At Google Marketing Live on May 20, 2026, Google introduced two ad formats built for AI Mode, its conversational search experience: Conversational Discovery ads and Highlighted Answers. Both are powered by Gemini and place ads directly inside AI-generated answers, which Google framed as “a new generation of ads for the AI era of Search”. For B2B SaaS teams that have watched AI Overviews erode click-through on informational queries, this is the other side of the coin: paid placement inside the AI answer itself, rather than blue links pushed below it.
The important context is timing. As of September 2026 both formats are in testing in the United States on mobile and desktop, with no confirmed general availability date and no wider geographic scope. This is not a deadline to act on; it is a shift to understand and prepare for. This guide explains what each format is, how they differ, why they matter specifically for considered B2B SaaS purchases, how to think about measurement when the click happens inside a conversation, and what to do now so your account is ready when the format opens up.
What Conversational Discovery ads actually are
Conversational Discovery ads are ads that answer a user's specific question directly inside AI Mode, with creative that Gemini generates dynamically to fit the conversation. Google's own example is someone looking for ways to make their home smell like a spa, who then sees tailored creative highlighting a relevant product's features. The mechanism is a departure from how paid search has worked for two decades: instead of you writing headlines and descriptions that Google assembles into a responsive search ad, the model reads the intent behind a free-form question and produces the ad copy on the fly. Your role shifts from writing the ad to supplying the raw material, your products, positioning, and business information, that the model draws on.
That is a meaningful change in control. In a responsive search ad, you still own every headline that can appear. In a Conversational Discovery ad, Gemini generates the creative, and your leverage is upstream in the accuracy and clarity of the inputs. For B2B SaaS, where a misstated capability or a vague value proposition can actively repel a technical buyer, that raises the stakes on getting your product information and messaging exactly right, because the model will paraphrase whatever you give it into an answer you did not write word for word.
How Highlighted Answers differ
Highlighted Answers trigger in a different moment. When AI Mode responds to a query by producing a list of recommendations, for example the best tools for a particular job, a highly relevant, high-quality ad is eligible to appear as one of the items on that list, labeled as a Highlighted Answer. Where a Conversational Discovery ad responds to a single specific question, a Highlighted Answer inserts your product into a recommendation set the user is already scanning. Both carry a Sponsored label and, per Google, include an independent Gemini-written explainer alongside the creative so the user can see why the ad is relevant to their query.
For B2B SaaS this list placement is the more strategically interesting of the two, because so much of the software buying journey runs through exactly these comparison and shortlist moments. A buyer asking AI Mode for the best options in a category is doing the same work they would do reading a review site or a software directory like G2 or Capterra, and a Highlighted Answer is a paid route into that consideration set. The catch is Google's emphasis on “highly relevant, high-quality” eligibility, which suggests the format will favor products with strong relevance signals and clean data over those simply willing to bid, though the exact controls have not been detailed.
Why this matters more for considered SaaS purchases
AI Mode ads matter more for B2B SaaS than for impulse retail because SaaS buyers use conversational search to do genuine research, not just to find a product to buy in one click. A prospect evaluating, say, an incident management tool will ask AI Mode multi-part questions about integrations, compliance, and pricing models, exactly the kind of considered, specification-heavy queries where a dynamically generated answer that addresses their real question can carry more weight than a generic search ad. The format meets the buyer inside the research conversation rather than after it, which for a long, multi-touch journey is early and potentially influential.
It also connects to a problem this blog has covered from the organic side. As AI answers absorb more of the query, the classic worry is falling click-through, which we examined in our piece on how AI Overviews affect SaaS paid CTR and the risk of AI answers contradicting your paid ads. Conversational Discovery ads and Highlighted Answers are Google's commercial response to that shift: if the answer is where attention now lives, paid placement moves into the answer. For SaaS marketers that means the AI surface is no longer purely a threat to manage but also a channel to plan for, alongside the existing ads that already appear in AI Mode.
Measuring a click that happens inside a conversation
The measurement question is the one B2B SaaS teams should think hardest about, because it is where these formats are genuinely different. A click from a Conversational Discovery ad still behaves like any paid click: it carries a GCLID, lands on your site, and can be tracked with standard conversion tracking, enhanced conversions, and offline conversion imports from your CRM. On that level nothing changes, and your existing stack still works. The complication is that the format is designed to influence a conversation, and much of that influence will not resolve in a last click.
A prospect may read a Gemini-generated answer that features your product, absorb the positioning, close the tab, and convert two weeks later through a branded search or a direct visit that your reporting credits to another channel entirely. That is the same attribution gap that dogs all upper-funnel activity, and the discipline is the same: keep SQL and pipeline as your accountability metrics, not clicks or assisted conversions from the AI surface. Watch branded-search and assisted-conversion trends as directional evidence that the format is doing upper-funnel work, but hold spend accountable to pipeline. As always, the safest posture is to optimize toward SQLs, not raw leads, so bidding never learns to chase a cheap conversational click that never becomes revenue.
What to do now, while the format is still in test
Because both formats draw on your existing campaigns, assets, and product information rather than a new standalone campaign type, the preparation is not to build something new but to strengthen what the model will feed on. The accounts that perform when AI Mode ads open up will be the ones already running well-structured Search and Performance Max campaigns, with clean conversion data and precise messaging. Concretely: tighten your value proposition so Gemini generates from accurate material, keep product feeds and business information current, invest in offline conversion tracking so bidding optimizes toward real pipeline, and make sure landing pages directly answer the specific questions buyers ask.
None of that work is wasted if the format is delayed or never reaches your market, which is the whole point of treating this as preparation rather than a launch. Every one of those improvements raises performance on the campaigns you already run. It is also worth revisiting your campaign structure by funnel and intent tier, because a format that inserts you into research conversations sits naturally in the consideration layer, and an account organized around clear intent tiers will absorb it more cleanly than one that lumps all intent into a single campaign. Prepare the foundation; do not manufacture urgency around a date Google has not published.
Cautions and open questions
Several things about these formats remain genuinely unknown, and honest planning means naming them. Google has not confirmed a general rollout date, whether the formats will expand beyond the United States, or the precise controls advertisers will get, such as whether you can opt out, set separate bids, or see isolated reporting for AI Mode placements. It is also unclear how much creative oversight you will retain when Gemini generates the ad, which for regulated B2B SaaS categories like fintech or healthcare is not a small question. Until Google details those controls, any confident claim about how to run these ads is speculation.
The steadier position is to separate the durable signal from the uncertain detail. The durable signal is that Google is moving paid inventory into AI answers and rewarding relevance and data quality over raw bids, a direction consistent with everything else it has shipped in 2026. The uncertain detail is the exact mechanics, timing, and controls of these two specific formats. Prepare for the direction by improving the fundamentals that every AI-era format will reward, and wait for Google to confirm the mechanics before restructuring anything. That way you are ready if the format arrives fast and no worse off if it arrives slowly or not at all in your market.