ChatGPT Ads vs Google Ads: First $500

Every guide to ChatGPT Ads quotes the same number: $3–5 a click. That number is not a price. It is the maximum bid OpenAI suggests you start at, published in its own help documentation, and repeating it as though it were a cost is the single most common mistake in the coverage of this channel.

Here's the thing. The interesting difference between these 2 platforms is not price at all. It is how much of the buying decision you are allowed to make yourself.

The difference that decides everything else

On Google Ads you tell the platform what to match. You pick the words, you pick how loosely they may be interpreted, and you tell it what never to match. That control is the whole product, and every skill an advertiser builds sits on top of it.

ChatGPT Ads does not work that way, and the gap is wider than most write-ups admit.

Read OpenAI's campaign targeting documentation and the main campaign field is targeting.locations.include: a country, a region, or a designated market area. Since launch, platform targeting and custom audiences built from your own customer lists have joined it.

That is the complete list, and not one of those controls describes a query.

One level down, at the ad group, you get a field called context_hints — a free-text list of topics and phrases describing when your product might be useful. It looks like a keyword list. It is not one.

OpenAI's own wording is worth quoting exactly, because it is the sentence the rest of this page hangs on:

These hints help guide ad matching, but they are not exact-match keywords and do not guarantee delivery in specific conversations.

You can supply up to 2,000 of them. You still cannot make one fire.

Here's the kicker:

Selection is a relevance-weighted, second-price auction that reads the conversation's context and intent alongside your title, your body copy, and your landing page. Translation: the model decides whether you are relevant, and the strongest lever you own is not a bid. It is how legibly your offer is written.

That is the same conclusion the research on creative effectiveness reaches from a different direction — why creative is now the targeting instrument covers how the other platforms got here.

So what does that mean for you?

2 things. A tight, genuinely relevant ad on a modest bid can beat a vague one bidding higher, which is unusually good news for small budgets. And the discipline that makes Google accounts profitable — ruthless exclusion — has almost nothing to grip. Negative Phrases help, but only for advertisers who qualify.

What you can actually control

That third row is the one to sit with. On Google, the search terms report is where the money is found.

Case in point:

I once got access to a business-to-business fintech account running on autopilot with high-intent keywords and the wrong match modifiers, quietly pulling in consumer traffic nobody had excluded. Roughly 80% of the budget was going to people who were never going to buy.

The fix was obvious the moment I opened the search terms — and afterwards the company had to hire salespeople to keep up.

There is a way around it, and it is the first thing I would tell anyone running this channel. If you cannot see the query, build the report from the other end. Split your context_hints across separate ad groups, point each ad group at its own landing page, and let the destination do the reporting for you.

You never learn the exact wording. You do learn which conversation was worth paying for — and that is the half of a search terms report that actually changes a decision.

What a click actually costs

OpenAI recommends starting at a maximum bid of $3–5 per click, and sets the default maximum for impression buying at $60 per 1,000. Both figures come from OpenAI's own help documentation.

Those are ceilings you set, not prices you pay. In a second-price auction you pay what it takes to beat the next advertiser, which on a young platform with thin competition is often well under your maximum.

The honest answer is that nobody outside the platform can tell you your number, because it depends on your category, your geography, and how relevant your ad reads.

What I can tell you is what we see. Across the local businesses I work with — medical clinics, automotive, retail, and high-billable professional services — cost per lead usually lands somewhere between $30 and $150.

I know that is a wide range. It is wide because those are genuinely different businesses, and any narrower number would be dishonest.

We publish what we actually pay for ChatGPT clicks, by vertical.

One more thing worth knowing about budgets: daily budgets now have a floor. OpenAI's standard minimum in the US is $25 a day, and it varies by market and currency.

What you can measure

Ads Manager reporting currently covers impressions, clicks, spend, click-through rate, average cost per click, average cost per thousand impressions, and conversions. That is the whole list.

