Ads have arrived inside ChatGPT. For years the advertising question for a small business was whether to pay for a spot above the search results. Now there is a second, newer one: whether to pay to show up inside an AI assistant's answer. This week it got more pressing, as OpenAI expanded its ads with conversational sponsored agents and new tools for building campaigns.
Owners are already asking us about it. Our honest read is that it is worth understanding now and, for most small and local businesses, not yet worth moving money into. Here is why, and what to do instead.
This is not search advertising with a new logo
A search ad answers a query. Someone types a few words and you bid to appear next to them. Those words say a great deal about what the person wants, and years of data tell you roughly what a click is worth.
Advertising inside an assistant differs in almost every way that matters. People don't type three words; they hold long conversations, ask follow-up questions and work through a decision. Advertisers don't get to see those conversations — which is right for privacy — so instead of choosing keywords you give the system hints about the context you want to appear in, and trust it to judge. And when an ad can itself hold a conversation, it stops being a placement and starts being a representative of your business that you are not in the room with.
None of that makes it bad. It does mean the instincts built on search ads don't transfer cleanly, and nobody yet has the history of results to say what works.
Before you pay to appear in an assistant's answer, find out what the assistant already says about you for free.
Fix what the assistant says about you first
For most businesses, the bigger opportunity is the one that costs nothing: showing up well in the answers people already get. Ask the assistants your customers use about your business, your service and your part of town. Are you mentioned at all? Is what they say accurate?
Those answers come from what is publicly readable — your website, your business profile, your reviews, what others say about you. If your site is quietly turning AI crawlers away, or your pages don't state plainly what you do and where, a paid placement just papers over a gap you could close for free. It is the same instinct behind the questions we keep hearing from owners around here about editing their site and doing their own SEO with AI: get the foundation right, then decide what to spend on.
If you do decide to test
Some businesses should test early — especially those whose customers genuinely research a purchase in conversation, and who can afford to learn. If that is you, treat it as an experiment rather than a new channel.
Decide what the ad is for before you start. Introducing your business to someone still weighing options is a different job from catching someone ready to buy, and the two should be judged differently. Fund the test from money set aside for learning rather than pulling it from advertising that already works. Write down what success looks like before you see any results, because early numbers from a new format are easy to read generously. And keep it small and time-boxed, so a disappointing month becomes a lesson rather than a loss.
If an ad can hold a conversation on your behalf, read what it says. Check the claims it makes about your prices, availability and guarantees exactly as you would check a new hire's first week on the phones.
Our read, for now
This will matter. Assistants are becoming the place people start a purchase, and paid placement inside them was always going to follow. But it is early, the formats are still settling, and the businesses that benefit most will be the ones already showing up well in the unpaid answers.
So spend this quarter making sure an assistant can read your site and describe your business accurately. Watch how the ad formats develop — exactly the kind of news worth twenty minutes a week — and revisit when there is real evidence. If you would like help getting the foundation right, talk to us.
