanity metrics.
Category 3: Brand Accuracy and Trust KPIs (Risk)
Do AI answers stay correct, compliant, and on-message? These three KPIs protect the brand from what AI gets wrong.
AI Answer Accuracy Rate
The percentage of sampled answers that describe your offering correctly across features, limits, pricing, and compliance.
How to measure: Weekly sample of priority queries; score each as accurate, partially accurate, or inaccurate.
Why it matters: Errors create sales friction and compliance exposure. AI summaries compress nuance in ways that can introduce both.
Citation Consistency
The percentage of answers that cite the same canonical pages for the same question type.
Why it matters: Consistency reduces drift and makes updates easier to control at the source.
Policy/Compliance Flag Rate
The percentage of answers that surface risky claims — regulated statements, guarantees, security assertions, or competitive claims.
Why it matters: AI summaries strip context. Governance needs a way to measure what's slipping through.
Category 4: Execution Speed KPIs (Team Throughput)
How fast do you ship improvements and learn from results? These three KPIs keep the team honest about pace.
Time-to-Update for Priority Pages
Median days from insight to shipped page update, covering content, markup, and internal links.
Why it matters: AI surfaces move fast. Slow release cycles lose ground.
Content Refresh Yield
Visibility gain per refreshed page.
Why it matters: Ties team effort directly to measurable outcomes.
Instrumentation Coverage
The percentage of priority pages that have clear conversion events, structured data where useful, clean metadata, and internal link hubs.
Why it matters: Gaps in measurement create false narratives about what is and isn't working.
How to Instrument This in 2–3 Weeks
Step 1: Build an "AI Assistants" channel group in GA4. Use Google's custom channel group feature to create a dedicated AI channel. Start with the source domains you already see in your referral report — chatgpt.com, perplexity.ai, copilot.microsoft.com, gemini.google.com — then expand as new sources appear. Use the Traffic acquisition report to monitor sources and outcomes by channel.
Step 2: Segment Search Console the right way. Google includes AI feature traffic in Web totals, so perfect segmentation isn't possible natively. Work with what you have: segment by branded vs. non-branded queries, query intent (informational vs. commercial), and page groups (answer pages vs. decision pages).
Step 3: Build an Answer Visibility Panel. A spreadsheet is enough to start. Use these columns:
- Query
- Engine surface (AI Overview / assistant)
- Present (Y/N)
- Citation (Y/N)
- Prominence score (1–5)
- Target URL
- Notes (entities, errors, missing nuance)
Run it weekly for 12 weeks. Trends show up fast, and you'll have enough data to act on by week six.
What CMOs Should Do Right Now
On revenue: Expect CTR noise and plan for it. Judge search performance by answer visibility, pipeline per 1,000 impressions, and qualified visit rate — not raw clicks. Pair each priority page with a target question set and report page-level pipeline per impression monthly.
On speed: Stand up the AI Assistants channel in GA4 this week. Route these visitors to short, intent-matched paths — product or solution page, then proof, then CTA. Set a 14-day release cadence for page changes and track time-to-update and refresh yield against it.
On risk: Run a weekly accuracy audit on 50 queries. Track accuracy rate, citation consistency, and policy flag rate in a shared doc. Assign fixes to owners within seven days and re-sample the same queries to confirm the correction landed.
Your 90-Day KPI Pilot
This is the recommended starting point — scoped so you can run it without a new tool budget.
- Pick 50 priority queries that span the full funnel.
- Stand up the GA4 AI Assistants custom channel group.
- Build the weekly Answer Visibility Panel and score prominence. Use GEO-style word count and position weighting when you need more rigor.
- Report monthly on: Answer Presence Rate, Citation Share, AI referral conversion rate, Pipeline per 1,000 impressions, and Accuracy rate.
Start the pilot on three cornerstone pages. Ship updates every two weeks. Review results at day 45 and day 90, then scale to the next 20 pages.
Frequently Asked Questions
What is the single most important KPI to start with for GEO/AEO measurement?
AI Answer Presence Rate. It tells you whether your brand is showing up at all in AI-generated answers for the questions your buyers are asking. Everything else — citation share, prominence, and conversion — builds from knowing you have a presence to optimize.
How do I track AI referral traffic in GA4?
Create a custom channel group in GA4 under Admin > Data Display > Channel Groups. Add a rule that matches session source to AI domains you're already seeing in your referral report (chatgpt.com, perplexity.ai, copilot.microsoft.com, gemini.google.com, etc.). Then monitor it in the Traffic acquisition report alongside organic search and direct.
Why is CTR dropping even when my rankings haven't changed?
AI Overviews and AI Mode answer questions directly on the SERP, which means users get what they need without clicking. This is especially common on informational queries. CTR dropping while impressions hold steady is a signal that an AI feature is intercepting traffic, not that your content is underperforming.
How many queries should I track in my Answer Visibility Panel?
Start with 50 to 100 queries, prioritized by buyer intent and funnel stage. That's enough to surface meaningful patterns within six to eight weeks. Expand to 200 once you have your scoring process dialed in.
What counts as a "convertible" page for Citation Target Quality?
Any page where a visitor can take a meaningful next step toward becoming a customer: product pages, solution pages, pricing, demo requests, case studies, and integration pages. Blog posts and general educational content are not convertible in the same way, even if they drive awareness.
How often should I run an AI accuracy audit?
Weekly, on a consistent set of 50 priority queries. Consistency matters more than frequency. Running the same queries on the same cadence lets you catch changes quickly and confirm that fixes actually stuck.
