AI Answer Engines Are Eating The Search Engine World

Do Google's Troubles Also Mean SEO Is No Longer An Investible Traction Channel?

Over the last few weeks I’ve been contemplating and drafting this article to posit that Google search is on the verge of being so disrupted, that it could actually impact the viability of SEO as a traction channel for your business.

Given the complete dominance of Google for more than two decades and the complete inability of any competitor to make a dent in it, it seems a preposterous claim that frankly I have been nervous to make.

But coincidently, a new study posted by Leipzig University this week and reported elsewhere does actually support at least half of my thesis, so I thought I would complete and post it here.

But rather than only focussing on the decline of Google search, let’s look at the rise of an additional opposing force - the new breed of AI-Powered “Answer Engine” - and then focus on what it means for your ability to acquire customers through search.

A Shifting Paradigm for Search and Answers

For a quarter of a century, Google search has remained virtually unrivaled as the de-facto gateway to the internet and the global oracle for all questions and answers.

After more than 25 years, the basic paradigm remains mostly unchanged. When you enter a search term, the search engine mostly does not answer your query directly, it primarily organizes and presents other web pages that might suit your needs.

There are several types of search query, but more than half of all searches are not just users navigating to a given site by searching for it by name, they are asking a question and seeking an answer.

This combined ~60% of “Informational, Problem-Solving and Decision-Making” search is not only worth more than $80bn per year in revenue to Google, it has also long created an valuable opportunity for startups to acquire potential customers.

Users searching for “how to…” or “best software for…” are essentially qualifying themselves as a lead with active interest in that topic, while opening their consideration to a good answer. So by practicing problem-space or solution-space SEO, a startup can get in front of qualified leads, provide value by answering the question, then pivot the visitor’s interest into their specific solution.

For more than 20 years, this strategy has as effective and reliable a flywheel traction strategy as Google’s dominance itself.

But if some game-changing force were to disrupt users from using search engines, it would not only be painful for Google, but also for the vast ecosystem of companies that rely on SEO for finding qualified customers.

Amazingly, for the first time in 25 years, this disruption actually looks set to occur.

The Already Declining Search Experience

They say bad incentives create bad behavior, so it should be no surprise that the allure of affordable, compounding customer growth from SEO has incentivized shady SEO practices.

You know exactly what I’m referring to - We’ve all clicked through to pages that were clearly optimized to monetize our interest over answering our question. God forbid you should ever try to ask Google a cooking question, for example:

Such destination sites with overwhelming adverts are especially bad, but you’ve probably also experienced clicking through to an article with a perfect title, only to find it was clearly written to appeal to Google’s index bots, rather than a human.

All of this combined to a gradual decline in search experience, which has led to a decline in user expectations.

The AI-Powered Search Engine Decline

What has thus far been a gradual decline is about to drop off a cliff, thanks to the arrival of AI writing products.

There have been hundreds of AI-powered writing tools for content marketing launched in the last year alone, most of which primarily focus on content that is optimized for SEO. The basic proposition is that they will spit out perfectly SEO optimized articles faster than you can blink.

This is a a dream proposition for the SEO manager, but a nightmare for the reader.

We can expect an explosion in “AI Dogshit” content that is high on SEO score and low on reader value, or even low on accuracy. Combine the powerful incentives of SEO dominance with the power of AI SEO technology and it is no surprise to see behavior like this:

Spoiler alert: The 1k replies are not very kind.

Coming back to the big picture, the point is that:

AI will accelerate the existing decline of the search experience by destroying the quality of the destination results.

This also starts to open the door for something radically better for users.

Enter The AI-Powered "Answer Engine”

The advent of AI, particularly LLM powered experiences in the last year introduces a new category of experience, that is better described as an “Answer Engine.”

These change the quarter-century-old search paradigm by providing the answer instantly, rather than presenting destinations that the searcher must sift through to find their answer.

The early leader of general purpose answer engine is probably Perplexity. Take a look at how different this answer experience is from the Yorkshire Pudding example I shared previously.

Beyond Perplexity, we can see the paradigm of instant answer, rather than search and find applying in:

• Domain specific use cases, like Consensus, which provide instant answers to research queries

• General LLM ChatBots like ChatGPT, Claude or Bard.

• Custom GPTs within OpenAI’s GPT store that are trained and connected to specific data for more informed results

• The Rabbit - Perhaps a bellwether of future hardware in an AI-powered world?

• AI integrated into existing products, such as Notion AI, which I am using Q&A and fact checking as I write this very article right now.

Yet another thing I didn’t need to Google for!

While all of these examples are quite different experiences, they share an ability to provide a much more efficient route to an answer than searching on Google, which means all of them will conquest searchers from Google.

The Tipping Point

AI will power answer engines to get better while also powering search engines to get worse.

Once consumers start to experience how much more effective the answer engine experience is, search engine usage will decline. This is already happened among tech early adopters.

For me, this change was even faster than Jeremiah’s: My Google usage halved from the very first day I started using Perplexity. Could it be that the previously invincible search engine is suddenly vulnerable?

How Might This Play Out?

It would be foolish to suggest that answer engines will kill off search engines without appropriately respecting the fact that this is Google we are talking about here.

