The advent of the AI era means the future of search is in flux. Not merely because the big names have rushed to market with substandard products, though they undeniably did. The Geminis of this world are merely the result of Google and foes behaving largely as you would expect, not wanting to get left behind.
The sudden arrival of an “answer bar” above the search results represents a mere subplot compared to the seismic storied shift that could be set to come. What it revealed, was the tech industry behemoths not only do not know how to nurse in the inevitable upheaval, but they have no real idea of what this change will look like when it takes something resembling its final form.
So if not them, who?
The answer might be a company called Perplexity AI.
You’ve probably heard of Perplexity AI, even if you’re yet to avail yourself of their service. For those still completely in the dark, it’s “an AI-powered research and conversational search engine that answers queries using natural language predictive text.”1
The app’s own front page makes clear the company would prefer it to be known as a “research assistant” – we’ll see how successfully that catches on – while in its recently made public advertiser pitch deck the company calls its product an “answer engine.”
Unlike Google and Bing and other search engines, which are all bolting an AI function onto an existing search product, and making a right mess of it, Perplexity is AI first.
That means Perplexity AI has been free of the requirement to bulldoze existing internal (and external facing) functions Perplexity AI has been able to start from fresh foundations, build on suitable terrain without worrying about, I don’t know, lead pipes and unexploded ordinances (Look, we’re not actually builders, alright?)
“For nearly two decades, the way we searched for information online was defined by this idea of ten blue links,” reads the aforementioned new pitch deck aimed at advertisers. “We’ve designed our product from the ground up so AI will read relevant information from the web and synthesize it for you in a conversational way.”
Whether or that sounds actually exciting, or just relatively, it brings us to the nuts and bolts of it all: the money.
Because while SEOs have right reason to freak out about what this new uni-response world might mean, us in the advertising game know that online advertising isn’t going anywhere. That’s not just a guess. Perplexity AI professes “global ambitions to redefine search” and if it even if it is able to deliver on them, managing to reinvent or at least reshape the way we approach interneting each day, it’s all going to have to be paid for somehow. At the moment most of the signs point to their retaining the advertising model to do just that.
(We say “most” because the above screenshot shows that, initially at least, they’re offering a multi-tier consumer model and B2B “Enterprise” offering.)
How is Perplexity AI’s ad offering shaping up?
Perplexity AI says “online advertising is changing.” It’s natural to assume, based on that statement, they will be doing something we haven’t seen before. And while Perplexity AI knows it has to set itself apart from Google in order to achieve any sort of foothold in this space, what they’ve put in front of us is more familiar than not.
It’s going to look different to Google, even if they weren’t telling us explicitly, there’d be no doubt about that. The points of an AI-based “answers engine” is to deliver you “the” answer to your query, rather than to offer you a selection of starting off points in the shape of links to multiple different websites, or ecommerce options.
Take away those links, though, what are you left with?
Space.
As long as the screen remains the window to the web (how old does that wording make me sound?!) there will be a certain amount of what we used to call (and some still do) real estate that will be filled, one way or another.
The “answer” itself will of course occupy some of that surface area, but it’s likely to be a case of less is more. If it resembles one of those recipes that starts with a seven hundred word preamble before arriving at the ingredients and instructions users aren’t going to be interested, engagement will be poor, people just won’t come back.
Infiltrating “the answer” with paid material will have a disengaging effect. First and foremost an AI answer engine needs to give you what you came for, in as simple a form as possible, in the exact place you would expect to see it.
As long as it sticks to those basic tenets, that should still leave a fair amount of screen to fill.
So what does Perpexity plan to do with it?
For starters they’re promising it won’t be wall-to-wall ads. Though perhaps par for the course in the modern age, it would be more than a bit jarring for a company priding itself on relevance and “intelligence” to become Piccadilly Circus. if we wanted that we’d all be reading local newspaper websites.
No, it needs to be clean and complementary to the primary product. Material supplementing your answer will be part of the offering and that can include sponsored inventory, such as “branded explanatory text.” Brands will be able to buy “related questions” that will appear below the original query.
And this is just for starters.
If it all sounds a lot more like display advertising than PPC, it should. Although not referred to in the deck itself, the company told Digiday, that they’re planning on working at least partly on a CPM model.
That might sound like a backward step. It shouldn’t. Anyone worth their salt in this game will tell you the click (particularly the last one) has been over-weighted as an attribution element for, ooh, a good decade now. Engagement has value, whether that’s in the form of a visit or a view. Of course all engagement isn’t equal in value but it’s the job of the attribution modellers and the analytics experts, not the advertisers, to unpick it.
Perplexity AI’s pitch deck is actually light on the delivery detail. We perhaps shouldn’t be surprised because at this stage what matters is less how the advertising product is going to compare to the Google’s, but how Perplexity AI is going to compare to Google.
At this point what will matter to potential advertisers will be whether Perplexity AI can peel off users from the big G to make it worth their while shaving off a proportionate, or higher, amount of their own ad spend.
The numbers are certainly heading in the right direction. U.S. queries are said to total 230 million monthly – an 800% increase on a year ago. To put that in context, however, Google’s AI overview tells us:
The United States sees roughly 40–60 billion Google searches each month. Google processes around 99,000 searches per second, which is more than 8.5 billion searches per day.
The company, however, has pedigree of personnel. Its founders Aravind Srinivas, Denis Yarats, Johnny Ho and Andy Konwinski, come from senior roles at Open AI, Meta, Quora and Databricks respectively. Backers include Jeff Bezos (yeah, that guy) and Garry Tan, CEO of Y Combinator. An investment from Japanese tech banks SoftBank valued the company at $3 billion.
Again, for context, as of yesterday Alphabet’s market cap was $2.06 trillion.
Additionally Perplexity AI faces structural barriers. There’s a good chance reading this article on a property developed and managed by one of the companies its looking to displace. You probably even went looking for them using one of the very tools they hope to usurp. If you created an account with them, maybe you even did so using your Google login.
But even Google wasn’t Google once. It achieved its pre-eminent position by doing something that people wanted done better than any other company. If Perplexity AI can deliver on the grand promise it stands a chance.
It’s a big ask, of course it is. If it happens at all, it won’t be overnight (so “about to”? no) but all empires fall. Slowly then all at once.
