Every few years, a new ad channel shows up, and the same pressure follows. Someone in the room says, “Our rivals will be there, so we should test it now.” It sounds smart. It also puts real money behind a channel no one has measured yet.
ChatGPT Ads are the latest case. People use the tool to look up suppliers and solve work problems, so showing up there feels like an easy win.
I wanted facts, not hype, so our team ran ChatGPT Ads for some of our clients. In our experience, the targeting felt loose, the leads did not fit the businesses and even the ads we saw as users felt off.
Those results point to one idea I want to put on the table. A conversation is not the same as buyer intent. ChatGPT Ads sell a topic, not a buyer, and that gap decides whether a channel can generate leads. This article shares my view, backed by what our team saw in client tests.
Key Takeaways
- Our view: ChatGPT Ads match a topic, not a buyer, so today they work as a brand experiment, not a lead generation channel.
- Targeting: In our client tests, targeting was too loose for firms that need qualified leads.
- Lead fit: The leads we got did not fit the businesses we ran them for. A Workshop Digital test drew 739 clicks and no qualified leads.
- Ad match: The ads we saw as users did not match what we asked. A SE Ranking study found about 14% of ChatGPT Ads were off topic for the prompt.
- Privacy paradox: According to OpenAI, advertisers get only aggregate results, never the chats, so the privacy that builds user trust also limits the signal brands can use.
- Audience limit: OpenAI says ChatGPT Plus, Pro, Business, Enterprise and Education plans do not show ads, so part of the B2B audience is out of reach.
- My test: I judge every new ad channel on Signal, Fit and Proof, and ChatGPT Ads fall short on all three today.
- Better move: Keep any ChatGPT Ads budget small and capped, and put your main effort into earning a place inside answers with GEO (generative engine optimization).
What Are ChatGPT Ads and How Do They Work?
ChatGPT Ads are paid spots that appear below answers inside ChatGPT. The platform does not match keywords. It matches ads to the topic of the chat, and advertisers can only guide that with a context hint. Brands supply the ad and budget, but the platform decides when an ad fits.
OpenAI says it picks ads using the chat topic, past chats and past ad interactions, and gives advertisers only aggregate results such as views and clicks.
ChatGPT Ads are still in the beta phase. Ads show only to Free and Go users, while paid plans stay ad free. That limit matters a lot for B2B.
Curious how paid channels should fit your growth plan?
What Did We See When We Tested ChatGPT Ads for Clients?
We saw three problems. The targeting was loose, the leads did not fit the businesses, and the ads we saw as users did not match our chats. For a firm that needs good leads, those three issues make ChatGPT Ads hard to justify as a lead channel.
The targeting felt too loose
Good targeting means the right buyer sees the right message. In our tests, that match was weak. The platform seemed to read each chat in broad strokes, so our clients’ offers reached people with little reason to care.
The leads did not fit the business
The leads we got did not fit the businesses we ran the ads for. More leads did not help, since a lead that does not fit never turns into a sale. It also eats sales time and makes reports look better than they are.
Even the ads we saw felt off
We also used ChatGPT as everyday users, and the ads shown to us did not match what we were asking. If trained marketers notice that gap, a busy buyer sees it faster. It can also hurt trust in the brand behind the ad.
What this meant for our clients’ budgets
We hold every paid channel to the same questions. Does the lead fit the offer? Can the sales team use it? Would we put more budget behind it? On the proof from our tests, we could not say yes for ChatGPT Ads.
That does not make the platform a failure. At this stage, ChatGPT Ads are a place to learn, not to scale, so learning money should not share a plan with lead money.
| Area | What marketers expect | What we observed |
|---|---|---|
| Targeting | Intent-led audiences that match the offer | Loose targeting that was not precise enough for our clients |
| Lead relevance | Inquiries that fit the business | Leads that were irrelevant to the businesses we tested |
| Ad relevance for users | Ads that match the conversation | Ads shown to us that were not relevant to our questions |
| Overall readiness | A channel ready to scale | An early experiment that needs monitoring |
Source: Wolfable client tests and internal observations during our ChatGPT Ads tests.
Bottom Line: In our tests, ChatGPT Ads delivered reach, but not the relevance a lead channel needs.
Why Is a Conversation Not the Same as Intent?
A search query states a need in a few words, so the platform can read intent. A chat is open, long and full of side topics. ChatGPT Ads match the topic, but a topic shows no buyer. I call this the topic-buyer gap, and it is why targeting felt loose in our tests.
