ChatGPT Ads step by step: How to set up your first campaign in Ads Manager

Before you start: What you need

ChatGPT Ads launched in Norway on 24 August, and setup goes quickly when you have four things ready in advance:

• Your legal company name, exactly as registered and as shown on your website
• Your organisation number and a payment card for billing
• A square logo of at least 256 by 256 pixels, since it appears as the icon in the ad itself
• A landing page relevant to what you are advertising

Check two more things before spending time on registration. First, that your industry can actually advertise: OpenAI's prohibited lists include alcohol, gambling and a range of financial services. Second, if you work with an agency: The account should be created by you as the business, not by the agency. The agency is invited in as a user afterwards, because there is no agency structure yet like Google and Meta have.

Step 1: Create your account and get verified

Go to ads.openai.com and sign in with an OpenAI account tied to your work email, or create a new one. Registration has three parts:

1. Business Details. Company name, website, logo and industry category. A small but important tip: Use exactly the same company name as on your website. Mismatches here are one of the most common reasons applications get rejected.
2. Account Details. Country, currency and time zone, plus accepting the terms.
3. Identity verification. Run by a third party, confirming that the business and the person creating the account are real. Approval can take anywhere from the same day to a couple of weeks depending on industry, so do this well before you plan your first campaign.

Once approved, two settings remain before ads can serve at all: Under Settings and Account info you add the account name as it should appear in ads, plus the logo. Under Settings and Billing you add the billing address and payment card. Consider pointing the invoice email to your finance team's inbox rather than your own.

If several people will work in the account, invite them under Users. The roles are clearly split: Owner has full access, Marketer can build campaigns but not touch billing, Analyst sees reports only, and Finance sees financials only.

Step 2: Understand the structure

Ads Manager uses a structure you will recognise from Google and Meta, with three levels:

• The campaign defines objective, budget, dates and geographic targeting
• The ad group collects ads around a theme or intent area, and this is where context hints live
• The ad is the creative itself: Title, copy, image and landing page

You can build campaigns in the guided flow in the interface, or upload many campaigns at once with a completed schema template if you are launching at scale. For your first campaign I recommend the guided flow, it forces you through every choice.

Step 3: Set up the campaign

Click Create campaign and give it a name that says what it is, for example "Demo - Norway - Clicks". Then make the important choices:

Objective. You have three options: Reach, where you pay per thousand impressions, Clicks, where you pay per click, and Conversions, where the system optimises towards conversions you have defined. Start with Clicks. Conversions requires tracking to be in place first, and we get to that in step 6.

Budget and dates. You can set a daily budget or a total for the whole campaign. In the US, daily budgets have started at around 25 dollars, so the threshold for testing is low. Choose a start date, optionally an end date, and remember you can always pause or adjust along the way.

Geography and platform. Choose Norway. Note that reporting currently happens at country level, so if you want to compare regions, create separate campaigns. Optionally you can also narrow by platform, meaning iOS, Android or web.

Customer lists. You can upload your own lists to include or exclude audiences, but they require at least 25,000 matched users. For most Norwegian businesses, skipping this at the start is the practical choice.

Step 4: The ad group and context hints

Now comes the part that is most different, and honestly the most important in this whole guide.

In the ad group you first set a max bid. OpenAI recommends starting at 3 to 5 dollars per click, and if you set the bid below roughly 3 dollars, you get a warning that the ad may not serve. That is the floor in practice.

Then you write context hints. Since there are no keywords in ChatGPT Ads, this is the most important lever you have for steering where the ad appears: Free-text descriptions of which conversations, topics and questions your ad is relevant for. A good rule of thumb is 5 to 15 hints per ad group. For our own part, we write hints like:

• "how to become visible in ChatGPT and AI search"
• "business evaluating AI visibility tools"
• "marketer wondering how AI describes their brand"
• "how to advertise in ChatGPT"

Notice that these are not keywords in the classic sense, but descriptions of conversations. The best advice from early B2B advertisers is to describe the user's situation rather than your product category: "team struggling to track deliverables" hits better than "project management software". The US experience is otherwise split: Short, keyword-like hints have delivered more often and cheaper, while full questions seem to reach slightly different conversations. Test both in separate ad groups and let the numbers decide. And remember that hints guide the system, they are not exact rules, so do not be surprised if the ad shows up in conversations you had not predicted.

