Measuring AI Search: the 2026 KPIs from the Latest SEJ Webinar
Summarize this article with AI
The missing-click trap
A client calls me on a Tuesday morning. He runs an online shop, 800 SKUs, 4 years old. Since September, his organic traffic dropped 22%. Revenue didn’t move an inch.
He’s lost. I tell him: « It’s not traffic that fell. It’s the entry point that changed. »
When I examine his citations in ChatGPT, Perplexity, and Gemini responses, I count 3,200 mentions in one month. Zero in GA4. Zero in Search Console.
The trap is right there. Users don’t need to click anymore. They get their answer straight from the AI interface. Your brand shows up, your recommendation gets cited, your influence works—but the click never happens.
The Search Engine Journal webinar confirms it. Heather Campbell explains: your brand can appear in 1,000 AI responses and your analytics tool sees nothing. Nothing.
I see this with 12 out of 15 clients I audit each week. Traffic drops. Conversions hold. That’s a symptom of strong AI presence.
My client asked me: « But how do I prove this to my boss? » I told him: « Stop talking about sessions. Show citation count. Connect it to sales. »
He tracked 47 commercial queries over 3 months. For 32 of them, his brand was cited in AI Overviews or ChatGPT. On those 32 queries, revenue climbed 19% over the period. No correlation with organic traffic.
We’re no longer in an attribution funnel. We’re in an influence ecosystem.
Invisible signals: what GA4 will never see
Classic metrics were born in a world of blue links and clicks. That world is disappearing. Today, a user types a question into ChatGPT. Reads. Doesn’t click. Acts.
Which signals must you capture?
The SEJ webinar names three: citation rate, share of voice in AI responses, and brand mention frequency by tool.
Citation rate: out of 100 queries tested, how often does your brand appear? My e-commerce client was at 4% a year ago. Today, 34%. +750%. Not because he bought links. Because he built solid semantic clusters—citable by AI.
Share of voice is your presence against competitors. If ChatGPT recommends three brands on a query, how often is yours in that trio?
Mention frequency is raw volume. How many times is your brand cited in a month? On which tool? In what context?
These three signals aren’t technical metrics. They’re influence indicators. They’re the new levers you can pull.
The mistake would be treating them as marketing KPIs. They don’t replace conversions. They precede them. They build trust upstream.
I don’t measure citations to look good. I measure appearance frequency to estimate the volume of influenced decisions. It’s a proxy.
The webinar drives home: without these signals, you’re blind on the real funnel. You don’t know why your pipeline stays full while Google Analytics cries.
And you don’t need 50 tools. Start with 10 to 20 key queries, a manual tracking sheet, or a monitoring tool like Brandwatch or Profound. What matters is consistency.
I recommend weekly tracking. AI moves fast. A citation can appear and vanish in 5 days.
You’re capturing momentum. Not a snapshot.
Measuring influence: the two methods that actually work
Knowing you’re cited isn’t enough. You must translate that visibility into commercial impact. The SEJ webinar proposes two complementary methods: incrementality and media mix modeling.
Incrementality compares two groups. One exposed to AI citations (e.g., users who saw your brand in AI Overviews during a given period). One not exposed. You measure the conversion gap.
I tested this with a SaaS client. We isolated a specific urban segment where AI Overviews was live on mobile. Conversions there were +7% higher. No other variable changed. Proof that AI influence exists, even without clicks.
Media mix modeling goes deeper. It adds AI citations as a channel in a statistical model that links marketing spend to revenue. You feed in ad spend, SEO, social—then add an « AI citations » variable. The model quantifies each lever’s contribution.
According to Heather Campbell, this combo is « the only way to land a defensible number in a budget meeting. »
Concretely, for an e-commerce site, I built a simple model: monthly citations, average order value, site conversion rate. Linking this data across 18 months, we found that every 500 additional citations drove roughly +3.2% incremental revenue. Order of magnitude.
It’s not perfect. But it beats denying the channel exists.
The trap is waiting for magic tech. It doesn’t exist. Incrementality and MMM take work. But it’s the only path to measurable AI ROI.
And I see it: clients who adopt these methods get budgets others are denied. They speak the language of results.
The monitoring layer: where to start tomorrow
Everyone talks about tools. I’m not a software seller. I build systems that run. What I deploy with clients rests on three tiers.
Tier 1: citation tracking. Define 20 queries that matter to your business. Not the most volume-heavy. The most commercial. Consideration, comparison, recommendation queries.
For each query, test it once a week on ChatGPT, Perplexity, Gemini, Google AI Overviews. Note: does my brand appear? Yes/No. What type of mention (direct citation, recommendation, plain mention)? What tone?
Don’t chase completeness. Chase consistency. In 4 weeks, you’ll have an actionable citation rate.
Tier 2: linking to business metrics. Once you have your citation series, cross it with weekly conversion data. Look for patterns. A spike in citations precedes a spike in branded traffic or direct queries? That’s cause and effect.
Tier 3: scale. When the link is proven, you move to a formalized incrementality test, then simplified MMM. No data scientist needed upfront. A solid analyst and Excel suffice.
