In short: We do not just recommend AI advertising, we run it ourselves. Since shortly after ChatGPT Ads opened in Norway, we have had our own campaigns live for both of our products, and the early results are encouraging: Thousands of impressions in relevant conversations, a steady stream of clicks at prices below the US reference levels, and the first booked meetings straight into the calendar. Here we share what we set up, what we have learned, and why we believe those who start now get a head start that will be expensive to catch up on.
Why we jumped in immediately
ChatGPT Ads opened in Norway on August 24, and we had decided long before: We would be among the first advertisers. Partly because we believe in the channel, but mostly because we cannot make a living telling businesses to be early in the AI surfaces while sitting on the fence ourselves. Everything we write in our guides on AI advertising, we wanted to test with our own budget and our own numbers.
The competitive situation made the choice easy. Few advertisers are live, prices are early, and every week in the channel produces learning that does not exist in any guide yet. That window does not stay open for long.
What we set up
The setup follows exactly the recipe from our step by step guide, because that guide is written by the people who actually did the work:
- Separate campaigns per product, with one clear need per ad group, instead of one big campaign trying to cover everything
- Click-based buying from the start, as we recommend in the guide on buying models, with conversion data building up along the way
- Context hints instead of keywords, short descriptions of the conversations we want to be present in
- Dedicated booking pages and full conversion tracking with the OpenAI pixel from day one, so every booking can be traced back to the campaign that created it
The results so far
We will share more thorough numbers once the campaigns have run long enough for them to mean something, but the direction is already clear. The ads have received thousands of impressions in conversations where people are actually evaluating what we offer. The clicks come steadily, at prices below the reference levels from the US launch. And most importantly: The channel has already delivered the first booked meetings, tracked all the way from ad impression to a confirmed slot in the calendar.
What pleases us most is who is arriving. The traffic consists almost entirely of people who did not know us before, exactly the way the channel is meant to work. ChatGPT Ads is a new-customer channel, and it is doing its job.
Four lessons from our own campaigns
- Write context hints as short, natural sentences. Our experience confirms the advice from the context hints guide: What delivers is one to two sentences describing the user's situation or the question they actually ask, with one need per ad group. Single keywords give the system too little to work with, and long essays lock it into rigid phrasings. The sentence in between is the sweet spot.
- Write the ad as an answer, not a billboard. The ad sits below an answer from an advisor, and the copy that behaves like part of the conversation gets the clicks. Recycled slogans from other channels fall flat, exactly as the format suggests.
- Start with clicks, aim for conversions. Click-based buying gave us controlled risk and fast learning, while the pixel collects the conversion data that makes conversion-optimized bidding possible over time. That order, CPC first and oCPC once the data is in place, is in the buying models guide for a reason.
- Tracking from the first krone changes everything. Because every booking can be tied to a campaign and an ad, we optimize on what actually matters instead of staring at click counts. Without that setup, "good results" would be a feeling, not a measurement.
Why starting early matters so much
Everything we learn now, we learn cheaply. Competition for the impressions is low, click prices are early, and every adjustment of context hints and ads builds a bank of experience competitors will have to buy later, at higher prices and with less room for trial and error. We have seen this pattern before: Those who learned search advertising in the early 2000s paid less for their customers for years afterwards. AI advertising is at that same moment now, and that is why we would rather share our experiences early than wait for perfect numbers.
Want to test the channel yourself?
What we have done for ourselves, we now do for our customers too: Account setup, campaign structure, context hints, ads and conversion tracking, plus ongoing optimization. If you are curious what ChatGPT Ads could mean for your business, book a meeting and we will walk through the opportunities together. And if you want to read up first, the pillar guide on AI advertising is the place to start.