Buyers stopped starting on Google

Watch how anyone under forty actually researches a purchase now and you’ll notice something the dashboards haven’t caught up to yet: the search box isn’t the first stop anymore. The first stop is a question, typed into ChatGPT or Perplexity or Claude, phrased like you’d ask a knowledgeable friend. “What’s the best tool for X if I care about Y?” The model answers in a paragraph. It names three or four options. And the buyer’s shortlist is now those three or four names, assembled before a single one of those companies knew a buyer existed.

That’s the whole shift, and most of GTM is still optimizing for a funnel that now begins one step earlier than they’re measuring.

The funnel grew a new first step

For twenty years the top of the funnel was a ranked list of blue links, and an entire discipline (SEO) grew up around getting into it. The mechanics were legible: keywords, backlinks, rankings, clicks. You could see your position. You could watch it move.

The new first step is different in a way that breaks the old instruments. When a model answers “what should I use for X,” it isn’t returning a ranked list you can climb. It’s synthesizing an opinion from everything it has read about you and your category. You’re either in that synthesis or you’re not, and there’s no page two to be stuck on. There’s just the answer, and the absence from it.

SEO got you onto the list the buyer chose from. The new game is being named in the answer that builds the list. Those are not the same skill.

What “optimizing” even means now

Call it AEO or GEO: answer engine optimization, generative engine optimization, pick your expansion. The label matters less than the mechanic underneath it, which is genuinely new:

• The unit isn’t a ranking, it’s a mention. Did the model name you, unprompted, when someone described your category? That’s the thing to move.
• The corpus is everything, not your site. A model’s picture of you is built from your docs, yes, but also Reddit threads, comparison posts, forum answers, other people’s writing. Your owned pages are one input among hundreds.
• The measurement is a distribution, not a number. Ask the same question ten times and you’ll get slightly different answers. “Are we cited?” isn’t yes/no; it’s how often, in what position, next to whom. Share of voice, not rank.

None of this fits the SEO dashboard, which is exactly why there’s an opening. The teams treating “am I in the answer?” as a measurable, movable metric are building the muscle everyone else will scramble for in a year.

Where I’d put my attention

If I were starting from zero today: measure it: pick your ten highest-intent category questions and track, weekly, whether the major models name you and beside whom. Shape the corpus: go where your category is actually discussed and be useful there, in public. And write the definitive thing: for each question that matters, publish an answer so clear that both a human and a model would rather quote you than reassemble it from scratch.

The buyer’s journey grew a new first step, and it’s the most decisive one. Most companies aren’t in the room for it yet. The window where that’s a fixable disadvantage rather than a permanent gap is open right now. It won’t stay open.