What I see: too many teams want to start at Tier 3. They drown. Start at Tier 1. The tracking sheet is your foundation.
A B2B client did this for 3 months. She tracked 15 queries. She discovered her brand was always recommended on Perplexity for software-choice questions, never on ChatGPT. Armed with that, she refocused content optimization for ChatGPT. Two months later, she was cited on 9 of 15 queries. Her leads jumped 22%.
No magic. Just method.
From traffic to revenue: reconnecting the threads
When you look at raw numbers, some of my clients’ organic traffic dropped 15–25%. A short view says: « SEO is dead. » A finer look shows something else.
Users still arriving via the site are more mature. They’ve already been exposed to your brand in an AI response. Their purchase intent is stronger. Conversion rates climb.
I saw an online course site lose 18% of sessions in 6 months. Yet revenue stayed flat. Why? Conversion rate went from 2.1% to 2.9%. Average order values grew.
AI filtered out the curious. It let the buyers through.
That’s the real link between AI visibility and revenue: residual traffic quality improves. Not accident. Natural filter.
To reconnect the threads, you must shift the internal narrative. Don’t say « SEO brings X sessions. » Say « AI visibility contributed to X conversions at 30% lower acquisition cost. » Back it with incrementality tests.
A finance client of mine presented his latest budget using AI citations as a variable in MMM. He proved 14% revenue contribution, even though the channel didn’t exist in the original media plan. Budget approved in 20 minutes.
You’re no longer selling SEO. You’re selling measurable influence.
Now I build monthly reporting on three columns: AI citations, residual organic traffic, conversions. With an arrow connecting them. The CEO sees the story.
What I observe with my clients: orders of magnitude
I don’t believe in averages. I believe in cases. Here are three I’ve seen recently.
Case 1: auto parts e-commerce. 47,000 sessions/month. Before GEO, 120 AI citations per month. After 8 months on semantic clusters and structured markup, 3,200 citations per month. Organic traffic slightly down (-5%). Revenue up +12%. Average order value: +$8.
Case 2: B2B consulting firm. 14 AI citations per month to start. Took 4 months to realize it had to publish sourced, data-driven studies that AI loves to cite. Result: 230 citations per month. 8 qualified leads directly traced back to conversations that started with « I saw your study cited by ChatGPT… »
Case 3: tourism site. No strong organic traffic, online bookings only. AI Overviews began showing their property in recommendations. Unmeasurable traffic. But direct phone calls jumped 30% in 3 months. Coincidence? No.
These numbers aren’t promises. They’re observed orders of magnitude.
What’s striking: in each case, traditional metrics told nothing. Only citation tracking revealed the lever at work.
And in each case, hard foundational work paid off. No shortcuts.
Every sector has its own high-impact AI queries. Whoever finds theirs gets 6 months ahead.
Action to take tomorrow
You don’t need a steering committee, an RFP, or six-figure budget. You need three simple moves.
1. Pick 10 commercial queries. Those that trigger purchase intent, demo requests, meeting bookings.
2. Measure your current citation rate. Test each query on ChatGPT, Perplexity, Gemini, AI Overviews. Note once per week. Build a sheet.
3. Run a simple incrementality test. Take a region, segment, or channel where you can isolate AI exposure. Compare conversions to a similar non-exposed segment. You’ll have your first defensible number.
In 30 days, you’ll have more ammunition than 90% of marketing teams.
SEO in 2026 isn’t won on rankings anymore. It’s won on citations. Those who’ve figured it out are moving ahead.
I work with clients on this launch. The live audit I offer takes an hour: we review your semantic clusters, measure your citation rate on 10 queries, and you leave with a 30-day action plan.
Don’t chase perfection. Chase proof.
How many citations did your brand receive last month?
Live Audit: your citation rate in 1 hour
I review your semantic clusters, measure your citation rate on 10 queries, you walk away with a 30-day action plan. I don’t sell you the method. I show you which pages trigger AI citations.
Book a strategic call — 45 minFrequently Asked Questions
Should I still track organic traffic in GA4?
Yes, but as a secondary indicator. Organic traffic no longer captures your full brand influence. It measures direct visits, not decisions made after an AI response. Add citation tracking to get the complete picture.
How do I measure AI citations without expensive tools?
A tracking sheet with 10–20 queries tested weekly on ChatGPT, Perplexity, and AI Overviews is enough. Note yes/no and context. Consistency beats comprehensiveness.
Is incrementality accessible to a small business?
Absolutely. Compare conversions from a geographic segment exposed to AI Overviews against a non-exposed segment, over the same period. No data scientist needed: an analyst and Excel work fine at first.
What tools do you recommend for tracking AI mentions?
I don’t sell tools, but my clients use platforms like Brandwatch, Mention, or specialized solutions like Profound. What matters is capturing citation frequency and context, not just volume.
Is traditional SEO dead?
No. Classic SEO still feeds the AI systems that crawl indexed pages. But click-driven management is outdated. You must integrate AI influence signals into reporting and budget decisions.
