THE BUYER IS AN AGENT: How to sell to something with no eyes – and why SEO, GEO and AEO just became one job.

Your next customer may never look at your website. A machine will read it, compress it, and hand a human three options. You are no longer charming a person in a chair – you are being quoted by something with no patience and no taste for adjectives.

PART ONE: HOW OFTEN IS THIS ACTUALLY HAPPENING?

1. Today: 45% of consumers already use AI in the buying journey (IBM, Jan 2026) – 41% research, 33% reviews, 31% deal-hunting.

2. Next season: 60% of shoppers expect to use an agent within 12 months (Kearney). By 2030, half of online shoppers will, carrying ~25% of their spend.

3. The traffic already arrived: AI-referred retail traffic grew 393% year over year in Q1 2026 (Adobe); Shopify orders from AI search rose 13x.

4. And it converts: AI-referred visitors convert ~42% better than traditional search – +37% revenue per visit, +87% time on site.

5. B2B is the bigger stadium: Gartner expects 90% of B2B purchasing to route through agents by 2028: $15 trillion, machine to machine.

6. The trust gap is the strategy: 65% of Americans trust AI to compare prices; only 14% let it buy unsupervised. The agent makes the shortlist, the human signs it.

7. You can’t see it happening: About 70.6% of AI referral sessions log in GA4 as “direct” – you are read far more than your dashboard shows.

PART TWO: THE WORLD CUP PART

8. Nobody gets signed off a highlight reel. An agent scouts you like a federation scouts a striker: minutes played, conversion rate, injury history. “Award-winning” is not a stat.

9. Semi-automated offside doesn’t hear the crowd. It reads coordinates. Your hero video and parallax scroll are crowd noise. JSON-LD is the coordinate.

10. It’s a group stage, not a final. Agents shortlist three to five. FAQ schema makes content ~3.2x likelier to surface; full Product + Review schema hits ~61.7% citation.

PART THREE: WHO DOES WHAT ON MONDAY MORNING

11. Designers: beauty that can’t be parsed is decoration. Real headings, real text – 44.2% of LLM citations come from the first 30% of a page. Answer first, then be gorgeous.

12. Developers: JSON-LD for Organization, Product, Offer, FAQ. Server-render the facts. Allow GPTBot, ClaudeBot and PerplexityBot – a quarter of the top 1,000 sites block them. Ship llms.txt: hours, not weeks.

13. Marketers: trade adjectives for numbers. Princeton’s GEO research measured up to 40% visibility gains from citing statistics and sources. About 86% of AI citations trace to brand-controlled pages.

14. Owners: pick one thing you can prove you are best at, publish the proof with a date on it, keep the date fresh. Vagueness is now a pricing disadvantage.

The referee is a machine now. Argue with it, or hand it clean footage and let it put your name on the scoreboard. 88% of marketers have already started; the other 12% are about to learn what relegation feels like.

Leave a Reply