Browsers vs buyers: the core problem
Conversational AI is a research surface first and a buying surface second. When someone opens ChatGPT, they are usually working through a problem, learning a topic, drafting something, or satisfying curiosity. Only a minority are actively evaluating a purchase. Your ad, however, is matched to the topic of the conversation, not the buying intent behind it. That means you pay to reach everyone talking about your category, buyers and browsers alike.
15.6%
Share of ChatGPT prompts with commercial or transactional intent; the other 84% are research, curiosity, or task completion
The data makes the imbalance concrete. In an analysis of more than 50 million ChatGPT prompts, Profound classified only 9.5% as commercial and 6.1% as transactional. The rest split across informational (32.7%), generative or task-completion (37.5%), navigational (2.1%), and no clear intent (12.1%). Roughly 84% of ChatGPT prompts are not shopping-related at all, so if you target a broad topic and let delivery run, most of your impressions land on conversations that were never going to convert.
The flip side is the opportunity. Compared with traditional search, purchase intent on ChatGPT is rising fast. SparkToro’s analysis of 332 million Google queries found just 0.69% were transactional and 14.5% commercial. On ChatGPT, transactional intent has climbed to 6.1%, a roughly ninefold jump, as people ask the assistant to compare products, find deals, and shortlist vendors inside the chat.
9× more transactional
Transactional intent on ChatGPT (6.1%) vs traditional Google search (0.69%)
This matters because wasted ad spend is already the industry’s largest silent tax: the Association of National Advertisers found that only about 44% of every programmatic dollar reaches consumers effectively. Conversational ads give you a new signal to fight that waste. The conversation itself reveals which stage the buyer is in, something a two-word Google query cannot. Use that signal to concentrate budget on the roughly one in six conversations that can actually buy, and you turn a structural disadvantage, paying to reach researchers, into an advantage.
The rest of this guide is about capturing that signal. If you have not set up topic clusters and context hints yet, read the targeting guide first for the mechanics of matching ads to conversations without keywords. Here we focus on one question: how do you spend to reach buyers, not browsers?
The intent ladder
Not every relevant conversation is equally valuable. The single most useful mental model for conversion targeting is the intent ladder: the stages a person moves through from not knowing they have a problem to being ready to buy. Borrowed from Eugene Schwartz’s classic awareness stages, it maps cleanly onto how people talk to ChatGPT.
The five rungs
A problem-unaware user does not know anything is wrong yet. A problem-aware user feels the pain but has not looked for a fix. A solution-aware user knows categories of solutions exist and is exploring them. A product-aware user is comparing specific vendors. A purchase-ready user has decided and needs pricing, a trial, or a final nudge. Each rung deserves a different ad angle and converts at a different rate.
| Stage | Example ChatGPT prompt | Best ad angle | Relative CVR |
|---|---|---|---|
| Problem-unaware | “Why does my team keep missing deadlines?” | Name the problem; educational, no hard sell | 1.0× (baseline) |
| Problem-aware | “How do I stop tasks from slipping through the cracks?” | Frame the category as the fix; introduce your approach | ~1.5× |
| Solution-aware | “What kinds of tools help teams manage projects?” | Position your product in the category; differentiate | ~2× |
| Product-aware | “Is Asana or Monday better for a 12-person agency?” | Head-to-head proof, reviews, free-trial offer | ~3× |
| Purchase-ready | “Does ClickUp have a free plan and how do I sign up?” | Direct response: pricing, trial, start-now CTA | 4×+ |
2× to 4× higher CVR
Queries with transactional language convert two to four times higher than purely informational ones
The relative conversion figures above are directional, anchored to a consistent finding across search data: queries with transactional language (“buy,” “pricing,” “near me”) convert two to four times higher than purely informational ones. The practical implication is a budget rule: the higher a cluster sits on the ladder, the more you should be willing to pay for it, because each click is worth multiples more.
Most advertisers make the opposite mistake. Problem-unaware and problem-aware conversations are the most common, so broad targeting floods budget into the cheapest, lowest-converting rungs. Reaching buyers means deliberately weighting spend toward the top two rungs, product-aware and purchase-ready, even though those conversations are rarer and more competitive. You are not trying to reach the most people; you are trying to reach the most buyers.
Conversion-stage context hints
Context hints are the plain-language descriptions you give the platform to tell it which conversations your ad belongs in. For conversion targeting, the goal is to write hints that describe a buyer at a decision point, not a topic. Specificity is not only about relevance; it directly affects cost.
ChatGPT’s auction is relevance-weighted, as detailed in our CPC vs CPM bidding strategy guide. A hint that precisely matches a purchase-ready conversation earns a higher relevance score, so you win the impression at a lower effective bid. A broad hint scores lower on the conversations that matter, so you overpay to reach researchers. Specific, bottom-funnel hints raise relevance, raise CVR, and lower cost at the same time.
