You ask ChatGPT for the best project management tool for a small team. Three seconds later, you have an answer. A clean paragraph, three named tools, a reason for each. You pick one and move on. You never opened a single website. You never saw a comparison page someone spent three weeks writing.
Now flip it around. Somewhere, a founder built one of those tools. They wrote that comparison page. And they will never know whether the chat search you just ran mentioned them, paraphrased them, or left them out.
That is the answer layer. For most brands, it is unmeasured.
You cannot fix what you cannot measure
In part one, we called this attribution loss. The click is going away, and the visit with it, so you lose the one signal you trusted to tell you the truth. You watch your organic traffic slip even though your rankings look fine, and Google Search Console gives you no clean reason why. In part two, we got under the hood. Humans skim and feel. Chat engines extract claims. Structure is the new prose.
So here is the turn. You stop reacting and start building. But you do not start by writing. You start by measuring, because you cannot engineer your way into a conversation you have never heard.
You would never run paid ads without a tracking pixel. You would be spending money blind. Chat search is the same problem, except almost nobody has the pixel yet. So step one is one plain question: when someone asks a chat engine about my category, what does it say about me? Cited by name, mentioned in passing, or invisible? Until you can answer that, every other move is a guess.
Stop writing pages. Start engineering claims.
Once you can see the answers, your job changes shape. The old job was to write pages. More pages, longer pages, pages stuffed with the right phrases. The new job is smaller and harder. You write claims a machine can lift cleanly and repeat without breaking.
A claim is a single, checkable statement. "Our onboarding takes about ten minutes." "We support teams from two people to two hundred." "We are SOC 2 compliant." Short. Specific. Easy to quote.
Compare that to the mush most pages are made of. "We pride ourselves on a smooth onboarding experience tailored to your unique needs." A human skims past it and feels vaguely warm. A machine reads it and finds nothing to extract. There is no fact in there to repeat, so the engine reaches for a competitor who said the quiet part plainly.
This is the gap from part two made real. Two brands, same quality. One wrote in claims. One wrote in fog. The one writing in claims gets quoted.
So go through your own pages with a cold eye. For every paragraph, ask one thing: what is the one fact a stranger could repeat after reading this? If you cannot find it, you have not written a claim. You have written a mood.
Win the extraction
Writing the claim is half of it. The other half is making sure the machine can grab it without tripping. A few things move the needle, and none of them are exotic.
Put the answer near the question. If a page asks "How much does it cost?" the number should sit right under it, not four scrolls down past your founding story.
Use real structure. Headings that name the thing. Lists when you list. Tables when you compare. A claim sitting inside clean markup gets lifted far more reliably than the same claim buried in a wall of text.
Repeat your core facts in plain words across the site, not once on an "About" page. Engines weigh things they see stated consistently. Say your key facts the same way more than once and you give them a reason to trust the pattern.
And date your facts. "As of 2026" signals a claim is current. Stale pages get passed over.
None of this is a trick. You are doing the thing every good writer should have done anyway: saying true things clearly, where a reader can find them. The reader is just a machine now, reading on behalf of your actual customer.
Earn the citation across every engine
Here is the part people miss. There is no single answer layer. There are several, and they do not agree.
Ask ChatGPT, Gemini, Perplexity, and Claude the same question about your category and you will often get four different lineups. One names you. One mentions you without a link. Two skip you. Each engine reads the web differently, trusts different sources, and weighs structure its own way. A win on one is not a win on all. Check only ChatGPT and you might feel great while Gemini quietly hands every lead to your competitor.
There is a second wrinkle every technical buyer raises first: these models do not answer the same way twice. Ask once and the result is noise, not a measurement.
That is why we built Prizvox to measure, not to guess. We ask each engine your category's real buying questions, several times each, across ChatGPT, Gemini, Perplexity, and Claude. Then we score how often you show up cited, merely mentioned, or invisible, so the number survives the fact that these models never answer the same way twice. Your visibility score is citation frequency across those repeated runs, broken out by engine, not a vibe with a decimal point. You also get the specific changes to make on the specific pages that are failing, and if you report to clients, a client-ready PDF in one click.
Here is what "specific" looks like. One client's pricing page opened with "Flexible plans built around the way your team works." Across forty runs, Perplexity cited a competitor on the cost question every time. We flagged the line. They changed it to "Plans start at $19 per user per month, billed monthly, no contract." On the next audit, Perplexity began quoting that exact sentence back. Same product. One claim, made checkable. The win came from the change, not from the story.
The payoff
We started with a loss. The last click is being abolished, and you are quoted in rooms you cannot see into. Then we looked at the machinery. Engines extract claims, and structure is the new prose. Now you have the rebuild, and it comes down to one sentence.
The brands that win the answer layer are the ones machines find easiest to quote.
Not the loudest. Not the biggest. Not the ones with the most pages. The most quotable. The ones who said true, specific things, in clean structure, consistently, across every engine, and then measured their work instead of hoping.
That is the whole game now. You measure where you stand. You engineer the claim. You win the extraction. You earn the citation. And you do it again next quarter, because chat search keeps moving and so do you.
The click built the last era of the web. The quote is building this one. Start writing things worth repeating, and start measuring whether they get repeated. Run your first audit and see what four engines actually say about you.