They type a question. Maybe it's "who are the best providers of [the thing you sell]." Maybe it's "what should I look for when evaluating [your category]." Maybe it's something more specific, a problem they're trying to solve, and your category is one of the answers.
ChatGPT generates a response. A few companies get named. A few get described. A few get cited as the ones to consider.
Your company is in that answer or it isn't.
That moment didn't exist five years ago. The buyer never typed your category into a search bar. They never landed on a SERP. They never clicked a result. They never visited your website. They never filled out a form. They got an answer, formed an impression, and either added you to the shortlist or didn't. All of it happened in an interface you don't measure, on a platform you don't pay, in a layer of the buyer journey most marketers haven't built for. 73% of B2B buyers now use AI tools like ChatGPT and Perplexity in their research process, and the trend is accelerating.
This is the layer the rest of this article is going to make you reckon with. But to get there I have to walk you through the two layers you already know, because the failure most B2B marketers are running into right now is not that they're missing the new layer. It's that they're operating in only one of the layers they already had, and they don't realize the channels they're using are not the channels they think they're using.
The three layers
There are three layers to B2B visibility right now. Most marketers are operating in one of them and calling it a strategy.
Demand creation is the layer that reaches buyers before they know to search. The buyer is not in market. The buyer may not know your category exists. The job is to plant the idea, build familiarity, and put your brand in front of the right person at the right title at the right company, repeatedly, over time. The channels that do this work are LinkedIn paid, Meta, programmatic display, newsletter sponsorships, podcast placements. The metric that matters is not click-to-form. It's whether the buyers who came through this layer convert at higher rates when they later show intent.
Intent capture is the layer that catches buyers who are already searching. The buyer has a problem, has decided to look for a solution, and is typing a query into Google. The query itself is the qualification. A buyer searching "best [category] providers" is in a different place than a buyer searching "[your competitor] vs [other competitor] pricing." The keyword is the signal. This is the layer most B2B marketers are most comfortable in because the math is legible and the buyer is actively raising their hand. It's also the layer that fails first when category demand is thin, because if nobody is searching, there is nothing to capture.
Answer presence is the layer I opened with. The buyer is researching, but the research is happening inside an AI interface. ChatGPT, Perplexity, Claude, Gemini, AI Overviews on Google. There is no click. There is no spend. There is no campaign. The work is structural. It happens in how your content is written, how it's organized, how it's marked up, and whether it's the kind of content an AI can extract, synthesize, and cite. Most marketers have not built this layer at all. They are still writing content for Google's algorithm and assuming AI will figure it out. AI is figuring it out. It's just not figuring out them.
Three layers. Each one operates on different rules. Each one requires a different kind of investment. Each one is measured differently, fails differently, and pays off on a different timeline.
The marketers who are getting crushed right now are the ones who built one layer, called it a paid strategy, and are wondering why the program isn't working. The marketers who are quietly winning are the ones who figured out that these are not interchangeable channels, they're sequential roles in how a buyer actually moves toward a decision.
The failure: a campaign that generated zero qualified leads
A B2B services company in a niche category runs a LinkedIn paid campaign. It generates zero qualified leads. Not underperformed. Zero.
The numbers tell the story without needing to be exact: cost per lead nearly tripled in the same campaign period. Form completion fell by more than half. And when you pull the audience data, the companies engaging with the ads are consulting firms, medical device companies, nonprofits. Not a single decision-maker from the actual target buyer profile. The campaign is reaching people. The people are just entirely wrong.
The diagnosis everyone reaches for is "LinkedIn doesn't work for us." It's a reasonable conclusion if you're looking at the numbers in isolation. It's also wrong.
The real diagnosis was simpler and more expensive: an intent capture conversion ask on a demand creation channel, measured like a Google Search campaign.
Think about what LinkedIn is at its core. A buyer on LinkedIn is not searching for a solution. They are not in decision mode. They are scrolling. They are reading. The intent that defines a Google Search query does not exist on LinkedIn. The buyer has not raised their hand. You are interrupting them.
And what was the conversion ask? A demo request. A primary conversion ask that requires a buyer to be ready to talk to sales, on a channel where buyers are explicitly not ready to talk to sales, targeting a niche B2B category with almost no existing search demand.
Form completion didn't collapse because LinkedIn's ad product got worse. It collapsed because the ask didn't match where the buyer was. This is the layer mismatch problem, and it's more common than most marketers want to admit, because it means the channel didn't fail. The strategy failed. Those are very different conversations to have with leadership.
The rebuild: three campaigns, same time, different intents
When the diagnosis is layer mismatch, the fix is not a better creative. It's not a lower bid. It's rebuilding the campaign around the right channel for the right stage.
Three campaigns ran simultaneously, each mapped to a different stage of buyer intent. Not sequentially. Simultaneously. Buyers are not synchronized. A buyer who is ready to make a decision this quarter cannot wait six weeks while you finish your awareness flight.
This is where most B2B paid strategies get the sequencing wrong. Running awareness first, then consideration, then conversion is a planning convenience. It reflects how a marketing team thinks about campaigns, not how buyers move through a decision. If your conversion campaign is not in market when that buyer searches, you are not in the auction. Someone else is.
Top of funnel. Broad, exploratory, problem-oriented keywords. "How to reduce manual data entry costs." The destination is not a product page and not a demo request. It is content. The conversion ask is a download, not a sales conversation. The goal is not to convert this buyer today. It is to get them into the remarketing audience at a moment when they are receptive to information and not ready for a pitch.
Middle of funnel. Comparison-driven, solution-oriented keywords. "Automated data entry software comparison." The destination is a page built to help the buyer evaluate. The conversion ask is a secondary action: a consultation request, a scorecard, an interactive tool. Something that requires more commitment than a content download and less than a sales call.