Good vs weak hints
| Weak hint (broad) | Why it wastes spend | Stronger conversion-stage hint |
|---|---|---|
| “project management” | Matches students, job seekers, and researchers | “10-to-50-person agencies comparing project tools and ready to switch this quarter” |
| “CRM software” | Mostly informational; catches career and how-to queries | “small sales teams evaluating a paid CRM to replace spreadsheets” |
| “running shoes” | Browsers, window shoppers, and gift researchers | “runners choosing a marathon shoe under $180 to buy before race day” |
| “email marketing” | DIY and free-tool seekers dominate | “ecommerce brands seeking a paid alternative to their current email platform” |
Layer hints by stage
Do not collapse everything into one hint. Create a separate hint, with matching creative and landing page, for each rung you target. A purchase-ready hint points to a pricing or trial page; a product-aware hint points to a comparison page. Mismatched stage and landing page is one of the most common reasons ChatGPT clicks fail to convert: a purchase-ready user dropped onto a generic homepage instead of a signup flow bounces, and you have paid for nothing.
Reading conversion signals
Even within a single topic cluster, individual conversations vary in intent. You can read that intent from language. These linguistic cues are the same ones performance marketers have long mined from search query reports, and they transfer directly to conversational surfaces, where they appear in richer, more explicit form.
The highest-intent cue is switching language (“alternative to X,” “migrating from”), because the person already pays a competitor. Budget language (“under $X,” “pricing”) signals someone who has decided to spend and is sizing the check, and comparison language (“best X for Y,” “X vs Y”) marks a product-aware buyer narrowing a shortlist. Readiness, local, and urgency cues round out the set, each mapped to an example in the table below.
| Signal type | Example phrasing | High-intent example by industry |
|---|---|---|
| Comparison | “best X for Y,” “X vs Y” | “best CRM for a real estate team” |
| Budget-qualified | “under $X,” “pricing,” “how much” | “project tool under $15 per user” |
| Switching | “alternative to X,” “migrating from” | “alternative to Mailchimp for ecommerce” |
| Readiness | “sign up,” “free trial,” “get started” | “how do I start a free trial of a booking app” |
| Local | “near me,” “in [city]” | “bookkeeper near me for a small LLC” |
| Urgency | “this week,” “before,” “deadline” | “need a payroll tool before next pay run” |
The reverse is just as important: cues that signal a browser. Language like “how does X work,” “what is X,” “free way to,” “DIY,” and “for a school project” marks research and low intent. You cannot always keep your ad out of those conversations through positive targeting alone, which is why the next lever, negative topic exclusions, is essential.
Negative topic exclusions
Negative topic exclusions are how you subtract the browsers your positive targeting inevitably pulls in. On Google these are negative keywords; on ChatGPT they operate at the semantic level, so a single exclusion covers many phrasings of the same idea. For conversion targeting, exclusions are not optional cleanup, they are a primary lever for lowering cost per conversion.
The pattern to exclude is consistent: conversations that are topically adjacent to your product but structurally incapable of converting. Five categories cover most of the waste, and they recur across nearly every vertical.
| Vertical | Conversations to exclude | Why it wastes spend |
|---|---|---|
| B2B SaaS | Students, job seekers, “build my own” DIY, career advice | No budget authority or purchase intent |
| Ecommerce | “how to make,” DIY tutorials, free-only seekers | Actively avoiding a purchase |
| Finance / fintech | Definitions, homework, “is X a scam,” academic research | Research and reassurance, not buying |
| Local services | Out-of-area users, DIY repair guides, hobbyists | Outside service area or self-serving |
| Courses / education | Free-resource seekers, “free download,” piracy queries | Unwilling to pay for the product |
$293 billion
Estimated annual digital ad spend that produces no measurable business impact, roughly 37% of budgets, much of it from poor targeting
Build your exclusion list from data, not guesses. After two to four weeks, review which conversation themes generate impressions and clicks but no conversions, and add them to your negatives. Because exclusions work semantically, you need far fewer rules than a Google negative-keyword list to cover the same ground: describe the type of conversation once, and the system recognizes its variations.
Geographic and audience constraints
Conversion targeting also has to account for who is even eligible to see your ad and what levers the platform does not give you.
Who sees ChatGPT ads
Ads currently serve to logged-in, US-based users, 18 or older, on the Free and Go ($8/month) tiers. Plus, Pro, Business, and Enterprise subscribers do not see ads. OpenAI has signaled expansion to Canada, Australia, and New Zealand during 2026. For conversion targeting this has a subtle implication: the Free and Go audience skews toward individuals and small teams rather than large-enterprise buyers, so bottom-funnel angles built around self-serve signup and quick time-to-value tend to outperform enterprise-procurement angles.
