We Spent Real Money on ChatGPT Ads for Small Business. Here's Why You Should Avoid Them For Now.
Updated: Aug 20
Updated August 2026 with corrections and results from a second campaign. See the update at the end of this post.
Here's a sentence I didn't expect to write in 2026: I bought ads inside ChatGPT. And then I watched $500 vanish in about an hour with, as far as I can tell, absolutely nothing to show for it. No conversions, almost no traffic, nothing I could verify.
So consider this your field report from someone who actually clicked the buttons, entered the credit card, and ran a real campaign for a real business (ours). If you've been wondering whether ChatGPT ads are worth it for a small business, I have good news: I did the flailing so you don't have to. Grab a coffee. This one has a plot twist, and it isn't a happy one.
A quick note on why we even tried this. We run Out of the Box Advisors, a small business coaching firm, and we've been helping owners make smart marketing decisions since 2012. We also believe in testing things ourselves before we recommend them (or warn you off them). AI is reshaping how customers find businesses, something we dug into after Google's I/O 2026 announcements, so when OpenAI opened its ad platform to small businesses, we felt obligated to kick the tires. With our own money, on our own campaign.
What follows is exactly what happened.
Unboxed Wisdom: What $500 Bought Us
Short on time? The whole experiment in one breath:
We spent $500 on ChatGPT ads for our own coaching business. It disappeared in a little over an hour.
The cost per click looked great (around $3.50). That's very reasonable for our industry. Except low CPC that results in zero conversions is worse than a high CPC that works.
We got almost nothing we could verify. Of the 140 clicks OpenAI reported, only about 13% ever reached our site. Zero leads. Zero consultations.
The billing was strange. We were charged five separate times in partial amounts, and the campaign data didn't appear in the dashboard for about a day.
Setup was smooth, but the platform felt raw. OpenAI calls it a beta. Based on our experience, that's generous.
Our verdict: not yet. We wouldn't spend another dollar here right now, though we may test again once it matures. More on that at the end.
Wait, ChatGPT Has Ads Now?
Yes. As of 2026, ChatGPT shows ads, and small businesses can buy them. After years of being ad-free, OpenAI began placing sponsored results inside ChatGPT conversations and, in May 2026, opened a self-serve Ads Manager that removed previous minimum-spend requirements.
Translation: you no longer need a Fortune 500 budget to advertise inside the world's most popular AI chatbot. Any small business with a credit card can now jump in, which is exactly what makes this worth understanding.
If you've used ChatGPT lately on a free or lower-tier account and noticed a suggested product or service woven into an answer, that's the ad product. According to OpenAI's own announcement, advertisers can now create ChatGPT ads through a beta self-serve Ads Manager, with cost-per-click bidding and what they describe as expanded measurement tools.
The pitch is compelling on paper: reach people at the exact moment they're asking an AI for help solving the very problem your business fixes.
And to be fair, the strategic logic is real. We've written before about how AI is changing the way customers find small businesses. People increasingly ask ChatGPT the questions they used to type into Google. So the idea of showing up inside those conversations isn't crazy. The idea is sound. However, the version of it we paid for isn't there yet.
What Is It Like to Set Up a ChatGPT Ad Campaign?
The setup itself is genuinely easy. The interface, however, feels like it was built in a hurry. Creating our campaign was fast and the steps were clear: define your business, set a budget, launch. If you've ever built a Google or Facebook ad, you'll feel at home in minutes. The information they ask for is straightforward and the flow is smooth. That part, credit where due, was painless.
Now for the slightly awkward part, with a bit of irony? The Ads Manager looked and felt cheap.
The best way I can describe it is that the interface had the vibe that it was AI generated from a single prompt. It was technically functional, but unrefined, and definitely with none of the polish you'd expect from, you know, the AI company themselves. If you were to tell me they had an intern whip this up in a matter of minutes I would believe you.
Buttons and layouts felt thrown together and just bland overall.
Now, they do label the product a beta, so some roughness is fair. But there's a difference between beta and unfinished, and this leaned toward the latter. If I'm being blunt, it felt closer to an alpha wearing a beta name tag.
None of that would matter if the thing worked. A clunky interface that prints leads is a great deal. So we launched a campaign, set our $500 budget, and settled in to watch what the most hyped platform in tech could do for a small business coaching firm.
I figured we had days of data ahead of us.
The budget didn't survive lunch. That's where the wheels came off.
Do ChatGPT Ads Actually Work? Our Personal Experience.
