Q&A From Winning Strategies for AI Ads, ChatGPT Ads & the Next Era of Paid Search

Our webinar, Winning Strategies for AI Ads, ChatGPT Ads & the Next Era of Paid Search, garnered interest from more than 270 marketers and paid media professionals interested in how AI is actually changing advertising and what they should be doing about it.

Brafton’s Director of Paid Media, Zach Shah, joined by Chief Services Officer Dave Snyder, discussed ads in ChatGPT, new formats in Google AI Mode, and the growing role of automation in paid search.

The audience brought questions that got to the practical decisions behind those headlines. Here are our answers!

1. Does Giving Algorithms More Control Create Risk? Where Should Paid Media Teams Focus Instead?

Zach called out a noticeable pattern during the session: Through each paid search era, advertisers have handed algorithms more control; first with Smart Bidding in 2018, then Performance Max in 2021, and now with ad creative generated per query directly within AI answers.

But does it create any real risk? Yes, some of which we already know about. An algorithm can optimize toward the goal you give it, but it cannot decide whether that goal reflects what your business actually values. If the conversion signal is a form fill, it may find more form fills even when few turn into qualified opportunities.

But for AI-generated ad creative — the latest control advertisers are starting to hand over — there’s a new risk: AI-generated creative could repeat or amplify incorrect information about your business. 

As platforms handle more bidding, matching and creative assembly, paid media specialists should spend more time on the inputs and the results:

• Defining a valuable conversion.
• Connecting qualified leads or sales back to campaigns.
• Keeping product and site information accurate.
• Testing the offer and landing page.

Automation can make a good strategy more efficient … but it can also make a poorly defined one more efficient, too.

2. Could Other LLM Platforms Challenge Google for Advertising Spend?

Google is such a massive player that it’s difficult to say for sure. Google’s total revenue last year was just north of US$400 billion, but network advertising revenue dropped 2%.

By comparison, AI-specific ad spend in 2026 is estimated at US$32 billion, with about $1 billion coming from ChatGPT.

Google has an established search advertising business, mature buying and measurement tools, and access to commercial queries at scale.

LLMs like ChatGPT offer something a bit different: Someone may be working through a problem, comparing options or narrowing a decision over several turns. That context could make a relevant ad useful even when the person never typed a conventional search keyword.

The buying tools are developing. OpenAI has introduced CPC bidding, a beta self-serve Ads Manager and conversion measurement, while continuing to describe ChatGPT advertising as an evolving program. Those capabilities make testing more practical, but they do not establish how much spend will ultimately shift from Google.

Our view: Treat LLM advertising as a channel to evaluate against a specific objective, rather than moving budget because the audience is growing. Keep funding proven search activity, then run a contained test where your product fits the kinds of decisions people make in a conversation. Then, compare lead quality, sales and incremental reach with your existing channels.

3. What Matters Most for a Busy Team With Limited Resources?

If time, budget and bandwidth are limited, start with work that improves campaigns you already run. In most cases, this work will also help you prepare for newer ad formats if and when you decide to run a test or start a new campaign:

1. Check your conversion signals. Make sure campaigns can distinguish a valuable action from a low-quality lead. If possible, connect CRM outcomes or sales data to advertising decisions.
2. Make your offer easy to understand. Review key product or service pages, pricing and eligibility details, landing pages and, where relevant, product feeds. AI systems increasingly use those inputs to understand what you sell and build relevant ads.
3. Choose one measured test. Pick a campaign, audience need or product with a clear success metric. Set a budget and review date before trying a new format.

These actions are far more useful than opening an account on every new ad platform. Google’s newer AI Mode ad formats draw on advertiser creative and product or service information to explain relevance; improving that source material can support both current campaigns and future tests.

4. Will Someone Respond Differently to an Ad in an LLM Conversation Than to One Beside a Search Result?

Probably, but we would treat that as a hypothesis to test, not an established performance rule.

A short search often communicates intent through a few words. A conversation may reveal the constraints behind a decision: budget, intended use, timing or the features someone is weighing. An ad that speaks to one of those needs could feel more timely. It could also feel intrusive if it interrupts a sensitive discussion or looks too much like part of the answer.