Step 5: Create the ad itself

The format is simple: A title of up to 50 characters, a description of up to 100 characters, a square image of at least 256 by 256 pixels, and a landing page. In addition, the company name and logo from your account settings are shown. Here is ours: The title is "Get visible in AI answers", the description is "We map and improve how ChatGPT describes your business. Book a free demo", the image is our logo on a clean background, and the landing page is the demo page.

OpenAI's own guidelines for the creatives boil down to four principles worth taking seriously:

• Build for coverage, not one message. Many variations of title and copy per offering give the system more chances to find relevant conversations. Each variation should take a new angle, not repeat the same message in other words.
• Write like a helpful answer, not a billboard. The ad sits below an answer from an adviser. Concrete value beats slogans: "Cut reporting from three days to three hours" wins over "Transform your business".
• Link to the most relevant page, a product or topic page rather than the homepage, and add UTM parameters to the link, for example utm_source=chatgpt and utm_medium=paid. That lets you follow the traffic in Google Analytics independently of OpenAI's reporting.
• The landing page must be technically open. OpenAI requires that the page does not block their bots, OAI-AdsBot and OAI-SearchBot. If you have followed our guide on technical SEO for AI search, this is already in order.

A practical time-saver: Ads Manager can suggest ready-made ad drafts based on your website's metadata, with image, title and description you can edit and use. A good starting point, but polish the copy yourself.

Step 6: Set up measurement

Before publishing, get tracking in place. OpenAI has its own pixel you add to your website, where you define events like form submissions, booked demos or purchases, preferably with value. For those who want to send events directly from their own systems, there is also a Conversions API, plus third-party integrations. Once the pixel is live, you can both see conversions in reports and switch the campaign objective to Conversions, so the system optimises towards what actually matters.

The reports give you impressions, clicks, spend, CTR, average cost per click and conversions, at campaign, group and ad level, as tables, charts over time or CSV export. What you do not get is insight into which conversations triggered the ad. That is a black box for now, and exactly why your own UTM tracking matters so much: It lets you compare the channel directly with Google Ads and Meta in your own numbers.

Step 7: Publish and monitor

Review the summary and publish. The ad goes through approval before it serves, and afterwards the status field is your friend: If an ad shows as "Not serving", hover over the status to see the reason. Much is solved by editing and resubmitting, while some rejections are due to policy or the account missing verification or billing details.

Five pieces of advice from those already running

1. Start small. With daily budgets of a few hundred kroner you learn a lot without much risk, and prices are still low because few Norwegian advertisers are active.
2. Test systematically. Run different context hints and ad variants in separate groups, because there is no playbook yet. Those experimenting now build the knowledge competitors will have to buy later.
3. Treat the first weeks as learning. Experienced B2B advertisers recommend viewing the first 60 to 90 days as a learning investment, with enough budget for the numbers to actually mean something.
4. Judge the channel on its own terms. Results vary wildly: Some industries report conversion rates beating Google Ads, while other tests have shown shorter visits and far weaker conversion than search traffic. Give the channel its own goals and a few weeks before concluding.
5. Check the match. Low CTR suggests your hints hit the wrong conversations, while many clicks with poor engagement suggest the landing page does not deliver what the ad promised. Both are fixable, but only if you look for them.

Finally: Ads are half the job

Ads buy you into the conversation, but they do not replace ChatGPT actually knowing and recommending your brand when someone asks for advice. The organic recommendation and the paid placement reinforce each other, just as SEO and Google Ads have done for twenty years. If you want to know how the AI models describe your business today, that mapping is what we do for a living. And if you are also getting started on the paid side of the Google ecosystem, we have written a matching step by step guide for Google Ads, including the AI ads on their way there. And if you want the whole paid AI landscape in one place, from the surfaces and prices to the measurement, our pillar guide on AI advertising is the place to start.