Conversions come through a measurement pixel or a server-to-server conversions application programming interface, the same shape as Meta's. Static tracking parameters survive the click, so if you tag your landing page URLs the traffic shows up in the analytics you already run.

Compared with Google, you are giving up the query view and most of the audience view. You are keeping the thing that actually matters, which is whether a click turned into a customer.

No joke.

If your conversion tracking is not already trustworthy, fix that before you spend a dollar here — a channel you cannot measure at query level is a channel you must measure at outcome level, or not run at all.

It isn't either/or

I get asked which one is better, and it's the wrong question. This isn't black and white.

What matters is coverage: the right amount of spend on each platform where your buyers actually are, so that when someone goes looking for a company like yours, you show up clearly. An empty platform is not a saving. It is a missed opportunity.

Here's where ChatGPT surprised me.

I've seen it work well even on a small spend, purely for brand mentions. People ask ChatGPT about a company by name, or about anything close to it, especially when they are comparing options.

They also ask the questions a search box rarely gets: what the company actually does, whether it solves their problem, and how easy it is. An ad under that answer reaches someone who is already weighing you up.

So where does the first $500 go?

Alongside, not instead. That is the whole answer.

Budget is a variable, not a number. A fixed monthly figure is a planning fiction; what should actually move your spending is seasonality, your capacity to fulfil the work, and how many buyers are genuinely in the market this month.

Put another way:

So the question is never "should I move money from Google to ChatGPT." It is "can I afford a small, honest experiment while the channel that already produces customers keeps producing them."

Here is the one I can point at.

We ran ChatGPT Ads alongside Google for an automotive client, sized to be readable rather than to look serious.

It took 1–2 weeks before the numbers said anything I would trust. When they did, we kept it running and moved more budget into it.

One test in 1 vertical is not a law. It is 1 honest data point, which is 1 more than most of the pages you will read on this can offer.

Work through it in this order.

1. Is your proven channel dialled in?

   If your Google account still has obvious waste in it, that is where the next $500 earns the most. Fix the leak before you drill a new hole.

2. Can you measure an outcome?

   No working conversion tracking, no test. You would be buying traffic you cannot judge.

3. Is your country open?

   Self-serve Ads Manager now covers 63 countries. The table below the list has every one.

4. Is your category allowed?

   Several are not yet, and some of the biggest small-business categories are among them. Check the restricted list first.

5. Then start small and move quickly

   Size the test to produce a readable result, not to look serious. Scale what works, on evidence, even when that feels uncomfortably slow.

When ChatGPT Ads is the wrong call

I would not run it yet if any of these are true.

• Your buyers are in a hurry and nearby. Someone searching for an emergency plumber is not workshopping the decision with an assistant. Local search and Local Services Ads still win that moment.
• You have no conversion tracking. Covered above, and worth repeating, because it is the most common reason a test produces nothing but an invoice.
• Your category is restricted. Healthcare, financial services, and legal services are gated behind manual approval right now, and several categories are disallowed outright. I have one sitting in manual approval as I write this, submitted recently enough that I honestly cannot tell you yet whether the answer comes back in days or in weeks. Nobody publishes that number, me included. So start the approval before you plan the campaign around it, and check which regulators apply to your industry while you wait.
• Nobody is watching the account. Autopilot is how ad budgets die. That is true on every platform, and it is truer on one where most negative controls are a suggestion.

None of that makes it a bad channel. It makes it an early one.

And it gets better.

Buyers are arriving faster than advertisers, which is exactly the shape of a real first-mover advantage — but only for people who bring the same discipline they would bring to search. New channel, old rules.

Common questions

Tomas Kolafa

Founder, Bytown

14+ years running paid acquisition, managing $85M+ in ad budgets — co-founded the ad agency Growth Media, led marketing ($0–50M) at RVezy.com. I’ve spent the budgets, run the campaigns, and read the reports.