Opening: a number that lulls you — and misleads you
In 2026, marketers keep quoting one cross-industry benchmark: per Conductor's 2026 AEO/GEO Benchmarks Report, AI referral traffic averages just 1.08% of all site traffic, growing only about 1% month over month. That number tempts SMB owners to conclude: AI traffic is tiny, ignore it. But the same report holds another figure — within that 1.08%, 87.4% comes from ChatGPT alone. That is the real point: the pool is small, but the concentration is extreme.
To see why this isn't ignorable, look at how the market restructured over 12–18 months. The entry point is sliding from a list of blue links to AI answering directly: Google AI Overviews now appear on about 25% of searches on average, with huge industry variance — Health Care at 48.7%, Financials at 25.8%. When the answer is eaten by an AI summary, users stop clicking through; your organic traffic drops not because rankings fell but because the click itself vanished. The report analyzed 13,770 enterprise domains across 10 industries and 22 sub-industries, aggregating 3.3 billion sessions — a sample big enough that 1.08% is hard to dismiss as noise.
Set the peer research side by side and it gets clearer. Muck Rack's longitudinal data shows 82–89% of AI citations come from earned media, not brand-owned pages; on citation frequency, Perplexity attributes sources nearly every time (~97%) while ChatGPT does so only ~16%. Different engines behave wildly differently, and traffic sits mostly on the one that cites least. The category's trajectory isn't an AI-traffic explosion — it's a redistribution of the entry point, highly concentrated.
Will SMBs be affected? Yes, but the opposite of how most assume. This piece covers: what the numbers mean together, three things to do today, how AI-visibility trackers compare, and a no-SaaS DIY path to track it yourself.
Event detail: three numbers you must read together
Any single number misleads; the three together are the truth:
- 1.08% (traffic share): AI referral is a small pipe today — don't cut existing SEO and content investment for it.
- 87.4% (ChatGPT concentration): if you back only one platform's visibility, put resources on being cited by ChatGPT, not spread evenly.
- 25% / 48.7% (AI Overview coverage): this is what actually eats your clicks. In high-coverage sectors like health and finance, AI Overviews hit you far harder than AI referral traffic does.
Conductor's analysis rests on ~21.9M Google searches, 17M AI-generated responses, and 100M AI citations; the most-cited page type was blog content, with over 300,000 pages cited. In other words, AI citation favors deep, structured content — not homepage marketing copy. Cross-check the Google Search Central AI features docs for the official take on what suits AI summaries.
Immediate actions for three reader types
- Brand owners / SMB founders: build an AI-referral segment in GA4 (isolate chatgpt.com, perplexity.ai, gemini.google.com), measure your real share over two weeks, then decide how much to invest.
- Marketers / SEO: add structured data (FAQPage, Article schema) to your most-citable content types (deep guides, FAQs, comparison tables) to raise inclusion odds.
- Developers / agencies: package AI-visibility infrastructure (schema, clean semantic HTML, crawlable content) as a deliverable service, not buried inside a generic SEO line.
SaaS tool comparison
| Tool | Positioning | Price (monthly) | Best for |
|---|---|---|---|
| Conductor / Profound | Enterprise AI-visibility + SEO platform | Hundreds to thousands USD | Mid/large brands with a marketing team |
| Otterly.ai | AI-search mention/citation tracking | From ~US$29 | SMBs wanting low-cost entry tracking |
| Peec AI / Scrunch | Multi-engine citation monitoring + competitor view | From ~US$100 | Brands comparing across ChatGPT/Perplexity |
Shared limit: they tell you whether you're mentioned, but rarely attribute revenue. Treat prices as indicative — check vendor sites.
What they won't tell you
- AI citation equals business is marketing spin. Being mentioned by ChatGPT doesn't mean someone bought. How much of that 1.08% converts is something most tools can't answer — track it yourself in GA4.
- Concentration cuts both ways. Betting everything on ChatGPT is efficient, but the day it changes citation logic or inserts ads, your visibility evaporates overnight. Over-reliance on one engine is the new platform risk.
The no-SaaS SMB alternative
- Manual benchmarking: list 20 questions you most want to be found for; ask each monthly on ChatGPT, Perplexity, Gemini; log whether you're mentioned and which page is linked. A single Google Sheet does it.
- GA4 custom segment: isolate AI sources into a channel group; measure real referral traffic and downstream behavior — free, and first-party data you own.
- DIY structured data: add FAQPage/Article JSON-LD to deep content and validate with the Rich Results Test — no subscription needed.
FAQ
AI traffic is only 1.08% — do I still need GEO?
Depends on your sector. In high-AI-Overview fields like health and finance (48.7% / 25.8%), the clicks eaten by AI summaries far exceed the 1.08% average. Measure your own numbers before deciding — don't judge by the site-wide average.
ChatGPT or Perplexity?
By traffic, ChatGPT is 87.4% of AI referrals — most efficient; but Perplexity attributes sources nearly every time (~97%), friendlier to brands wanting citations. Focus on ChatGPT while monitoring citations via Perplexity.
Can I track this without buying a tool?
Yes. A GA4 AI-source segment plus a monthly 20-question manual benchmark captures ~80% of what matters, at zero cost, with first-party data you own.
What content gets cited by AI most?
Per the benchmark, blog-style deep content is cited most (300,000+ pages), not homepage copy. Write deep FAQs, comparison tables, and step guides with structured data for the best inclusion odds.
My take
Two extremes are fashionable: one shouts that AI search will replace Google, spend big on GEO; the other sees 1.08% and calls it a non-issue. Both are wrong. The real signal isn't traffic size — it's that the entry point is shifting from attributable clicks to unattributable citations: your brand gets mentioned but you can't measure it, gets summarized but no one clicks through. For ScriptWalker, that's a repositioning opportunity: rather than selling AI-traffic growth (the numbers don't back that promise yet), sell AI-visibility infrastructure — structured data, clean semantic HTML, a GA4 AI-source dashboard. The former is unreliable short-term traffic; the latter is an asset useful no matter which engine wins. Telling clients the honest 1.08% builds more trust than hype.
Sources
- Conductor: 2026 AEO / GEO Benchmarks Report (first-hand)
- Conductor (BusinessWire): Conductor Unveils 2026 AEO / GEO Benchmarks Report (first-hand)
- Google Search Central: AI features and your website (first-hand)
- Muck Rack AI citation study (third-party longitudinal data)
- GA4 docs: Channels and default channel group (third-party)