ChatGPT Ads: Getting in early is not a strategy

Generated with Avocode.Polygon 4

In brief

  • Since late August 2026, you can advertise in ChatGPT in Belgium. Only Free and Go users see the ads. Financial, legal, and medical services are excluded.
  • You measure results with familiar metrics like clicks, CPC, and conversions. You can't see which conversation your ad appeared in. And at €15 a day, you won't collect much data.
  • You can get on board, but treat it as a test with a clear hypothesis. Is your industry not allowed to advertise (yet)? Then make sure ChatGPT mentions you organically.

By Brecht Philips, AI Engineer at The Zoo

Remember the first websites? A banner on the left, a banner on the right, and one more at the top, blinking. Casinos, cell phone plans, and "Congratulations, you're our one millionth visitor!" Nobody asked for them, and nobody clicked on them. We've learned a lot since then. Advertising became contextual: an ad for hiking boots next to an article about the GR5 trail, or a search ad that answers what you just typed. Less shouting, more connecting with what's on someone's mind at that moment.

And now we're back at the beginning. AI is everywhere, and everyone wants to use it. Naturally, the companies behind it want to make money from it too, and not just through monthly subscriptions. Since late August, you can advertise in ChatGPT in Belgium. No blinking banners this time. The ads sit neatly below the answer, clearly labeled as sponsored. It all looks a lot more civilized than in the early days of the internet. But just like back then, nobody knows yet what actually works.

You've no doubt already seen the "be among the first" emails in your inbox. And I get the reflex. A new channel, little competition, and a starting budget of €15 a day. What could go wrong? Not much. But you won't get much out of it either, unless you first ask yourself four questions.

Question 1: Are you even allowed to advertise there?

Before you start crunching your budget numbers, check whether you're even allowed in. For now, OpenAI is keeping the door only slightly open. Only consumer industries are welcome: lifestyle and household products, local services, travel and experiences, digital products, and education. Financial, legal, and medical services are banned outside the US.

Are you a B2B service provider? Then you're probably still on the outside, no matter how good your campaign is. Is that a big deal? Maybe less than you think. Because even once the door opens, the question remains whether your ideal customer will ever see your ad. That brings us neatly to question 2.

Question 2: Does your target audience actually see these ads?

Ads only appear in the free version of ChatGPT and in Go, the low-cost plan. People on Plus, Pro, or Business don't see any.

On paper, that's not a problem: Free and Go together account for more than 90% of users. But who are those 90%? Do you sell bikes, trips, or kitchen appliances? Then your customer is probably among them. Are you targeting the IT manager at a mid-sized company? Chances are they use ChatGPT through a company license. Then your ad is invisible, no matter how good it looks.

So for B2B, that closed door isn't a big loss. Even with access, you'd be sitting at a lookout post in the jungle where your favorite herd never passes by.

Question 3: Does your customer actually use ChatGPT?

We like to pretend everyone talks to ChatGPT all day. But that's not the case. Many people type a question into Google and get an AI summary at the top. No app, no account, no chat.

And that's exactly where you can't advertise in Europe. Google is only testing ads in its AI answers outside the EU. Meta excludes Europe, and Perplexity has even scaled back its ads. The reason? It feared users would trust its answers less.

So if you want to advertise in an AI answer today, you really only have one option: ChatGPT. That makes it interesting, but not automatically relevant. Does your customer open ChatGPT to compare bikes? Or do they just Google it, like they've done for twenty years? Ask a few customers. It costs nothing.

Question 4: What are you actually going to learn from it?

For me, this is the most important question.

On paper, you get a solid dashboard. You see impressions, clicks, click-through rate, your average CPC and CPM, your spend, and conversions too, if you set them up. With UTM tags, you can also track the traffic in your own analytics tools.

Sound familiar? It should. You measure with the same metrics as on Google and Meta. The channel is new; the metrics aren't.

But what you don't see is just as important. You don't know which conversation your ad appeared in, or why. You can describe which topics or keywords your product fits, but OpenAI itself notes that it's not an exact match and comes with no guarantees. In Europe, only the current conversation counts for now, not what someone asked earlier. That's good news for your customer's privacy, but tricky for your analysis.