In our single test, no. We spent the full $500, OpenAI reported clicks, and not one of those clicks appeared in any of our own analytics. We generated zero leads. This is the heart of the story, so let me lay it out plainly and let you draw your own conclusions. The numbers OpenAI showed us and the numbers our own tools showed us didn't match. Like, at all!
The timeline, exactly as we watched it unfold:
The budget burned fast. Our entire $500 was spent in a little over an hour. For a small test campaign, that's fast. Uncomfortably fast.
The CPC looked healthy. We paid roughly $3.50 per click. In isolation, that's a perfectly reasonable number, which we'll put in context in a second.
The clicks didn't show up anywhere we could verify. We run detailed analytics through Wix, and we cross-check with Google Analytics 4. The traffic OpenAI reported sending us didn't appear in either one.
The data lagged by about a day. The campaign dashboard didn't populate with meaningful reporting for roughly 24 hours, so during the actual spend, we were flying blind.
The billing came through in five separate charges. Rather than one clean transaction, our card was hit five different times in partial amounts of the total budget.
The result: zero leads, zero consultations, zero anything. No form fills. No calls. No new business. Nothing we could tie back to the spend.
Now, fairness is a trait that I value, so let me take a moment and avoid just straight focusing on the negatives.
First, attribution on this platform is a known mess. Analysts have documented that ChatGPT ad clicks routinely lose their referrer data and can get quietly filed under "Direct" traffic in GA4 unless you set up special tracking. We knew that going in, which is why we didn't just hunt for a "ChatGPT" line in our reports.
We watched overall traffic, every source, Direct included. And here's the thing: our site gets about 1,000 visitors in a typical month. If the clicks OpenAI reported had actually shown up, that would've been a Black Friday crowd hitting a lemonade stand in a single hour. There was no line. Maybe a curious squirrel. Total volume barely moved.
And here's the postscript that makes this whole thing a little funnier. A few days after publishing this article, we put a modest boost budget behind it, the ordinary kind of promoted-post spend any small business does. That traffic showed up in our analytics the same day. Not subtly, either. Our daily sessions jumped to roughly three to four times our normal baseline and stayed there for the length of the campaign, then settled right back down when it ended. Same website, same analytics, same tracking setup that supposedly missed about 140 ChatGPT clicks. So our instruments work fine. They caught the boost instantly. They just never saw the traffic we paid OpenAI $500 for.
Second, the billing. When five separate partial charges hit the card, my first move wasn't to grab a pitchfork. I went and read OpenAI's help documentation, which says multiple charges can sometimes be payment pre-authorizations, and I gave them the benefit of the doubt. Then the reporting finally landed a day later and made it clear the charges were for the campaign itself, not holds.
At that point, with everything else stacked up, we didn't even bother reaching out to support. Some experiences answer their own support ticket. A lesson we chat with our clients about often.
And we aren't the only ones with reservations. An agency-side test by Enspire found the platform can only target at the country level (no state, city, or ZIP targeting), offers no conversion-focused bidding at all (objectives are limited to reach and clicks), and lacks a central business manager. Their verdict was the polite, agency-newsletter version of ours: still in its infancy. (Ours involved WAY more sighing.)
Another marketer's write-up carried the blunter title "We Tested ChatGPT Ads. The Results Weren't Pretty." So take our experience as one data point among a growing pile, most of them pointing the same direction.
Was the Cost Per Click Actually Good?
The CPC was reasonable on paper, and that's exactly what makes this a cautionary tale. Our roughly $3.50 per click sits right around the 2026 cross-industry average and is very fair for professional services and coaching. For context, WordStream's 2026 benchmarks put average search CPCs across industries in the low-single-digit dollars, with professional and education services generally landing a few dollars per click and truly expensive verticals like legal running far higher.
So on a spreadsheet, $3.50 looks like a win. But the lesson every small business owner should tattoo somewhere visible is this: a cheap click that doesn't turn into a real visitor isn't cheap. It's expensive. A $3.50 click that produces a genuine, tracked prospect is a bargain. A $3.50 click that never verifiably lands on your website is just $3.50 set on fire. In our case, about 140 times, and we could only ever account for a fraction of them.
Cost per click is a vanity metric if the clicks don't show up and convert. You can compare this to the real marketing metrics that actually matter, the ones tied to leads and revenue, not just activity.
This is one of the most common small business marketing mistakes we see: chasing a good-looking top-of-funnel number while ignoring whether it produces anything at the bottom. ChatGPT ads gave us a beautiful CPC and a completely empty pipeline.
