How AI answer engines will reshape the ad-supported web

If ChatGPT, Google Gemini, or other AI tools could eventually answer any ad tech question, many people will never discover this newsletter.

Specialized AI answer engines already deliver users information ripped from Ad Tech Explained, as you’ll soon see below. If these tools directly answer any ad tech question, I expect my search traffic to plummet eventually.

Roughly 40% of visitors land on adtechexplained.com from Google search, and 20% of Ad Tech Explained subscribers discovered the newsletter via Google. 

If that search traffic evaporated due to the widespread adoption of these AI tools, then a proportional 40% drop in display advertising revenue and a 20% drop in newsletter subscriptions wouldn’t be surprising.

Now, scale this out from a rinky-dink ad tech website to all publishers, and we are staring down a potential existential crisis on the Internet. One in which AI answer engines could cut publisher traffic and, consequently, advertising revenue nearly in half.

Gartner recently predicted that search engine volume will drop 25% by 2026, following a similar prediction that organic search traffic will drop 50% by 2028 due to users embracing generative AI search. That's a lot of potential traffic and monetization potential that could disappear soon.

What will the web look like after this evolution in user habits? How can publishers survive? And what is an answer engine? In this piece, we’ll explore these questions and more.

What is an Answer Engine?

An answer engine is an AI-powered chat tool that answers any question and could eventually replace traditional search engines like Google. They provide direct answers by browsing the web for you and then synthesize and summarize that information into a formatted natural language response.

I like the term "Answer Engine" since it succinctly positions a specific AI chat tool feature in direct competition with search engines. While ChatGPT can be considered an answer engine, tools like Perplexity compete directly with search engines like Google and more neatly fit the term—although rumors of a dedicated search feature for ChatGPT surfaced just this week.

To see an answer engine in action, check out this result from Perplexity. I asked it, "What is the ad selection API?" and it referenced Ad Tech Explained as the first source:

That's cool, but it also means most users don't need to visit Ad Tech Explained.

Consider the below examples from the Arc Search app, another answer engine from The Browser Company that may offer a glimpse of how most users will engage with the web in the future. 

I asked Arc Search, "What is Google PAIR?" and wouldn't you know it, my PAIR explainer is the first referenced content source. 

Arc search produces an entirely new piece of content by referencing my content. Depending on how much someone wants to know about PAIR, they might have no reason to visit Ad Tech Explained, eliminating my chance to gain a new user or earn advertising revenue. 

ℹ️ It is worth noting that both Perplexity and Arc Search leverage OpenAI models to power their products.

Publisher Headwinds in the Age of AI

Answer engines are not the only challenges facing publishers; they only pose one threat in addition to the other headwinds barreling down on web publishers:

1. Google is thrusting the Privacy Sandbox upon publishers, forcing them to completely upend how they earn advertising revenue from passerby web users (all while they seemingly don't care about their own Google Network revenue decline)

2. Made for Advertising sites (and Forbes) are sucking away ad revenue from publishers playing by the rules

3. AI-generated spam is flooding the Internet stealing valuable search rankings and ad revenue

4. Apple is developing a web eraser/ad blocker that will be built into Safari, the most popular mobile web browser in the United States.

5. Walled gardens continue to dominate time and attention, leaving little leftover for the open web

The ad-supported open web, one in which a rich diversity of publishers can flourish, is likely facing an extreme contraction. But what does this mean for the future? 

The future of an AI-driven web

So, let's recap. AI tools could soon start siphoning publishers' traffic, and big tech behemoths are kneecapping their capabilities to monetize remaining users. Where do web publishers even go from here? There are only a few options:

1. Cease to exist

The most morbid reality is that many publishers who rely on advertising revenue will cease to exist as they see their traffic and business decline. 

Some may view this as a positive, as a large portion of the "open" Internet is a cesspool of listicles, endless slideshows, and low-quality content packed in so tightly with ads that you can barely breathe. 

The real tragedy is the potential limitation of access to journalism. Like it or not, advertising funds journalism and creates a model where anyone can access well-reported and researched information for free. Without advertising, journalistic outlets only have one other sensible business model: subscriptions.

1. Pivot to subscriptions 

If you don't subscribe to The New York Times, The Wall Street Journal, or even Business Insider, you have most likely faced the frustration of clicking a link with an interesting-sounding headline only to face the dreaded paywall.

I explored the ethics of advertising vs. subscription business models a couple of years ago if you want to dive deeper into this topic. In that article, you will discover the unfortunate side effect of ad-supported content no longer being a sustainable business model: information is available only to those who can afford it.

