People used to type “best running shoes” into a search bar and scroll through link after link. Now they open ChatGPT and say, “I’m training for a half marathon, I overpronate, and I have wide feet. What should I buy?” That second version tells a brand almost everything it needs to know about the shopper. And for the first time, brands can show up inside that conversation.
OpenAI launched ads in ChatGPT earlier this year, and the early campaigns are already teaching us things that don’t match what we expected. This guide covers what ChatGPT Ads are, how they work, what the first wave of results is showing, and how to get started, with learnings from a webinar Pacvue hosted with OpenAI and Kepler.
Why Conversational Commerce Changes the Game
People Are Researching and Buying Inside AI
AI adoption is moving faster than any consumer technology in history. ChatGPT has reached 900 million weekly active users, and the curve is still climbing. For context, AI is being adopted roughly 10 times faster than the PC and 5 times faster than the internet.
That shift is already changing how people shop. In Pacvue’s 2026 Funnel Rewired report, a survey of 1,008 US consumers, 53% said they use AI tools to research products and 28% now turn to AI for shopping research every day.
OpenAI’s own data shows about 20% of ChatGPT conversations carry shopping intent, spanning categories like retail, home, beauty, travel, cooking, auto, electronics, and fitness. That’s a lot of buying decisions happening in a place most brands aren’t present yet.
From Keywords to Context: Richer Intent Signals
A typical Google search runs three to four words. A ChatGPT prompt is a paragraph.
Think about someone shopping for a college dorm. The old signal was a keyword: “college dorm essentials.” The new signal is a full brief. “My son is starting college in the fall. He’s living in a small dorm in the Northeast, and I want to make sure he has what he needs without overbuying.” Now you know the objective, the constraints, the context, and the tradeoffs the shopper is weighing.
What ChatGPT Ads Are and How They Work
How an Ad Appears and Where It Runs
An ad in ChatGPT shows up as a sponsored card below the model’s response. It’s clearly labeled “Sponsored” with a clickable favicon, and it includes a headline, a short description, and a 1×1 square image you provide. OpenAI recommends keeping the headline around 16 characters and the description around 32, so the message has to be tight.
Ads never interrupt an answer or change it. OpenAI has been firm on this through its ad principles: answer independence, conversation privacy, choice and control, and long-term user value. Ads only appear where they’re appropriate and safe, filtered out of sensitive content. Users can tap “About this ad” to see why they’re seeing it and what gets shared with advertisers, which includes broad performance stats, not their chats.
Right now, ads are delivered to users 18 and older across Free and Go plans in the US, Canada, Australia, New Zealand, the UK, Japan, and South Korea.
How Targeting and Context Hints Work
ChatGPT Ads don’t use demographic or behavioral targeting. They’re matched to the conversation itself. The system reads three things: your ad copy, your landing page, and your context hints.
Context hints are where you describe the conversations you want to show up in. They live at the campaign level and act like a brief for the matching engine. You’re telling ChatGPT Ads your objective and theme in plain language, not stuffing a keyword list. One clear objective per campaign works better than five competing ones. Everything below it, your ad groups and individual ads, should stay aligned to that theme.
Your ad copy carries weight, too. The headline and description are part of how the system decides which prompts your ad fits, so conversational copy that answers a question tends to match better than a promotional line about a sale.
Buying, Bidding, and Measurement
You can buy on CPM for reach or CPC for clicks. CPC launched recently and runs on a second-price auction, so if you bid $20 and the next competitor bids $10, you pay just above their bid, not your max.
Measurement is still maturing. Ads Manager Beta (aka OpenAI’s Ad Manager) shares impressions and clicks today, with more metrics on the way. You can track with static UTM parameters on your landing URLs, and you can measure conversion impact through the pixel and Conversions API.
What’s Working for ChatGPT Ads
Creative and Copy
Lead with a clean visual. Images beat no images, but the 1×1 format is small, so skip text inside the image and use recognizable branding.
Write copy like you’re answering a question, not pitching a product. Promotional lines like “50% off this week” tend to fall flat in a conversation. Helpful, specific copy that matches what someone is actually asking matches better and earns more clicks.
Context Hints and Ad Group Structure
Use plenty of context hints, and organize them by theme. A furniture brand might split ad groups into living room furniture, small-space couches, and sectionals, then watch which conversations deliver. Separating brand from non-brand also helps, since you can tailor copy and landing pages to each.
If you’re struggling to scale, pull your context hints higher in the funnel. Brands that started too narrow on branded terms found more room to grow by moving into broader, non-branded conversations where the demand actually lives.
Landing Pages, Crawlability, and Tracking
Check crawlability before you launch. If OAI-SearchBot or ChatGPT-User are blocked from your landing pages, your ads won’t serve at all. Seeing zero impressions? Your robots.txt is the first place to look.
Brands are testing sending traffic to Amazon, Walmart, Instacart, and their own DTC sites to see what converts best. Right now, the best way to track performance is with UTMs at the campaign, ad group, and ad level, so you can tie performance back when fuller attribution arrives.
How to Get Started with ChatGPT Ads
Plan, Launch, and Manage in One Platform
Activating a new channel usually means another login, another report to reconcile, and another silo. Pacvue handles ChatGPT Ads inside the same platform you already use for Amazon, Walmart, Target, and 100+ other channels, so it feels native from setup to reporting.
That matters for a few reasons. You can extend Budget Manager, Dayparting Scheduler, Ads Rules, and Commerce Rules to ChatGPT Ads, the same automation layer running the rest of your retail media. You can make strategy calls using retail performance signals, which connects your conversational AI spend directly to commerce outcomes. And you can review ChatGPT Ads next to your full portfolio in one view, with AI-driven recommendations across every channel.
How Much to Budget
Most brands are starting with a budget of $10K to $25K per month. That’s enough to baseline delivery, test creative, and learn how your context hints perform before you decide how much to invest. Some advertisers are already spending hundreds of thousands, but the smart play is a focused test budget first, then iteration.
Treat the budget as evergreen for as long as you can. The matching engine needs runway to learn, similar to the warm-up period on any new search platform. Keep it live, feed it creative, and refine your hints as you go.
The Future of AI-Native Advertising
ChatGPT Ads are early, but that’s the point. The brands learning this channel now, with real budget and real creative discipline, are building an advantage that compounds. Conversational AI is where a growing share of shopping decisions are already happening, and the path from prompt to purchase keeps getting shorter. Showing up in those moments, with the right message, is a strategy worth testing before everyone else catches on.
We went deep on all of this with OpenAI and Kepler in a recent session, including the full pet brand and sports nutrition results, a live walkthrough of campaign setup, and a Q&A on budgets, verticals, and what’s coming next. Watch the on-demand webinar to see the data and the platform in action.