Bottom of funnel. Vendor-specific, intent-loaded, decision-stage keywords. Brand searches. Competitor comparisons. The destination is the primary conversion ask. The bid is higher because the keywords are lower volume and higher value, but every click is a buyer in a decision window.
And here is a detail that gets missed more often than it should: the keywords are the words buyers actually search, not the words the company uses internally to describe what they sell. A company might call their service "intelligent document processing." A buyer might search "automated data entry software." If your keyword strategy is built around internal brand language, you are optimizing for searches that don't exist. That distinction sounds obvious. It is almost universally ignored.
RLSA is the connective tissue most marketers never build
Buyers who visited your site through an earlier campaign get tagged. When they later search on Google, your campaign recognizes them and bids higher to make sure you show up. You are not treating them like a cold prospect. You are treating them like someone who already knows you exist.
That sounds like a retargeting tactic. It is actually a structural decision about how your campaigns relate to each other. The demand creation campaigns at the top of the funnel are not just building awareness. They are building the audience pool that the conversion campaign bids against.
Without demand creation in market, that pool doesn't exist. The conversion campaign is bidding cold on every search. You are paying full price for buyers who have never heard of you, competing against brands they may already know.
The mechanics start with the audience. You need a remarketing tag firing on every page of your site, not just the conversion pages. Most marketers tag the demo page and call it done. Then you build tiered audiences, not one undifferentiated list: site visitors in the last ninety days at a modest bid adjustment, multi-page or asset-download visitors higher, high-intent page visitors highest. The audiences layer onto the search campaigns rather than running as separate campaigns, so the same keyword costs a different bid depending on the searcher's relationship to your brand.
The window has to match the buyer journey, not the default. A B2B services company with a six-month sales cycle should be running a 180-day window at minimum.
This is why ToFU budget is structural, not optional. Marketers who cut demand creation to protect conversion budget are cutting the pipeline their conversion campaign depends on. Performance does not drop immediately. It drops two quarters later, when the audience pool has aged out and the conversion campaign is bidding cold against searches that used to convert. By the time the data shows it, the damage is done.
It also explains why most B2B paid attribution gets the math wrong. The conversion campaign catches the buyer. The demand creation campaign created the buyer. Attribution shows the catch. It rarely shows the creation.
Answer presence: the buyer may never click
Now we come back to where this started. The buyer opens ChatGPT, types a question, and your brand is in that answer or it isn't.
Most marketers have not built this layer at all. They are still optimizing for keywords and click-through rates, still measuring success by traffic. None of that captures what happens when a buyer asks an AI a question and never visits your site. AI Overviews now reduce organic click-through rates by 58% on the queries where they appear, and that's just the searches still happening on Google. The searches happening inside ChatGPT, Claude, and Perplexity don't appear in your analytics at all.
Structure in self-contained sections. The AI is not reading your page the way a human does. It is extracting fragments that can stand alone in an answer generated somewhere else. If your section requires the previous three paragraphs to make sense, it will not be cited.
Write declarative headers. "What is intelligent document processing" gets cited. "The Future of Workflow Innovation" does not. The AI is matching the buyer's question to your header.
Make the first sentence the answer. Not the setup. Not the context. If the AI extracts only the first sentence of your section, that sentence should still answer the question.
Mark it up. FAQ schema. HowTo schema. Article schema. The crawler is not guessing, you are telling it. And FAQ pages are not an afterthought, they are one of the highest-leverage formats for this layer, provided they mirror the questions buyers actually ask rather than the ones you wish they were asking.
Picture two pieces of content on the same topic. The first is a 2,000 word post in flowing narrative prose, clever branded headers, an argument that unfolds across the page and requires reading start to finish. The second is eight sections, each with a declarative header mirroring a buyer question, each opening with a direct two-sentence answer, with schema and a comparison table. A human reader might prefer the first. The AI will cite the second.
AI does not cite content because the content exists. It cites content because the content is structured to be cited. Pages with well-organized headings are 2.8x more likely to earn citations in AI search results.
Answer presence is the layer where the buyer's mental model of your category gets shaped before they ever form a search query. By the time they search, the shortlist is already taking shape. The buyer who arrives at intent capture has already been through demand creation and answer presence, whether you were present in those layers or not. If you weren't, somebody else was.
One layer is not a strategy
In a category with thin demand, you cannot afford to operate in one layer. Intent capture only works when buyers are searching, and in thin-demand categories they are mostly not. Demand creation only works when you have something to convert buyers into when they show intent later. Answer presence only works when there is a foundation of content the AI can find.
The marketer who only runs Google Search is invisible to every buyer who hasn't formed a query yet. The marketer who only runs LinkedIn is reaching buyers who aren't ready to convert and measuring the channel by the wrong metric. The marketer who only invests in content is hoping AI will find it without doing the structural work to make it findable.
Demand creation plants the idea. Answer presence shapes the research. Intent capture catches whoever searches. RLSA ties the paid layers together. Every piece has a structural role.
If you take one thing from this article, take this: when a paid campaign fails, ask which layer you were operating in and whether the conversion ask matched. Ask whether the channel was being measured by the right metric. Ask whether the campaign was alone or part of a system. Most of the time, the channel didn't fail. The strategy did. That is a different conversation to have with leadership. It's also a more honest one.
The giveaway: a channel map
Two axes. The vertical axis is how much intent the channel requires from the buyer, low to high. The horizontal axis is how precisely you can target the right person, broad to specific. Every major B2B paid channel plots somewhere on the grid, and you can audit your own strategy against it in about thirty seconds.
It's free, it runs in your browser, and there's no form: the B2B Paid Layer Map.