Google has obliterated all competitors for the more than 2 decades, with a >90% share of all global searches making it The 3rd most valuable company in the world.

Google makes 58% of their revenue - a staggering $162.45Bn in 2022 - from search, which means that they are not going to quietly stand aside and let history’s most valuable cash cow die. Indeed it has been reported that Google’s leaders issued a “Code Red” alert to the business shortly after OpenAI’s release of ChatGPT.

Isn't Google Already Becoming an Answer Engine?

We should also recognize that some incremental features added to Google Search over time represent a slow transition to being an answer engine already, such as:

• Featured snippets that extract text from the top destination in the SERP

• Shopping results surfaced at the product level, not just at the site level

• “People also ask” content that provides further answers right in the SERP

However, the innovator's dilemma that Google faces is balancing user demand for instant answers against their business model's need to generate trillions of click-throughs per year.

Enter: Bard

Google’s own LLM Chatbot created to rival ChatGPT is also an answer engine in the same way.

While Bard currently trails ChatGPT in most people’s assessment, Google have an $80bn incentive to catch up and find ways to work paid search results into the experience gracefully.

This is the source of the previous stats on share of search query types

Their challenge is that while the early user experiences offered by Perplexity et al do include destination pages as sources, they are of secondary prominence to the inline answer. And if the user’s question has already been answered, how many will bother to click through at all?

It’s hard to imagine a design solution that delivers on user answers without reducing article click through, but I still wouldn’t bet against Google figuring it out.

---

What Does This Mean for SEO as a Traction Channel?

In this article I’ve focus on the impact of the Search Engine > Answer Engine shift on SEO, but it is of course just as relevant to paid search as a channel too.

But while paid search ads deliver results instantly, it’s commonly acknowledged that investment in organic SEO efforts can take 6-12 months to pay back. So if it seems possible that search traffic could tip into meaningful decline within the next 12 months, you need to start considering the implications for your business now.

There are serious concerns now that Google is highly likely to be disrupted in maybe even the next five to ten years (2024 to 2029 or 2034). Google’s search quality is rapidly declining and we have new reasons to understand why. Google search is getting worse and it’s literally the least of their worries.
Michael Spencer -

AI Supremacy

Should You Invest in SEO as a Traction Channel?

As a user I welcome the arrival of Answer Engines like Perplexity and Consensus, and only see positive improvements to my life. But as an advisor to AI startups seeking sustainable traction, I am significantly concerned, but not gravely.

Reasons to Remain Optimistic for Search as a Traction Channel

• Google nails search integrated into Bard
  Both as a “Search with google” option and with links added to clickable images teased into answers - I wouldn’t bet against Google figuring out a new UX that achieves the balance

• OpenAI includes browse with Bing
  Though Bing is historically tiny compared to Google, ChatGPT is far bigger than Bard. So the extent to which OpenAI choose to increase prominence of 3rd party destinations within the Chat window via Bing, click throughs weighted by Bing visibility could start being more meaningful.

• Top 3 results have extra power
  Bard, Perplexity and others still present destination pages to the user, but there are far fewer of them. 3-5 for Perplexity, Bard depends on nature of query. This means that being within the top few results for a topic will remain powerful, while being further down the list will diminish in value

• Licensing stipulations
  Though first releases of ChatGPT and Claude were somewhat wild west, numerous copyright claims from media sites is forcing the LLM providers to negotiate terms with destination sites, which may include the requirement to provide source links, restoring the destination sites for answers

• Winning search engines probably helps winning answer engines
  Given that most answer engines are trained on the web, increasing the visibility and authority of your product online in ways that good SEO professionals would advise may also increase the chance of being featured as the right answer within an answer engine. My go-to writers on SEO Devesh and Benjy at Grow and Convert explain this point well.

Top Questions to Ask Yourself Now

• What topics can you realistically win top 3 billing for?
  Given the disproportionate power of the top 3-5 results in an answer engine world, it is more important than ever to focus building visibility and reputation around highly relevant and niche topics where you have a realistic chance of being the top result

• Can you create image or video answers for the topics you wish to defend? The ability to create images or videos, as well as data tables and graphs directly within the answer engines lessens the opportunity for gaining clicks from certain image based search. However quickly creating the exact result you want is still tricky. So search for text based content is likely to be disrupted sooner than image or video content

• What is the demographic and technographic of your user?
  While early adopters like Jeremiah and I in the valley are changing their behaviors quickly, we are a minuscule fraction of the Googling world. So if your target audience is technographicaly advanced not likely to exhibit early innovator behavior, they may finish 2024 without having ever heard of Perplexity or any other answer engine, so you probably have plently of click juice left in the SEO squeeze.

• What is the role of SEO in your business and how else could you hedge that traction? If you are primarily using SEO to acquire customers via problem space or solution space search, what other tactics could you start experimenting with now in case you need to pivot to them later? Perhaps I’ll write about this soon. Subscribe below to stay informed.

Please Speak Up!

If you have any questions, thoughts or disagreements, please let me know with a comment below or DM me on LinkedIn or Twitter.

Or if you liked this and found it valuable, please help me out by hitting that heart, restack it or share it with a friend. It means the world.