So what conclusion do you draw if your campaign doesn't deliver much? That your ad wasn't good? That your audience isn't there? Or that the user already got their answer in the chat and didn't need to click through? The dashboard tells you what happened, not why.

And do the math. OpenAI recommends starting with a bid of roughly €2.50 to €4.50 per click. With the minimum of €15 a day, that gets you three to six clicks. That's not enough to draw solid conclusions after a week.

There's one more thing you don't see: which conversation your brand appeared under. OpenAI keeps ads away from sensitive conversations, but safe doesn't mean suitable. Someone asks: "My flight got canceled and I'm stuck in Barcelona, what now?" Right below: "Up for a last-minute city break?" Brand safe? Absolutely. Brand suitable? Not even close. Completely within the rules, but no travel brand wants to be there. And can you specify where you don't want to appear? You can't.

There are no European case studies to fall back on yet. If you get in now, you're an early adopter. That's fine, as long as you know you're paying to learn, not to win.

Back to square one or a step forward?

In fact, advertising in a chat is the ultimate contextual ad. Someone asks a question and you appear below the answer. You can't get closer to the moment than that.

But a chat is a different place than a website. It feels more personal, almost like talking to an advisor. Will people accept ads there, the way they do next to a news article? Or will it feel like a salesperson is listening in? OpenAI has to strike a balance: enough ads to make money, but not so many that users drop off or start doubting the answers.

Nobody knows today how this will evolve. Will it stay a small block below the answer? Will in-chat checkout, which already exists in the US, come here too? And what happens when AI itself decides which product to recommend? It took the internet years to go from blinking banners to smart contextual ads. With advertising in AI, we're only on day one. Anyone who claims to know where this is headed has a better crystal ball than we do.

Test, don't gamble

So my advice isn't "don't do it." It's "do it deliberately":

  • Decide in advance what you want to learn. A hypothesis, not a hope.
  • Do honest math. The €15 a day is an average over a week. Some days cost twice that. Count on around €450 a month. At €2.50 to €4.50 per click, you'll get little data, so plan your test over several weeks.
  • Measure from day one. Set up conversion tracking and use UTM tags. That way you can also see what happens after the click in your own analytics tools.
  • Don't judge too quickly. Your spend data lags up to 8 hours behind. Look at your numbers after a week, not after an hour.
  • Compare with what you're already doing. Only by comparing it with your other channels will you know whether it adds anything.

ChatGPT Ads isn't hype, and it isn't a miracle cure. It's a new stretch of jungle where nobody knows the trails yet. Feel free to go exploring. But bring a map: know who you're looking for, where they are, and how you'll tell whether you've found them. Or better yet, don't head out alone. Even the pack at The Zoo doesn't know every trail yet, but we'd love to explore it with you.

FAQ: Advertising in ChatGPT

What exactly are ChatGPT Ads?

ChatGPT Ads are paid ads that appear below a ChatGPT answer and are labeled as sponsored. Which ad you see depends mainly on the conversation you're having at that moment, your language, and your approximate location. In other words, it's contextual advertising in a chat environment.

Do ads influence ChatGPT's answers?

According to OpenAI, no. The ads are separate from the language model, and advertisers can't steer the answer. They also don't get to see conversations or personal data, only metrics like impressions and clicks.

That's the official line, and it's crucial to the entire model. As soon as users start doubting the neutrality of the answers, ChatGPT loses exactly what makes it valuable. Perplexity already came to that conclusion and scaled back its ads.

In which conversations are ads not shown?

OpenAI doesn't show ads in sensitive conversations, such as those about personal health, mental health, and politics. As an advertiser, you can't verify this: you don't see the conversations.

Does advertising in ChatGPT generate leads, or mainly brand awareness?

The honest answer: nobody knows yet, not even OpenAI.

In theory, you're close to the moment of decision. Someone is comparing options, and you're right below. That sounds like leads. In practice, this often happens: the user already gets their answer in the chat, remembers your brand, and only looks you up later via Google or directly. You won't see that conversion in the ChatGPT dashboard.

So for now, think of ChatGPT Ads as a mid-funnel channel: something between awareness and conversion. Want to know whether it generates leads for your sales team? Then connect your campaign to your CRM and look beyond the click.

Sint-Truidersteenweg 142
3500 Hasselt
[email protected]