No demographic targeting
Unlike Meta, ChatGPT offers no demographic or interest targeting: no age, gender, income, job title, or lookalike audiences. You cannot say “show this only to marketing directors at companies with 200 or more employees.” At first this looks like a limitation. In practice it pushes you toward a better signal, stated intent. A person who writes “I need a payroll tool for my 12-person LLC before next pay run” has told you more about their readiness to buy than any demographic proxy could infer. Your job is to describe that buyer in a context hint, not guess who they are from a profile.
Geographic narrowing
You can narrow within the US by region, which matters for local and regionally licensed businesses. A conversion-focused local advertiser should combine geographic narrowing with local signals (“near me,” city names) and exclude out-of-area conversations outright, so a Dallas HVAC company does not pay to appear for a homeowner in Seattle.
Optimizing toward conversions
Targeting gets the right people in front of your ad. Optimization is how you and the platform learn which of those people actually convert, then shift spend toward them. On ChatGPT this capability is still maturing, but the moves you make now compound.
Set up conversion tracking first
You cannot optimize toward conversions you cannot measure. Install conversion tracking before you scale spend, so every signup, lead, or sale is attributed back to the conversation type that produced it. Our conversion tracking pixel and API guide covers both pixel and server-side setup. Without it, you optimize to clicks, which is how budgets end up funding browsers who click but never buy.
Feed conversions back
Once tracking is live, review conversions by topic cluster and stage every week. Reallocate budget away from clusters that generate clicks but few conversions toward those with proven conversion rates. This is the same discipline that separates efficient Google accounts from wasteful ones, applied to conversation topics instead of keywords. Pair it with our ROI measurement guide to tie those conversions back to revenue rather than stopping at the click.
Prepare for CPA bidding
ChatGPT ads launched with CPM (early 2026) and added CPC (April 2026). CPA bidding, where you set a target cost per acquisition and the platform optimizes delivery to hit it, is expected later in 2026. CPA is the end state for conversion targeting: it lets the algorithm decide, auction by auction, which conversations are worth bidding on to reach your cost target. The preparation is simple but time-sensitive: run campaigns now, accumulate conversions, and keep your tracking clean so the model has quality signal to learn from.
1.5 to 4× higher CVR
ChatGPT ad clicks vs Google Search clicks in matching verticals, because users self-qualify in the conversation
The payoff for getting this right is large, because the raw material is unusually good. Early advertiser data shows ChatGPT clicks converting 1.5 to 4 times higher than Google Search clicks in matching verticals (First Page Sage; Digiday), largely because users describe their problem, budget, and requirements before they ever click. Conversion optimization compounds that advantage; sloppy targeting squanders it.
Forecast before you spend
Every technique above shares one weakness: you usually learn what converts only after you have spent money finding out. Lapis closes that gap by letting you forecast and pre-select high-intent angles before launch, so you fund winners instead of discovering them.
800M+ weekly users
ChatGPT’s weekly active user base, the pool your conversion targeting is drawing buyers from
Performance Forecasting
Lapis Performance Forecasting predicts CTR and expected leads for each angle and audience before you spend, so you can rank creative and intent clusters by projected conversion value and fund only the high-intent ones. Instead of running ten variations for two weeks to discover the three that work, you narrow the field up front and put budget behind the projected winners, turning small improvements in which conversations you fund into large differences in cost per conversion.
AI Audience personas and segments
Because ChatGPT gives you no demographic targeting, the burden of knowing your buyer shifts to your context hints and creative. Lapis AI Audience builds detailed personas and segments from your product and market, then generates the bottom-funnel angles and context hints that match each purchase-ready segment. It turns the question “who is my buyer?” into concrete conversation descriptions you can target directly.
One prompt, every platform
From a single prompt, Lapis generates conversion-ready ads for Meta, Google, Reddit, LinkedIn, and ChatGPT in under three minutes, each formatted to the platform’s specs (for ChatGPT: a 50-character headline, a 100-character description, and a correctly sized image). Competitor Tracking reveals which angles rivals run so you can target underserved high-intent clusters, Campaign Studio lets you refine copy in natural language, and Web Analytics plus conversion tracking close the loop from prompt to conversion.
Lapis is YC F25, holds a 5.0 rating on G2, and has generated more than 10,000 campaigns across 30+ industries. Plans start at Free ($0), with Basic at $99/month and Pro (recommended) at $599/month; Enterprise pricing is custom.
Get started
Reaching buyers instead of browsers on ChatGPT comes down to a repeatable loop: map conversations to the intent ladder, write bottom-funnel context hints, read conversion signals, exclude the topics that never convert, and optimize toward a measured conversion goal. Advertisers who run this loop concentrate their budget on the roughly one in six conversations that can actually buy, and let competitors pay to educate the other 84%.
Start with Lapis to forecast high-converting angles, generate conversion-ready ChatGPT ads, and target buyers from day one. For a full platform overview, read our complete guide to ChatGPT ads; for ready-to-use prompt templates that capture buying intent, see our buyer-intent prompts for ChatGPT ads; and to measure what your targeting produces, use our ROI measurement guide.