Picture two people asking about the same subject. One wants a quote this week. The other is writing a school report. A topic match cannot tell them apart, yet only one of them matters to your sales team.
A second force sits behind this, and I think it is the heart of the matter. Users trust ChatGPT with open, private questions. OpenAI says advertisers get only aggregate results, with no access to chats or personal details. So the more private the chat, the weaker the signal a brand can use. I call this the privacy paradox of conversational ads.
Getting inquiries that do not fit your business?
The controls reflect it. As Casey Burridge found, each ad group takes one context hint, with no keywords, no negative targeting and no search terms report. A brand can describe the chats it wants, but it cannot exclude the ones it does not, or see where its ad ran.
The wider data fits what we saw. SE Ranking tested 50,006 commercial prompts in 20 US niches. Ads showed on 25.94% of them, and 14.35% of those ads were off topic. That is about one in seven, and it matches what we noticed as users.
I expect matching to improve. I doubt that topic matching alone can close a gap that privacy creates.
Are Other Marketers Seeing the Same Thing?
Several published tests from other advertisers show the same pattern: plenty of clicks, weak targeting controls and few qualified leads. These are small, self-reported samples, so I treat them as signals, not proof. Still, they point the same way as our own results and as the platform’s design.
Workshop Digital spent $2,319 on three B2B campaigns and got 739 clicks and no qualified leads, while Google Search delivered two.
GravityKit’s Casey Burridge saw a similar pattern. Paid ChatGPT visits averaged 5 seconds of engaged time and produced no purchases. His verdict: “Intent was the problem, not loose thumbs.”
On lead fit, Floyd Blaikie found that only 5 of the 146 organizations behind 336 paid clicks matched his ideal customer profile, Search Engine Journal reports. And MediaPost notes that SE Ranking’s own test drew over 97,000 impressions and 1,263 clicks, but very few sign-ups.
Search Engine Journal’s review adds the wider mood from its first six months: no universal benchmark for “good” exists yet, and Ads Manager has no Auction Insights or impression share. That is what an early product looks like.
| Source | What they tested | What they reported |
|---|---|---|
| Workshop Digital | Three B2B campaigns, $2,319 | 739 clicks and no qualified leads |
| GravityKit | Three-week test, about $550 | No purchases, 5 seconds of engaged time |
| SE Ranking | Two weeks, four countries | 1,263 clicks and very few sign-ups |
| Floyd Blaikie | B2B campaign, visitor identification | 5 of 146 organizations matched his ICP |
Source: published advertiser reports. These are self-reported, small samples with different goals, so treat them as signals, not benchmarks.
Bottom Line: Different advertisers report the same signal: clicks arrive, but fit does not.
Where Do I Stand on ChatGPT Ads Today?
I see ChatGPT Ads as a channel to watch, not a channel to scale. My view rests on what our team saw in client tests and on one core idea: a conversation is not intent. I urge small tests with strict checks.
As Sunny Patel, Marketing Head, Wolfable
Good marketing puts the right message in front of the right person at the right time. In the tests we ran for clients, the targeting was not tight enough to do that.
A conversation is not the same as intent. People tell ChatGPT everything, from homework to hiring plans. A topic match cannot tell a buyer from a bystander, and lead generation needs a buyer.
The leads that came in did not fit the businesses we were running ads for. That tells me the problem starts upstream, with who sees the ad. Better creative cannot fix an audience that was wrong from the start.
Here is the paradox I keep coming back to. The more private a chat is, the less signal a brand can use. Matching will get better, but I doubt it closes that gap soon. Until then, keep any ChatGPT Ads budget small and separate, and judge it only on leads your sales team confirms.
To be clear, I do not call the platform a failure, and neither does our team. I ask for discipline: test small and protect the channels that already work.
Want an honest review of where your ad budget works hardest?
Why Do Irrelevant Leads Cost More Than They Seem?
Poor-fit leads cost more than their price per lead suggests. They eat sales time, skew reports and can steer auto bids toward the wrong audience. A cheap lead that never fits your business costs more than a pricey lead that does.
We see this often with clients in manufacturing, healthcare and finance. Their sales cycles are long and their offers are specific. One poor-fit lead can take a specialist real time to sort out.
That is why we judge paid channels by lead fit first. Our Google Ads management approach ties setup, landing pages and tracking to sales goals, so volume never hides a poor fit.
How Should Leaders Judge Any New Ad Channel?
