ChatGPT Ads is here, and over the past month I’ve been testing the platform across a number of different clients and use cases.
As someone who spends a significant amount of time working with Google Ads, Microsoft Ads and Meta Ads, I was particularly interested to see how OpenAI would approach paid advertising – and whether ChatGPT Ads would simply become another version of the platforms we already use.
The short answer is: it feels quite different.
There are parts of ChatGPT Ads that I really like. The platform is refreshingly simple, campaign creation is quick, and the way advertisers can describe their target audience and advertising context feels far more natural than traditional keyword-based advertising.
But it is also very clearly a platform in its early stages.
Conversion tracking, geographic targeting, optimisation tools and advertiser support are all areas where I think there is significant room for improvement.
So, after around a month of experimenting with ChatGPT Ads, here are my first impressions.
ChatGPT Ads Manager is refreshingly simple
One of the first things that struck me was how straightforward the interface is.
Anyone who regularly works inside Google Ads will know how complicated the platform has become. There are campaigns, ad groups, assets, audiences, keywords, match types, bidding strategies, conversion goals and countless settings and recommendations.
There is a huge amount of power there, but it also means creating and managing campaigns properly can become quite technical.
ChatGPT Ads feels very different.
The interface is clean and relatively intuitive, and I was able to start building campaigns without feeling like I needed to spend hours learning where everything was.
For businesses considering advertising on ChatGPT, I think that’s a real positive.
There is far less friction involved in getting started.
Ad group targeting is probably my favourite feature
The area I find particularly interesting is how targeting works at ad group level.
With a traditional Google or Microsoft Ads search campaign, a significant part of the setup process revolves around keyword research.
You identify the searches you want to target, organise keywords into sensible ad groups, decide which match types to use, analyse search volumes and negatives, and then continually refine the campaign based on actual search behaviour.
ChatGPT Ads takes a much more natural approach.
Instead of building your targeting primarily around lists of keywords, you can provide context hints describing things such as:
- What your business offers
- Who the product or service is for
- The problems it solves
- Your main selling points
- The types of conversations where your offer may be relevant
If you already use ChatGPT regularly, the process feels very familiar.
You’re essentially explaining the advertising scenario in normal language.
For example, rather than spending time building extensive keyword lists around a particular professional service, you can describe the service, the intended customer and the situations where that person may need help.
That feels like a much more natural way of building advertising targeting.
It’s also important to understand that these context hints aren’t exactly the same as traditional hard targeting rules. But as a method of communicating the intended relevance of an ad, I really like the approach.
Creating ChatGPT ads is simple too
The simplicity continues when you create the ads themselves.
You don’t currently have the enormous collection of assets and options that have accumulated within platforms such as Google Ads.
Instead, the core ingredients are refreshingly straightforward:
Title. Description. Image.
For advertisers and agencies, that makes it possible to move from an idea to a live campaign relatively quickly.
Whether that simplicity remains as the advertising platform develops is another question.
Google Ads itself started life as a relatively straightforward advertising product before developing into the extremely sophisticated platform we have today.
For now, though, the simplicity of ChatGPT Ads is one of its biggest strengths.
Geographic targeting needs improvement
One area where I’ve found things more frustrating is geographic targeting.
For businesses operating nationally, this may not be particularly problematic.
For local businesses, however, I’ve found the current targeting options less flexible than I would like.
The geographic options available can vary, but during my own campaign setup I’ve found creating the exact local targeting I want less intuitive than it would be within Google or Meta.
For an agency working with businesses that may only operate within a particular town, city or service area, that’s important.
I’d like to see much more granular and intuitive geographic targeting as the platform develops.
It’s certainly something I’d check carefully before launching a campaign for a business where location is critical.
Conversion tracking has been my biggest challenge
This is probably my biggest reservation after my first month.
I’ve implemented conversion tracking across several of the campaigns I’ve been testing, and I don’t yet feel as confident in the data as I would like.
In some cases, I’m seeing GA4 attribute traffic and conversions to ChatGPT advertising, but I’m not necessarily seeing those conversions reflected within ChatGPT Ads Manager in the way I would expect.
That doesn’t automatically mean the ChatGPT Ads conversion tracking is wrong.
There can be perfectly legitimate reasons for discrepancies between analytics and advertising platforms.
Different attribution models, attribution windows, consent settings, browser restrictions, event configuration and reporting delays can all result in two platforms reporting different numbers.
Anyone managing Meta or Google campaigns will already be familiar with attribution discrepancies.
However, at the moment, I’ve found it more difficult to diagnose exactly why those discrepancies are occurring with ChatGPT Ads.
And that’s a problem.
If we’re going to optimise campaigns and make decisions about where clients should invest their advertising budget, we need to have confidence in the conversion data we’re using.
