Planning your first ChatGPT advertising test

If your business is interested in ChatGPT advertising and you want to understand what a worthwhile test involves, our Head of Paid Media, Richard Jackson, shares his expertise in this helpful guide. 

To Richard, a useful test means a clear purpose, sufficient investment and a way to analyse what the results really mean. 

Read our introductory guide if you’re still deciding whether ChatGPT ads are right for you and your business: Are ChatGPT ads worth exploring for your brand?

What should your first ChatGPT advertising test tell you?

When planning your first ChatGPT advertising test, consider awareness, consideration, and prospecting. Set realistic expectations from the get-go, rather than promising a specific outcome and expecting perfect results. 

If your business needs immediate, proven sales returns, a test may be harder to justify, and you need to distinguish between selecting a conversion objective and having enough evidence to confidently forecast sales. 

Questions your test could answer 

• Can the brand reach a relevant audience during research? 

• Can that initial interest translate into meaningful website traffic? 

• Is there enough evidence to justify further investment?

At Fusion, we can help you define the specific questions the test should answer for your brand, and set out how we measure that. This then shapes the campaign objective and the evidence we need to assess it properly. 

What investment does a meaningful test need?

I recommend starting with a £10,000 monthly budget as a comfortable starting point, or around £5000 for a smaller, more limited two-weekly test. 

You’d need a narrower scope with a test budget of £5000, with limited capability to compare different audiences or creative approaches.  

A £10,000 budget gives you more room to compare your audiences and is a more comfortable recommendation for those who want to truly see the impact the channel can make. The campaign also needs the opportunity to generate useful evidence, so a monthly test is ideal if you can. 

How we plan ChatGPT advertising tests at Fusion 

At Fusion, we scope advertising tests around your budget, audience and unique objectives.

There’s a lot of judgement involved in deciding what to prioritise and what the available investment can realistically achieve, and we’re here to advise you on the best route to get the most out of your budget. 

We’ll also be able to advise you if the budget you’ve got doesn’t support a test, even if it’s something we recommend. 

What goes into planning a useful ChatGPT advertising test?

To plan a meaningful test that gives you solid answers, you first need to understand your audience, where and how they research products, and how they interact with AI in general. 

Identifying the audience and research moments that matter

• The opportunity to advertise on ChatGPT depends on whether your audience uses conversational AI. Existing AI-referral traffic can pose as an additional signal to investigate.

• Use audience research, such as GWI, to assess relevant behaviour.

• It’s important to understand the conversations where a product or service could be relevant.

Take the travel and tourism industry as an example: potential customers may use AI to research destinations, compare holiday types, and narrow down options. 

In this case, we would research contextual themes to inform the testing approach and check targeting availability during planning. 

At Fusion, we combine real audience evidence, your brand's offering, and behaviour research to identify the most suitable campaign focus, going beyond simply transferring an existing paid search keyword list to a new platform. 

Choosing a focused targeting and creative approach

It’s also important to consider test size before getting into the details. We recommend keeping a smaller test more focused on one specific audience, as testing several audiences at once can spread your budget too thin and make findings harder to interpret. 

For creative, comparing advertiser-controlled and AI-generated creative can help you determine which lands best with your audience. If your brand has limited creative resources, AI-generated creative can be a good option, as long as quality and branding stay consistent. 

Agreeing how success will be assessed

First, as with any paid campaign, measurement should reflect the question and objective agreed at the start of the project. 

Ensure you distinguish beforehand between evidence of reach, website interest and sales, and use tracking parameters and conversion events as a starting point for discussing and planning measurement KPIs. 

However, remember that selecting a measurable objective doesn’t automatically define the campaign's wider commercial value, and you should still consider that.

Ultimately, the most valuable thing you can do before activity begins is agree on what the test needs to teach you. 

At Fusion, we help brands establish an appropriate measurement approach from the beginning, and interpret results based on the campaign's purpose and limitations. Reporting on the testing will then inform decisions about future activity and help shape strategy. 

For an initial test, I would recommend prioritising the following metrics: 

• CTR (Click-Through Rate) 

• CPC (Cost Per Click)

• CPM (Cost Per Mille) 

Because these campaigns are not lead gen and are more for awareness, and there’s no guarantee they’re going to return anything, we wouldn’t recommend tracking cost per conversion.

How do you decide what happens after the test?

After the test, use initial performance and audience findings to inform next steps and further recommendations. The possible next steps could be: 

• Refine the approach 

• Extend activity for a longer period 

• Explore another audience 

• Stop activity if the test identifies that ChatGPT advertising is not right for your business 

Remember, a small test should be interpreted within its scope, and it won’t necessarily answer every question about the channel or how your audience interacts with it. 

You’ll need to consider audience relevance, traffic quality and outcomes together. As the platform develops, this could create further testing opportunities, but you should always assess them against the brand’s objectives. 

How to decide whether to continue, adjust or stop the test 

At Fusion, we turn campaign findings into a solid recommendation. There’s a lot of judgement involved in identifying what deserves further investment and what needs to change, and we have years of experience in this at Fusion. 

We’ll use the testing review as a useful commercial decision point, rather than just a campaign results report. If you’re seeing a halo effect on other campaigns and brand awareness is visibly improving, accompanied by a strong CTR and increased website traffic, we would recommend continuing. 

If you don’t see these results, you may be targeting the wrong audience, which may mean we need to do more research and adjust the campaign, or stop altogether if the platform isn't suitable for your brand and audience.

Plan your first ChatGPT advertising test with Fusion

We can help you assess whether the channel fits the brand’s audience, objectives and available budget. 

To get started, we’ll conduct a strategic assessment, then design and launch the campaign. If the opportunity still needs further investigation to gauge its impact, we can provide an AI audit to assess how your brand is currently positioned and discussed on the platform before planning an advertising test. 

Get in touch to discuss your goals and budget, and we can help you decide whether a ChatGPT advertising test is the right route and advise what the project could potentially deliver.