Should Your Small Business Try ChatGPT Ads Right Now?
I would say with 100% confidence, NO. For most small businesses, this isn't the moment. If you have money you're genuinely willing to lose for the sake of experimentation, and you want bragging rights as an early tester, go for it with a tiny budget and clear eyes. But if you need every marketing dollar to work, and most small businesses do, we'd sit this one out for now and let the platform mature.
We aren't anti-ChatGPT, anti-OpenAI, or anti-AI. The opposite, actually. We think AI is one of the most useful things to happen to small business in a generation, and we fully expect AI advertising to become a serious channel. Early movers in Google Ads and Facebook did eventually win big by getting in cheap and early. That pattern is real, and it's the strongest argument for trying this.
But early doesn't mean reckless. The current version of this platform, at least the one we paid to use, wasn't ready to responsibly spend a small business's money. When your own analytics can't confirm the traffic you paid for, the single most important feature of any ad platform, trust in the numbers, is missing. Until that's fixed, we'd rather put your budget somewhere accountable.
Our take: Treat ChatGPT ads like a science experiment, not a marketing strategy, for now. If you must try, cap it at an amount you wouldn't miss, watch your own analytics like a hawk, and don't believe a number you can't independently verify.
What Should You Do Instead of ChatGPT Ads?
Put your money where you can measure it. For most small businesses, that means proven channels and a tight grip on your numbers. The whole reason our ChatGPT test stung was the lack of verifiable results. So the antidote is to focus your marketing budget on places where every dollar is trackable and tied to real outcomes, and to know your numbers well enough to spot a dud fast.
A few grounded moves that beat gambling on an unproven platform: lean on channels with mature tracking (Google, Meta, and good old-fashioned referrals), obsess over the metrics that actually predict revenue rather than vanity clicks, and revisit your plan regularly.
Our mid-year review guide is built for exactly that kind of critical check-in. And if paid ads are part of your mix, make sure they're set up and measured properly. That's a big part of what our small business marketing services exist to handle.
Is that advice boring? A little. But boring channels that pay the mortgage beat exciting ones that eat it. The point isn't to avoid new things; it's to avoid spending money you can't account for. New channels are worth testing, but tests should be small, measured, and clear about what they return.
What We Would Need to See Before Round Two
Because we do plan to retest eventually, here's our minimum checklist for when this platform deserves small business money: these are the specific gaps that made us walk away, and fixing them is what "out of beta" should actually mean.
Real conversion tracking and bidding. Today the campaign objectives stop at reach and clicks. Until you can optimize toward an actual lead or sale, you're buying activity, not outcomes.
Local targeting. Country-level targeting is useless for most small businesses. A local, family-owned business shouldn't be paying to reach the entire United States.
Same-day reporting. A budget should never be able to finish spending before the dashboard shows a single number.
Verifiable traffic. Clean, documented tracking so the clicks you buy show up in your own analytics, not just theirs.
Clean billing. One campaign, one clear charge, with an invoice a bookkeeper won't side-eye.
The Final Take: We Tested ChatGPT Ads So You Don't Have To
ChatGPT ads are real, they're available to small businesses, and one day they may be a genuinely great place to spend marketing dollars.
That day, based on our $500 experiment, isn't today. We got a great-looking cost per click, a burned budget, a confusing string of charges, zero verifiable traffic, and zero leads. As a first impression, that's a rough one.
We'll probably test it again down the road because we believe in the underlying idea and platforms improve fast. When we do, you can bet we'll report back with the same down and dirty good, bad, or ugly.
Update, August 2026
Two corrections and a follow-up.
We wrote that the clicks never showed up in our analytics at all. That was slightly too strong. Going back with the campaign date isolated, our direct traffic roughly tripled on the day the campaign ran, and every one of those visitors was new. Some of those clicks landed. Out of 140 we paid for, we can account for fewer than 20.
We also need to correct the record on our own setup. We tagged that campaign's links with UTM parameters, which is the standard way of telling your analytics where traffic came from. They didn't come through. That's part of why we couldn't identify the traffic, and it's a lesson worth passing on. Tagging your links isn't enough on its own. You have to confirm the tags survive the trip.
We should also note that our Google Analytics wasn't capturing traffic reliably during this period, which we found afterward while auditing our own tracking. Our Wix analytics worked correctly throughout, and it's the source behind every number here.