Given all the headwinds I outlined above, would you blame a publisher for taking refuge from the barrage behind a cozy paywall of predictable and sustainable subscription income? 

But let's face it: not every publisher provides enough value to warrant a subscription, and there is not enough of a market to support the number of publishers on the web today in the age of subscription fatigue. 

An avalanche of subscription services is weighing on consumers. Do you think most users will cancel Netflix to read some news?

1. Extract more advertising value from users

The good news is that a reduction in ad supply due to folding publishers would make the remaining inventory more valuable. Web publishers that survive extinction can implement strategies to thrive within the new constraints.

These strategies involve pursuing alternative ID solutions, building out support for Google Privacy Sandbox, and continuing to beg users not to disable their ad blockers.

Publishers need to make the most out of the remaining traffic and do everything they can to educate advertisers to direct spending away from MFA (Made for Advertising) sites and toward legitimate publishers. If MFA sites continue to increase available ad supply, there will always be a downward pressure on price, a deadly combination in an age of declining traffic. 

1. License Content to AI companies

The "if you can't beat them, join them" approach. How can publishers monetize the content scraped by AI tools? 

Reddit brokered a $60 million deal to license its content to Google to train its artificial intelligence models. OpenAI entered licensing agreements to feed its models with content from the Financial Times, Axel Springer, and more.

I expect similar deals to continue to materialize from large media companies as any AI model will benefit from a backlog and ongoing pipeline of high-quality content as traning data.

This could cause each AI company to rush to lock in content sources to exclusives, stratifying access to information and potentially injecting political or moral bias based on the sources they do business with. However, some AI companies and publishers may be waiting for the outcome of several pending lawsuits before they choose their path.

The New York Times sued OpenAI and Microsoft for copyright infringement, alleging that OpenAI used its content to train its models. The lawsuit came after the NYT failed to negotiate a proper licensing deal with OpenAI. 

This threat has not slowed down AI companies as the tools continue to develop rapidly. Microsoft has even promised any user of its generative AI tool, Copilot, that they would assume all responsibility and legal risks if a content owner challenges a user's Copilot creations on copyright grounds. 

Hoovering up content from any source is the name of the game, but AI companies are increasingly wary of the legal risks, as evidenced by OpenAI's CTO fumbling through an answer when asked if the company used YouTube or Instagram videos as a training source for its video generation tool, Sora.

OpenAI contends that their models don't store data like a database; rather, they learn from relationships in information to create something new. This thinking insinuates that they should not face copyright claims — a flimsy argument. Open AI continues with an equally shoddy analogy: 

I don't know many teachers who have mastered every subject and can instantly respond to any conceivable question asked by millions of people simultaneously.

Even if OpenAI continues inking licensing agreements, most publishers don't produce enough content to warrant a deal. Consequently, these publishers are at the mercy of answer engines building a compensation model for everyone else.

1. Answer Engine Optimization 

A new term I'm starting to see pop up is "Answer Engine Optimization," or AEO.

Answer Engine Optimization is the practice of optimizing and structuring content to increase its likelihood of being used as source material for answer engines. Like SEO is for search engines, AEO is for answer engines.

Even though the goal of these tools is to provide direct answers, they often provide sources from where they pulled information, so while the opportunity to gain referral traffic will be much smaller than traditional search engines, it is still an opportunity nonetheless to feature your brand and maybe lure a visitor to your property.

Additionally, depending on how the lawsuits from the last section play out, the judicial system may force answer engines to enter revenue-sharing arrangements with the sources of their information. 

While Perplexity does offer a paid "Pro" service, answer engines cannot gain the same scale as search engines today without a free option — which is where advertising could come into play. But is there even a clear and viable advertising business model similar to sponsored links on Google searches? 

On a recent episode of the Hard Fork podcast, Aravind Srinivas, CEO of Perplexity, poses this question (59:00) and doesn't know the answer himself. However, he does indicate that Google may be slow to introduce these features themselves, given that they have a massive search advertising revenue line to protect (classic innovator's dilemma).

When pressed about why publishers shouldn't be terrified of Perplexity and answer engines impacting publisher monetization potential, Srinivas could not provide any clear reason but did indicate a willingness to work with publishers on the problem. 

Publishers can only hope that legal verdicts will eventually require answer engines to compensate content sources, but how they will earn advertising income or pay out publishers remains to be determined.