Advertising has always followed attention. Search followed intent. Social followed discovery. Retail media followed purchase behaviour. Now, advertising is starting to follow something new: AI-powered decision-making.

Through the OpenAI open beta, advertising within ChatGPT is being opened up to selected brands and partners. For marketers, this represents one of the most significant shifts in digital media since the rise of paid search, because it places brands inside the research moment itself.

At M3.agency, we see this as an important early-stage opportunity for clients who want to understand how consumer discovery is changing, test into AI-native media, and build learnings before the channel becomes more competitive.

Why ChatGPT ads matter

People are already using AI tools to research products, compare services, ask for recommendations, shortlist suppliers and make decisions. That means the journey is no longer limited to traditional search results, review sites, social feeds or comparison pages.

A user might ask:

“What are the best running trainers for beginners?”

“Which CRM is best for a small business?”

“What should I look for when choosing a funeral plan?”

“What supplements help with energy and focus?”

In a traditional search environment, brands compete for visibility through keywords, rankings and sponsored listings. In a conversational AI environment, the opportunity is different. Brands can appear while the user is actively shaping their view, refining options and moving towards a decision.

That makes ChatGPT ads less like passive social discovery and more like a blend of search, display and contextual advertising, with the added benefit of appearing inside a high-intent research moment.

What the open beta currently looks like

This is still an early-stage advertising environment, and expectations need to be set accordingly.

The current beta is focused on contextual relevance. Ads are matched to the user’s prompt, rather than being driven by traditional audience targeting, demographic overlays or advanced geographic expansion.

In simple terms, the better the ad aligns with what someone is asking, the more likely it is to be eligible to appear.

Ads appear below the model response and are clearly labelled as sponsored. The current format includes:

• A 256x256 image

• A headline of up to 30 characters

• A description of up to 60 characters

• A landing page URL with static UTM tracking

This means creative discipline matters. Headlines and descriptions need to be clear, specific and closely aligned to the types of questions users are likely to ask.

Why relevance is the key advantage

In this beta, relevance comes before price.

The ad selection process considers how closely the creative, landing page and supporting keyword inputs match the user’s prompt. While the exact relevance model will continue to evolve, the principle is clear: generic messaging is unlikely to perform as well as tightly matched, contextual creative.

For brands, that means the best approach is not to submit one broad ad and hope it covers everything. Instead, advertisers should create multiple variants around:

• Product categories

• Use cases

• Customer pain points

• Comparison moments

• Problem/solution queries

• High-value keyword themes

• Buyer intent signals

For example, a B2B software brand may need different creative for “best CRM for small businesses”, “CRM for sales teams”, “CRM for customer service”, and “CRM alternatives to spreadsheets”. Each prompt reflects a slightly different mindset, so the ad creative should reflect that too.

What clients should expect from the beta

This is not yet a mature, fully transparent paid media channel so we’re taking caution with brands launching activity.

Reporting is expected to be limited, with delivery focused around impressions and clicks. Dynamic URL macros are not currently available, so tracking relies on static UTM parameters. Optimisation is manual, and share-of-voice style diagnostics are not currently available.

That does not make the opportunity less valuable. It simply means brands should treat this as a learning channel rather than a fully proven acquisition engine.

The best-fit clients are those with:

• Innovation or test-and-learn budgets (estimated £20,000 test budget)

• A clear appetite for early mover advantage

• Strong landing pages and clear propositions

• Multiple product or service use cases

• A willingness to learn while the platform develops

• A desire to understand how AI search may affect future demand generation

How M3.agency would approach ChatGPT ads

For clients, our recommendation is to treat ChatGPT ads as a structured pilot.

The starting point should not be “how quickly can we spend?” It should be “what can we learn that gives us an advantage over the next 6–12 months?”

A sensible test would include:

1. Identify high-intent AI prompts
   Map the questions customers are likely to ask when researching the category, comparing options or looking for recommendations.

2. Group prompts by intent
   Separate awareness-style research from comparison, consideration and conversion-led queries.

3. Build contextual creative variants
   Write headlines and descriptions that directly reflect the user’s likely question, rather than relying on generic brand messaging.

4. Use dedicated landing pages where possible
   Send users to pages that match the prompt theme closely. Relevance between prompt, ad and landing page will be critical.

5. Apply clear UTM tracking
   Use static UTM parameters to isolate ChatGPT ad traffic within analytics and reporting.

6. Review early signals carefully
   Focus on impressions, clicks, click-through behaviour, landing page engagement and any assisted conversion data available.

7. Scale only where relevance is proven
   Double down on the prompts, categories and creative angles that show early traction.

What this means for search marketing

ChatGPT ads should not be seen as a replacement for Google Ads or SEO. Certainly, the scope will increase with future ads in other LLMs too.

However, they are a clear signal that search behaviour is changing. Users are increasingly asking full questions, expecting direct answers and using AI tools to narrow their choices before they ever reach a website.

That creates both a risk and an opportunity.

The risk is that brands continue optimising only for traditional search results while more research shifts into AI environments. The opportunity is that forward-thinking brands can start learning how to show up earlier in the decision-making process.

For SEO, this reinforces the need for clear, authoritative, answer-led content. For paid media, it opens up a new contextual layer where creative relevance, prompt intent and landing page alignment become central to performance.

Our view

ChatGPT ads are not yet a channel every brand should move significant budget into immediately. The beta is early, reporting is limited and the mechanics will continue to change.

But for the right brands, this is exactly the kind of environment worth testing early.

The brands that learn now will be better placed when AI advertising becomes more established, more measurable and more competitive. They will understand which prompts matter, which messages resonate, and how their customers behave when discovery happens inside a conversation rather than a search results page.

At M3.agency, we are already looking at how clients can approach this intelligently: with clear test plans, relevant creative, controlled budgets and realistic expectations. Saying that, we value honesty, and we understand these early stages are about acting as guinea pigs for the platform to evolve. 

The question for brands is not whether ChatGPT ads will be perfect from day one. They won’t be.

The better question is: what can we learn before everyone else gets there?