I judge every new ad channel on three tests: Signal, Fit and Proof. Signal asks if the platform can tell who the buyer is. Fit asks if the leads match your offer. Proof asks if you can verify results outside the platform. In our tests, ChatGPT Ads fell short on all three.
Signal: can the platform tell who the buyer is?
Google reads a typed need. LinkedIn reads a job title. ChatGPT reads a topic. The stronger the signal, the less you pay for waste, and in our tests a topic was not enough.
Fit: do the leads match the offer?
This is where our tests failed first. A lead only counts when it fits the offer and the sales team can use it.
Proof: can you check results in your own system?
Platform reports are a start, not the end. Your CRM should confirm every result, and with no view of the chats that showed an ad, that check matters even more.
| Factor | Google Ads | LinkedIn Ads | Meta Ads | ChatGPT Ads |
|---|---|---|---|---|
| Signal | Search keywords and intent | Job title, company and industry | Interests, behavior and demographics | Chat topic, guided by one context hint |
| Who sees the ads | People searching on Google | Professionals on LinkedIn | Facebook and Instagram users | Eligible Free and Go users only |
| Advertiser control | Mature and detailed | Mature and detailed | Mature and detailed | Limited: no keywords or negatives |
| Proof | Established reporting | Established reporting | Established reporting | Basic reports, no search terms view |
| Best current role | Capture active demand | Reach named roles | Build awareness and demand | Small, capped experiment |
Source: Wolfable analysis and published advertiser reports. Platform features change often, so recheck before planning spend.
Bottom Line: Use established channels to capture demand, and use ChatGPT Ads only to learn.
Who Should Test ChatGPT Ads Now, and Who Should Wait?
Test only if you have a healthy lead engine, a capped budget and a way to judge lead fit with your sales team. Wait if you need steady lead volume, lack a sales feedback loop or are still fixing your core channels. Curiosity alone is not a plan.
It is reasonable to test if
- Your Google Ads, SEO and LinkedIn Ads already bring in good leads.
- You can cap spend at a level you can afford to lose.
- Your sales team will rate every lead for fit.
It is better to wait if
- You need steady lead volume this quarter.
- Your tracking and lead checks are still unclear.
- Your core buyers are senior people who may use plans without ads.
For manufacturers, clinics and financial firms, we urge caution. These buyers choose slowly and with care, so reach without fit adds noise. Every test needs an end date and a stop rule.
Ready to build lead quality into every campaign?
Where Is Conversational Advertising Heading?
I expect conversational ads to mature, but the winners will be brands that earn trust inside answers, not only brands that pay to appear next to them. Paid placement will have a role. Being the source an AI tool trusts will carry more weight.
This is where generative engine optimization, or GEO, comes in. It helps your brand become part of the answer when buyers ask AI tools about your category. ChatGPT can name brands with no ad at all.
Trust beyond your own site matters too. In my article on digital PR for GEO and AI search, I explain that outside mentions give AI tools more reason to trust a brand.
As Sunny Patel, Marketing Head, Wolfable
The bigger chance today is not to interrupt ChatGPT chats. It is to earn a place inside the answer. Keep a small test budget for ads, review it each month for lead fit, and put your main effort where buyers already trust you.
Why Does Wolfable Approach New Ad Channels Differently?
We judge every channel by the business result it creates, not by how new the platform is. That is why we test new formats on small budgets, check lead fit against our clients’ sales goals and say so plainly when a channel is not ready.
We specialize in manufacturing, healthcare and financial services, and we work from offices in India, the UK and the USA. Strategy, search, paid media, AEO, GEO and AI-led automation sit under one roof, so no channel choice is made alone.
The results show what that discipline can do. A healthcare provider we worked with saw 177% more patient leads at a 98% lower cost per lead after a scalable PPC plan. Results like that come from sound strategy, not from chasing the newest channel.
Final Thoughts: Are ChatGPT Ads Worth It Right Now?
ChatGPT Ads may grow into a strong channel. The format fits how people now research and decide. But a conversation is not the same as intent, and in our tests the targeting was loose, the leads did not fit and the ads we saw did not match.
So the answer to that pressure in the meeting room is simple. Being early only helps when a channel can reach the right people. A small test teaches you something. A big bet buys a pricey lesson.
If you want a straight view of where your paid and organic budget should go this year and next, talk to our team at Wolfable. We will help you build a plan that rewards fit, so you grow with the buyers who matter.
