There is no keyword report and no quality score to tune. The levers that exist are the prompts you surface on, the advertisers you sit beside, and the creative you put in the card. All three are observable.
Optimize the context hint first. It decides which conversations you enter, which is the single biggest determinant of whether an impression was ever winnable.
Order matters. Rewriting creative before fixing the context hint improves a card that is still appearing in the wrong conversations.
You cannot optimize a placement you cannot see. Start from the prompts that genuinely surface ads in your category, then ask honestly which of them your product answers better than anyone else.
See trigger prompts →Competition is the variable most advertisers never look at, because it is invisible inside their own account. If four established brands already answer a conversation well, that conversation is expensive attention.
See competing advertisers →This is the highest-leverage change available and it feels wrong, because impressions drop. That is the point: a narrower hint trades unwinnable impressions for winnable ones.
Write a better context hint →The card renders inside an answer. Benchmark your title and description against what advertisers are actually shipping, and cut anything that reads like a banner.
Compare your creative →Sometimes the campaign is fine and the category is the problem. A nearby, less contested niche often returns more for the same budget while the platform is still young.
Find open categories →A modest result in a crowded category can be a good outcome; a strong-looking one in an empty category can still be underperformance. Compare like with like.
What you can measure →There is no magic setting. ChatGPT ads have no keyword bids, no match types and no quality score to tune. Optimization here means improving how well you fit a conversation, and the evidence for that lives outside your account: who else appears, what they say, and which prompts surface ads at all.
What advertisers ask once the first campaign has run.
Work the levers that exist: the context hint that decides which conversations you enter, the creative inside the card, and the category you compete in. There is no keyword bid or quality score to tune, so optimization means improving fit rather than adjusting settings. Start by narrowing the context hint, because it determines whether an impression was ever winnable.
The context hint. It decides which conversations your ad is eligible for. A hint that is too broad buys impressions on questions your product does not answer, and no amount of creative or budget work recovers those.
Yes, and that is usually the intended outcome. Impressions on conversations you cannot win are not valuable. Narrowing trades volume for relevance, which is what a conversational surface rewards.
Change one variable at a time and give it enough volume to read. Because the platform is new and reporting is thinner than Google or Meta, frequent simultaneous changes make it impossible to attribute an improvement to anything.
Not inside your account beyond your own numbers. Externally, you can see which advertisers compete in your category, what their creatives say, and which prompts surface ads. We track Thousands advertisers and Tens of thousands creatives across Hundreds categories for exactly this purpose.
Questions about the data, a brand you expected to see, partnerships, or access to the intelligence layer. We read every message.