ChatGPT Ads Aren’t Perfect Yet — Here’s Why That Shouldn’t Stop You From Testing Them
Here’s the honest answer to the question every CPG and DTC brand is quietly asking: No, ChatGPT ads are not ready to be a core channel. But yes, you should be testing them right now. Waiting for a new ad format to be “perfect” before touching it is how brands end up two years behind. The question isn’t whether ChatGPT ads have issues — they do. The question is whether you have the framework to test them intelligently while everyone else is paralyzed by doubt.
Hesitation is a strategy, but it’s rarely a winning one.
Why “Wait and See” Is a Risk, Not a Safe Bet
Every major ad channel in the last decade — programmatic display, paid social, connected TV — went through an ugly adolescence. Viewability was a disaster. Attribution was broken. Brand safety was a mess. Brands that waited for the dust to settle missed the lowest CPMs, the least competition, and the steepest learning curves while their budgets were still small enough to absorb the lessons.
ChatGPT ads are in that exact moment right now.
The inventory is new, the audience intent is genuinely different from search or social, and the measurement infrastructure is still being built. That’s not a red flag. That’s an opportunity — if you walk in with clear eyes and a disciplined test structure.
What Makes AI-Native Ad Placements Different
ChatGPT isn’t a search engine and it isn’t a social feed. It’s a reasoning interface. When someone asks ChatGPT for a recommendation — the best protein powder for endurance athletes, a clean skincare routine for sensitive skin — they’re not browsing. They’re deciding.
That intent signal is qualitatively different from a keyword match or a behavioral target.
This matters enormously for CPG and DTC brands because the consideration window is compressed. A user getting a product recommendation from an AI assistant is often closer to purchase than someone clicking a paid search ad. That’s a big deal. It also means the cost of a bad ad experience — an irrelevant placement, a tone-deaf message — is higher than it is on a passive scroll format.
The Questions You Need to Ask Before You Spend a Dollar
Before any brand we work with tests a new channel, we run through a short, non-negotiable checklist. For ChatGPT ads specifically, those questions look like this:
- Can we isolate this inventory in reporting? If your measurement stack can’t separate ChatGPT ad conversions from your existing search or social attribution, you’ll never know what actually worked.
- What does the creative spec actually demand? AI-native placements may reward utility-forward copy over brand-awareness messaging. Do you have the creative flexibility to test that?
- What’s our incrementality baseline? According to Meta’s own research on incrementality testing, lift studies are the only reliable way to understand true channel contribution. You need that same discipline here — before you scale anything.
- What’s our exit criteria? Define what “this didn’t work” looks like before you start. A test without a stopping condition isn’t a test, it’s a spend.
- Are we testing the channel or the creative? These are two different experiments. Don’t conflate them or you’ll learn nothing clean.
This is the kind of strategic rigor our performance media work is built around — not assumptions, not hype cycles.
How to Measure Incrementality in Uncharted Inventory
Incrementality measurement in a new channel is hard. It’s also the only number that matters.
Last-click attribution in AI-native placements is going to be noisy, possibly for years. The user journey from ChatGPT recommendation to purchase likely spans devices, sessions, and channels. Standard UTM tracking will undercount. Platform-reported conversions will overclaim. This is not speculation — it’s the same pattern we’ve seen with every new inventory type.
The practical answer is to run a geo-based holdout test or a matched-market experiment before committing meaningful budget. Spend in half your target markets, go dark in the other half, and measure the delta in sales or site traffic. It’s not glamorous but it works, and it gives you a defensible number to bring to leadership.
Our strategy team helps brands design these tests so the learning is actually actionable — not just a slide deck that says “results were mixed.”
Don’t Let Doubt Be Your Competitor’s Advantage
The brands that will win in AI-native advertising over the next 18 months are not the ones who waited for a flawless product. They’re the ones who built measurement discipline early, accumulated creative learnings while CPMs were still cheap, and showed up to the channel with a point of view instead of a panic.
ChatGPT ads have real growing pains. So did every channel that matters now.
The move is to test smart, measure rigorously, and stay skeptical of anyone — including AI platforms — who tells you the results are better than they look.
FAQ: ChatGPT Ads for CPG and DTC Brands
What are ChatGPT ads?
ChatGPT ads are paid placements served within OpenAI’s ChatGPT interface, appearing when users ask questions or request recommendations. They represent a new category of AI-native advertising where brands can reach users at high-intent moments inside a conversational AI environment.
Are ChatGPT ads worth testing for CPG or DTC brands right now?
Yes — with structure. The inventory is early-stage, which means lower competition and more room to learn. But you need clean measurement setup, isolated reporting, and a defined incrementality test before you commit meaningful budget.
How do you measure incrementality for a new ad channel like ChatGPT?
The most reliable method is a geo-based holdout test: run ads in select markets, go dark in matched markets, and measure the difference in sales or conversions. Platform-reported attribution on new inventory types typically overclaims, so external validation is essential.
What’s the biggest mistake brands make when testing new ad formats?
Conflating creative performance with channel performance. If your creative isn’t built for the format and the results are poor, you’ll incorrectly conclude the channel doesn’t work — and miss the actual opportunity. Test them as separate variables.
Ready to test new channels without guessing? Junction 37 builds performance media strategies for CPG and DTC brands that are built to produce real answers, not vanity metrics. See how we work →
Chris Pyne, Founder, Junction 37 – 30+ Years in Performance Media