
ChatGPT
Ads
Since 24 August 2026, Norwegian advertisers can buy space in the middle of the conversation, where the purchase decision is actually made. ChatGPT Ads looks like search advertising on the surface, but works on a completely different logic underneath. We set up, write and run the campaigns for you.
ChatGPT Ads is OpenAI's advertising platform. The ad appears below the answer ChatGPT gives the user, clearly labelled as sponsored and visually separated from the answer itself. It consists of a headline, a short text, a square image and a link to your landing page.
Testing started in the US on 9 February 2026. On 24 August 2026 the format was rolled out in 31 European markets, Norway included.This means Norwegian advertisers are early in a channel that still has low competition.
This is the most important limitation to understand before you set a budget:
Both Free plan and Go plan users see ads.
Plus, Pro, Business, Enterprise and Edu are ad-free.
Accounts identified as under 18 do not get ads.
Ads are not shown in temporary chats or in the ChatGPT Atlas browser.
Ads are not shown in conversations about physical health, mental health or politics.
The reach is large, but skewed. In practice you reach the part of the market that doesn't pay for ChatGPT. For many B2C categories that is an advantage. For some B2B segments it is worth running the numbers before you move budget.
The difference lies in the signal. In Google you get two to four words. In ChatGPT the user describes their goal, their reservations and their budget in natural language, often over several messages. That gives far richer context, but you cannot target on words.
This does not replace Google. Google captures demand that is already phrased as a search. ChatGPT captures the consideration that happens before anyone searches at all.
The structure is familiar to anyone who has worked in Google Ads, but the content of each level is new.
Goal (impressions or clicks), budget, start and end date, and which countries you want to appear in.
Theme, max bid and context hints. The context hints live here, not at ad level.
Headline, ad text, image and landing page. Several ads per group let you test messages.
SPECIFICATIONS
Title: 16 to 24 characters recommended, 50 characters maximum.
Ad text: 32 to 48 characters recommended, 100 characters maximum.
Image: square, max 1200 x 1200 pixels.
The image URL must be publicly accessible without login, and the landing page must respond.
The recommended lengths are short on purpose. The ad is read in a small space right below an answer the user is already happy with. Write for recognition, not persuasion.
Context hints are where you tell the system which conversations your ad belongs in. They are not keywords. They are natural language describing the problem the customer is trying to solve.
“running shoes, jogging shoes, training shoes, neutral running shoes”
RIGHT
“person training for their first half marathon who needs cushioned shoes for long, easy runs”
This is the single move that separates a campaign that hits from one that burns budget on the wrong conversations. We have written a separate guide on how to do it: How to write context hints that work in ChatGPT Ads.
• The impressions goal is bought on max CPM. The clicks goal is bought on max CPC.
• Conversion-optimised bidding, geo-targeting and custom audiences arrived with the European launch.
OpenAI suggests USD 3 to 5 per click as a place to start. That is a starting point, not a Norwegian market price.
The auction is relevance-weighted and settled on second price. So the highest bid does not automatically win. The quality of the context hints, the ad and the landing page affects both how often you are shown and what you pay.
• OpenAI Pixel and Conversions API for conversion tracking.
• UTM parameters into Piwik PRO, GA4 or whichever analytics tool you already use.
• Ads Manager shows impressions, clicks, spend, CTR, average CPC and average CPM, as well as conversions.
Integrations with third-party measurement tools arrived with the European launch.
Advertisers only get aggregated figures. The conversations are not shared with you as an advertiser, and you cannot see which questions triggered the impression. That makes the landing page and conversion tracking more important than in channels where you can read a search terms report.
1. Discovery. We collect the customer's own language from sales calls, support, reviews and keyword data. That is the raw material for the context hints.
2. Account structure. One theme per ad group, split by need and buying stage, not by your product catalogue.
3. Context hints. Written from scratch, not translated from an existing keyword list.
4. Creative. Headline and text within the recommended lengths, and an image that still works when shown small.
5. Measurement in place before launch, not after.
6. Testing and pruning. We test one theme at a time, so the results can actually be interpreted.
• Your product solves a problem people describe in words before they buy.
• The buying journey includes research and comparison.
• You have landing pages that answer the need, not just the product name.
• Your sales are purely local and impulse-driven.
• You don't have conversion tracking in place. Without measurement this is a cost, not a test.
• Your budget is already too small to cover the demand you capture in Google.
At the European launch, buying went through OpenAI's Ads Solutions team, agency partners and technology partners, with self-serve in Ads Manager announced shortly after. We set up and run campaigns whichever route is open when you start.
No. Context hints are signals, not rules. OpenAI is clear that they are neither exact match, audience rules nor instructions to appear only in specific conversations.
No. You narrow the scope by describing precisely what the ad group is actually about. Vague context hints are the most common reason ads show up in the wrong conversations.
No. Google captures demand that is already phrased as a search. ChatGPT Ads captures the part of the consideration phase that never becomes a search. The two overlap less than you think.
Enough to get statistically readable figures on two to four ad groups within a month. We calculate it specifically based on category and bids, and tell you if the numbers don't add up.
Want to be early in a channel where competition is still low? We map the potential, set up the account, write the context hints and have measurement in place from day one.
Get in touch for a review of what ChatGPT Ads could mean for your category.