Another day, another Google announcement for AI search. Following our coverage yesterday about gen-AI-infused product search, Google has rolled out ‘AI Mode.’ This brings searches from isolated one-off endeavors to a more continuous and progressive string of questions. In other words, a conversation.
Specifically, AI Mode lets users ask complex and multippart questions with follow-ups. For example, you could ask for a step-by-step guide to fix a hole in your wall, then follow up to ask what kind of drywall saw you need. Google does this with a custom version of Gemini 2.0 with an advanced reasoning model.
Like Google AI overviews, results include generated descriptions that link to source material. This method both maintains some semblance of Google’s core paid search model (more on that in a bit), and offers some credibility in that it discloses its sources. Google has all the ingredients to pull this off.
Google also reports that in testing, queries are about 2x the length of regular searches; and about 25 percent of users ask follow-up questions. But these data only reflect initial usage: It’s likely that as users learn how to get the most out of AI search, the queries themselves will evolve in several ways.
To be clear, this isn’t free just yet. AI Mode is first rolling it out to Google One AI Premium subscribers, and is accessible for Search Labs users. We could see it rolled out more broadly in the next year. It’s also worth noting that this follows similar moves from Perplexity AI and ChatGPT’s new search feature.
Conversational Memory
One reason this caught our eye, besides its continued departure from the search status quo, is that we’ve speculated on this concept before. More accurately, giving credit where due, we were introduced to the idea by Yext Chief Data Officer Christian Ward. He was talking about this before anyone else.
Besides Localogy conference keynotes that go back years before generative AI caught everyone’s attention, he recently introduced to us this notion of progressive/continuous search. This is all about “conversational memory” he says, simulating how humans converse with each other.
“The most compelling aspect of this release is how it transforms the traditional search experience into an ongoing conversation,” said Ward of Chat GPT’s similar December announcement. “Unlike conventional searches that start and end with each query, ChatGPT maintains context throughout the interaction.”
All the above reminds us of another resonant Wardism. Paraphrasing, he says that a competitive edge has always been granted in our society to computer-savvy humans. But the same primacy and competitive edge – including in local search – will increasingly apply to human-savvy computers.
“This combination of conversational memory with search capabilities opens up new possibilities for how people discover and interact with local businesses and their communities,” Ward told us.
Offers, Not Ads
Stepping back, where is all this going? Many rightly surmise that if Google goes down this path, it will disrupt its own business model. Indeed, more refined and intelligent search engenders one correct answer. That means that SERP inventory – ten blue links preceded by sponsored results – dissipates.
But Ward theorizes that though there’s less ad inventory – one result versus several – this could be an opportunity for sponsored results that carry much higher premiums than a typical cost per click. AI-driven dialogues with a user can infer deeper levels of intent and thus higher-value leads for businesses.
This boils down to a quality versus quantity tradeoff. That’s to say that Google could make up for a diminished quantity of ads with higher-quality ads. Given deeper conversations with users, Google can hyper target sponsored content. More importantly, it brings Google towards offers rather than ads.
Importantly, those offers – delivered RTB-style – can carry higher premiums to the point where the value offsets the erosion of the current SERP model. Or at least that’s the idea at a high level. In the meantime, Google is stepping towards that end, and has partially answered the question of how it plans to get there.
But of course, this is a moving target and is evolving daily, as noted above. So to hear the latest, including from Ward himself, join us at Ohio State University later this month for Localogy’s flagship L25 event. Ward will be there to give us his signature TED-Style keynote on where this is all going.
Header image Credit: Priscilla Du Preez 🇨🇦 on Unsplash