There is an important distinction to make in the current experience: ChatGPT ads appear below a response, clearly labeled and separate from the answer. Advertisers cannot pay to change what ChatGPT says or see a user’s private conversation. Google is also testing AI Mode formats that present sponsored options with AI-generated explanations. These experiences should not be lumped together as though every ad sits “inside” an AI answer.

For advertisers, the question is whether the placement helps the person take a useful next step. Watch what happens after the click and how those visitors progress, rather than assuming a conversational placement will behave like a standard search ad.

5. How Should Advertisers Develop Context Hints: Search Terms, Customer Research or Sales Conversations?

Speaking about the actual mechanics of ChatGPT ads and targeting, Zach explained that, with the options currently available, advertisers cannot target specific keywords with their ChatGPT ads. Instead, you supply “context hints” that OpenAI matches thematically against the conversation.

“The craft here is describing the situation a person is in, not the phrase they would have typed,” said Zach.

To develop those context hints, advertisers should consider all three inputs — search terms, customer research and sales conversations — then check them against what customers actually buy.

• Search terms show the language people use.
• Customer research explains their needs and objections.
• Sales and support conversations reveal the situations that lead to a serious inquiry and the details prospects need before choosing.

The goal is to describe a real situation in which your offering helps, not to paste a list of keywords into a new field. Suppose you sell accounting software to retailers. “Growing retailers that need to reconcile online and in-store sales” gives the system more to work with than “accounting software, retail bookkeeping, POS.”

OpenAI describes context hints as ad-group-level information about what you offer, whom it helps and when it is useful. They inform relevance; they are not targeting rules or guarantees that an ad will appear in a particular conversation. Start with a few specific, accurate situations tied to a product or service. Review performance and revise the hints as you learn which needs lead to qualified outcomes.

6. How Do You Evaluate Creative When Gemini May Generate a Different Message for Every Query?

This is part of that handing-things-over-to-algorithms pattern we touched on earlier. 

If exact wording changes frequently, approving one perfect headline or trying to attribute results to a single sentence becomes less useful. So, what should you evaluate?

First, review the inputs and outputs. Are your source pages, product details, approved assets and offers accurate? Does the generated message describe the right benefit and take people to a relevant page? Spot-check examples across different search intents and flag claims or combinations your brand would not make.

Second, review business performance. Compare a controlled test with your existing campaign approach, using conversions, lead quality, revenue or another outcome that matches your goal. Look for patterns by query theme, asset and landing page. Google says AI Max reporting can show combinations of search terms, headlines and URLs, as well as performance data for optimized assets.

That will not answer every question about a generated ad, but it can help you judge whether the creative system is representing your offer well and improving the result you care about.

7. Do AI Ads Make It Harder for Less Established Brands, or Level the Playing Field?

This is an important question. We’ve heard so much about how AI in general has leveled the playing field for smaller businesses when it comes to things like marketing, but what about paid media specifically? 

The answer isn’t as clear as you’d maybe hope, as both forces are at work in their own ways.

Established brands may have more budget, stronger recognition and richer data to feed AI ad systems. Whereas a smaller advertiser with thinner product information or not-as-substantial conversion tracking could struggle to show its relevance, even if its offer is excellent.

But AI ad systems can also create openings for a less familiar brand that solves a specific problem especially well. For example, if a person describes a particular use case, a clear product page, credible proof and a strong match to that need may matter more than a broad message written for everyone. Smaller teams may also be able to update an offer, page or campaign more quickly when they see what works.

Still, agility alone is not a substitute for trust. If you’re a smaller brand interested in running AI ads, make it easy to verify what you sell, who it serves, what it costs and why someone should choose you.

What Should Paid Media Teams Take Away?

In short, the most useful next step is to improve what you can control: the quality of your conversion data, the clarity of your offer and the information AI systems draw from. With that foundation, it’ll be easier to test new placements against the same business outcomes you expect from any other paid channel.

For a closer look at the formats and decisions behind these questions, watch Winning Strategies for AI Ads, ChatGPT Ads & the Next Era of Paid Search. And if you’d like help deciding what to test first, get in touch with Brafton.