I expect this area to improve significantly as the platform matures, but right now I would be cautious about making major budget decisions based solely on the conversion figures reported within Ads Manager.
Advertiser support could be much better
This leads into another area I’ve found frustrating: support.
I’ve raised a number of support queries while working through tracking and campaign setup issues.
So far, I haven’t found the experience particularly helpful.
Responses can feel automated, and when you’re trying to diagnose a technical tracking problem, it’s difficult to get the level of hands-on assistance you sometimes need.
This is one area where I think established advertising platforms have an advantage.
Google Ads isn’t perfect by any means, but during onboarding and when dealing with certain account issues there is generally a much larger support infrastructure available.
With ChatGPT Ads, I’ve sometimes felt that you’re largely left to figure things out yourself.
For experienced advertisers that may be manageable.
For a small business attempting to set up its first campaign without specialist help, it could become frustrating quite quickly.
There’s a big question around B2B advertising
One of the most interesting questions for me is what ChatGPT Ads ultimately becomes as a B2B advertising platform.
Currently, ads are shown to eligible users on ChatGPT’s Free and Go plans, while users on Plus, Pro, Business, Enterprise and Edu plans don’t see ads.
That raises an interesting question.
What proportion of the most commercially valuable professional ChatGPT audience sits within those paid plans?
I don’t think we can simply conclude that this makes ChatGPT Ads ineffective for B2B.
There will clearly be business owners, decision-makers and professionals using the free version of ChatGPT.
But it potentially limits one of the things that initially makes ChatGPT advertising sound so exciting for B2B advertisers: reaching someone while they’re actively researching a business problem or considering a purchase.
If large numbers of heavy professional users are on ad-free paid plans, that could potentially remove part of that audience from advertisers’ reach.
It will be interesting to see whether OpenAI’s approach changes as the advertising product develops.
So, should businesses start testing ChatGPT Ads?
After a month, I certainly wouldn’t suggest moving large amounts of budget away from established channels and putting it all into ChatGPT Ads.
The platform is simply too new for that.
But I do think there’s a strong argument for controlled experimentation.
If a business already has a healthy paid advertising budget, allocating a small proportion to testing ChatGPT Ads could provide useful early insight.
We’re still at a stage where many competitors won’t have started experimenting with the platform.
That creates an opportunity to learn.
Which products work?
Which services generate engagement?
What types of context hints perform best?
What messaging resonates?
How does ChatGPT traffic behave once it reaches the website?
Those learnings could become increasingly valuable if ChatGPT develops into a significant advertising channel.
The important thing is to treat the budget as an experimental investment, measure what happens carefully and avoid drawing major conclusions from small amounts of early data.
Where I think ChatGPT Ads could get really interesting: ecommerce
One area I’m particularly interested in exploring further is ecommerce and product advertising.
My early testing here has been encouraging.
The nature of ChatGPT means people are frequently asking questions that demonstrate quite specific intent.
They’re researching products, comparing options, looking for recommendations and trying to solve particular problems.
That creates some potentially fascinating advertising opportunities.
Rather than interrupting someone who is scrolling through a social feed, there’s the possibility of reaching them while they’re already engaged in a relevant conversation.
Exactly how effective that becomes at scale remains to be seen.
But it’s the area of ChatGPT Ads I’m particularly interested in testing next.
I’ll be sharing a separate article and video looking specifically at my early experiences of using ChatGPT Ads for ecommerce and product promotion.
My first-month verdict
My overall impression of ChatGPT Ads after the first month is positive – but with some significant caveats.
The good: the interface is simple, campaign creation is quick, context-based ad group setup is genuinely interesting, and creating ads is refreshingly straightforward.
The less good: conversion measurement needs to become easier to trust and diagnose, geographic targeting needs more flexibility, optimisation capabilities are still relatively limited, and advertiser support needs improvement.
It feels like an early-stage advertising product.
And that’s not necessarily a bad thing.
The platform will almost certainly evolve considerably.
For advertisers, the interesting question isn’t whether ChatGPT Ads can replace Google or Meta today.
It’s whether it could become an important part of the paid media mix in the future – and whether learning how it works now gives businesses an advantage as the platform develops.
After my first month of testing, I’m certainly interested enough to keep experimenting.
Thinking about testing ChatGPT Ads?
At CW Digital, we’re already experimenting with ChatGPT Ads across different client types, campaign objectives and use cases.
If you’re interested in advertising on ChatGPT but aren’t sure where to start, get in touch and we can discuss whether it makes sense for your business and help you create your first campaign.
And if you’re already running ChatGPT Ads, I’d be genuinely interested to hear what you’re finding – particularly around conversion tracking, geographic targeting and campaign performance.