Then we ran it again in August. Same landing page, $175, tagged links that worked this time. What changed wasn't the cost per click, which was similar. What changed was delivery. The second campaign delivered 68% of its paid clicks to our site. The first delivered 13%. Measured by what we actually paid per visitor who arrived, the second campaign was roughly six times more efficient. We did register one conversion event, though nothing that felt like a lead on our end.
The clearest difference we can see is how the two campaigns spent. The first burned $500 and ten thousand impressions in under an hour. The second spread a smaller budget across a week. We can't tell you why that produced such different click quality, and OpenAI's reporting doesn't give us the data to find out. But it's the pattern we observed, and if you test this platform, pacing your budget looks like it matters a great deal.
Credit where it's due, too. When we published that checklist in July, it was a wish list. Since then OpenAI has shipped a good chunk of it: conversion tracking with optimized cost-per-click bidding, location targeting instead of country-only, and average daily budget controls that appear to actually hold. That's three of our five, faster than we expected. The one that still matters most, being able to verify the traffic you paid for, is the one we still can't check off.
So our verdict softens without reversing. Go small, tag everything, verify the tags on arrival, and don't let a budget spend itself in an hour.
We're leaving the original piece as written above. The correction is the useful part.
Want Small Business Marketing That Actually Works?
If you'd rather spend your marketing budget on things that produce results you can actually see, that's our whole job.
We've helped small business owners cut the guesswork since 2012. Book a free consultation and let's build a plan where every dollar has to earn its place, no vanishing clicks allowed.
Frequently Asked Questions
Does ChatGPT have ads in 2026?
Yes. After years of being ad-free, OpenAI began showing sponsored placements inside ChatGPT conversations in 2026, initially for users on the free and lower-cost tiers. In May 2026 OpenAI launched a self-serve Ads Manager and removed previous minimum-spend requirements, which opened the platform to small businesses for the first time. Higher-tier paid subscribers generally see an ad-free experience.
How much do ChatGPT ads cost for a small business?
There's no longer a large minimum spend, so you can start small. Pricing works on a cost-per-click model similar to other ad platforms. Across two campaigns we spent $500 and then $175, paying roughly $3.50 and $2.77 per click, both reasonable rates for our industry. But cost per click is only half the story. What matters more is how many of those clicks actually arrive on your site. In our first campaign only about 13% did. In our second, 68% did. Same platform, same landing page, very different economics.
Do ChatGPT ads actually work for small businesses?
It's very early, and results varied enormously between our two campaigns. In our first test, a $500 campaign that spent itself in about an hour, only around 13% of the clicks we paid for could be traced to our site, and we generated zero leads. A second $175 campaign in August, spread across a week with link tags that worked properly, delivered 68% of its clicks and produced a single conversion event. So the honest answer is that the platform can deliver traffic, but delivery quality swung wildly for us and OpenAI's reporting doesn't explain why. If you test it, go small, pace the budget across days rather than hours, and verify your tracking before you spend.
Is it worth advertising on ChatGPT right now?
For most small businesses we'd still say not yet, though our view has softened. The strategic idea is strong, since customers increasingly ask AI tools the questions they used to Google, and OpenAI has shipped real improvements since our first test, including conversion tracking, location targeting, and daily budget controls. What still gives us pause is verification: in our first campaign most of the clicks we paid for never verifiably arrived, and the reporting gave us no way to find out why. If you do test it, cap the budget at an amount you wouldn't miss, spread it over days, and confirm your tracking works before spending.
Why did my ChatGPT ads spend so fast or get billed multiple times?
We experienced this ourselves: our full budget was spent in about an hour and our card was charged in five separate partial amounts. OpenAI's help documentation notes that multiple charges can sometimes relate to payment pre-authorizations, but in our case the data later confirmed the charges were for the campaign itself. Fast spend can happen when a campaign gets a burst of clicks, but the lack of same-day reporting made it hard to monitor in real time. Watch your budget and your own analytics closely if you test the platform.
Why don't ChatGPT ad clicks show up in Google Analytics?
Usually it's a tracking problem on your end, and we learned this the hard way. ChatGPT ad traffic often loses its referrer information due to referrer policies and in-app browsers, so real visitors land in your analytics as Direct traffic rather than showing a ChatGPT source. UTM parameters are supposed to solve that, but in our first campaign we tagged our links and the tags did not survive the trip, so the traffic arrived unlabeled and we initially concluded almost nothing had come through. Going back later with the campaign date isolated, our Direct traffic roughly tripled that day and every one of those visitors was new. The lesson: tagging your links isn't enough on its own. Click your own tagged ad link before you spend real money and confirm the parameters still appear in the URL and register in your analytics.
