“Free 10-second gut check first, no email required. It tells you whether AI crawlers can even read your site.” — from the Suede launch campaign
Buying research used to have a shape everyone understood. A person with a problem typed words into a search box, received ten blue links, clicked a few, formed a shortlist, and bought. Every marketing discipline of the last two decades is a strategy for winning some stage of that shape. Rank higher. Write the comparison page. Earn the click. Convert the visit.
That shape is dissolving, and it is dissolving quietly, which is what makes it dangerous.
Where the buyers went
The migration has three parts, and each one removes a place where you used to be able to compete.
First, the question moved. A meaningful and growing share of buying research now starts as a conversation with an AI assistant rather than a keyword query. The buyer does not type “crm small business.” They ask, “I run a 12-person agency and our client tracking is a mess. What should we use?” That is a better question, and it gets a better answer: a synthesized recommendation with names, trade-offs, and reasoning. Notice what it does not produce. It does not produce a results page you can rank on.
Second, the answer moved. Even inside traditional search, the answer increasingly arrives before the links do. Google’s AI Overviews synthesize a response at the top of the page. The buyer reads the synthesis. Some scroll past it. Many do not. The industry calls the broader pattern zero-click search: the query is asked, the answer is consumed, and no website receives a visit. The click you spent twenty years learning to win no longer exists for that query. I am deliberately not quoting adoption percentages at you. The widely circulated statistics in this space are mostly undated and unsourced, and this book does not traffic in unverifiable numbers. You do not need a statistic. You need Chapter 1’s exercise: you watched a machine answer a buying question in your own category. That behavior is the migration, observed firsthand.
Third, the shortlist moved. This is the part founders underestimate. In the old shape, the buyer assembled their own shortlist from many sources, and a determined vendor had a dozen chances to get on it: rankings, ads, review sites, a colleague’s mention. In the new shape, the engine assembles the shortlist and presents it as finished work. Three to five names, with reasons. Buyers still verify from there. But the frame is set. Everything the buyer does next happens inside a list a machine wrote in four seconds.
Why you did not notice
If this migration is real, why doesn’t it show up as a crisis in your dashboards? Four reasons, and they compound.
Your analytics cannot see it. When a buyer asks ChatGPT about your category and you are not mentioned, nothing happens in any system you monitor. No impression, no lost click, no bounce. The event that mattered most, a qualified buyer receiving a shortlist without you on it, is invisible by definition. Your dashboards are not lying. They are measuring a road while traffic reroutes to one they cannot see.
Your SEO metrics still look healthy. Rankings degrade slowly, and ranking well and being cited by answer engines are related but different achievements, which is the entire subject of Chapter 4. A site can hold page one for years while being systematically absent from the AI answers layered on top of and beside those results.
The losses arrive as silence. A competitor winning a bidding war is loud. A buyer who never contacted you because a machine handed them three other names is perfectly silent. There is no lost-deal record, because there was no deal. Silence reads as normal. It is not normal. It is unmeasured.
Nobody in your company owns this. Ask your team who owns AI visibility and watch the pause. The SEO person owns rankings. Marketing owns campaigns. The question of what six different answer engines say when a buyer asks about your category sits in the gap between every job description you have written.
The gut check
This is why the first tool I ever shipped for this problem takes ten seconds and asks for nothing. At optimize.suedeai.ai you can check whether AI crawlers can even read your site. Not whether your content is brilliant. Whether the machines are able to fetch it at all.
I built the free check first because of what I kept finding: companies investing real money in content while their infrastructure silently turned AI crawlers away. A robots.txt rule written years ago for a different problem. A firewall or bot-protection layer that treats every non-Google crawler as an attacker. A site rendered so heavily in JavaScript that a text-first crawler fetches an empty shell. Each of these is invisible from inside the company. The site looks perfect in a browser. The dashboards are green. And GPTBot, PerplexityBot, and ClaudeBot are bouncing off the front door. Chapter 6 covers how to run this diagnosis properly, bot by bot.
The migration is the reason that check matters. When buyers asked humans and search pages, being readable by machines was a technicality. Now that a layer of machines sits between you and a growing share of your buyers, being readable by machines is the new being open for business.
What this means before we go further
I want to close this chapter by defusing the two standard reactions, because both are wrong in instructive ways.
The first reaction is panic: rip up the marketing plan, chase every AI optimization trick on the internet. Wrong, because as Part II will show, much of what wins AI citations is disciplined fundamentals, done for machines as well as people, and most of the tricks are noise.
The second reaction is dismissal: this is a fad, buyers will always verify, search is not dead. Also wrong, and note that the dismissal attacks a claim I am not making. Search is not dead. Your site still matters, arguably more than ever, since it is what the engines read and cite. The claim is narrower and harder to dismiss: a new layer now sits between buyers and vendors, that layer writes shortlists, and you have almost certainly never audited what it writes about you.
The buyers moved. The dashboards stayed. This book is about closing that gap.
Check this yourself right now
Open Perplexity and ask the question a real buyer would ask at the start of research:
“I need [the problem your product solves] for [your customer type]. What are my options and how do I choose?”
Perplexity cites its sources with links. Look at the citations, not just the names. Which pages taught the machine its answer? Is any of them yours? Screenshot it, date it, save it next to the one from Chapter 1.