The rise of AI-powered search summarisers like Perplexity.ai and large language models (LLMs) like ChatGPT, is disrupting traditional content marketing strategies. As users shift from web searches to these AI tools, content marketers must adapt.

We discussed the potential impact of deep learning-based tools on WWW search in our information delivery future trends post.

In this post, we discuss potential content optimisation strategies for AI search engines and LLMs.

Definitions

First, what is a search engine? In short, it allows a user to find information on the World Wide Web. You can read a detailed response to the question provided by Claude HERE.

Next, what is an AI-driven search engine? Again, a full response from Claude HERE.

Two points to note, first Google (and others) already use AI in their search engine process. Hence, they are already some way down this path. Second, the key distinguishing factor of AI-driven search engines is they learn from user interaction.

If you agree with Claude’s definition of an AI search engine, then compare it with Perplexity.ai’s definition of what it is HERE. To what extent does Perplexity.ai learn from interaction? Learning is a significant element in Claude’s definition.

For this reason and the fact that Perplexity.ai does not have its own index of information, we believe it is more a summarising engine, rather than an AI-search engine.

What is a Large Language Models (LLM)? You can read Claude’s definition HERE

Finally, what is Search Generative Experience (SGE)? Perplexity.ai definition HERE

Getting Found – Its A Moving Target

To clarify, what follows is our take on the situation as of today. There is an awful lot that could (and probably will) change.

Will SGE provide the same level of experience as Perplexity? Given Perplexity does not have its own index but piggybacks on search engine indexes and (to an extent) LLMs, will it survive?

What will happen to the LLM experience (ChatGPT, Claude and others) when Ads are added to LLMs? Voice interaction and voice search appear to be the next innovation; how will that change the mix?

What if AI fails to progress at its current pace? There are still three or four issues (we believe) that could still stop it in its tracks. Lots to ponder.

Getting Found On Perplexity

How does Perplexity decide which content it will use to provide a summarised answer to a user’s prompt? We don’t know, so we asked Perplexity for an answer. You can read it HERE.

Various businesses and individuals have published their findings on Perplexity results vs Google results for a given search phrase. The few we have read do not correlate. Context is important and to generate any sensible conclusion would require analysis of a vast set of queries. That is true even if the analysis is topic-specific.

There is a further problem. Our experience shows if you ask Perplexity.ai the same question on different days you will receive different answers. The answers (mostly) agree but will vary in content and depth. The links also change.

So how to ensure you receive a link to your content on Perplexity.ai for a given search phrase? First, it is important to understand you may not care. Some user queries are best dealt with by Google, Amazon, TikTok or other online sources. Examples include product searches where the user is in buyer mode, searches for local services and trivia (e.g. how old is George Clooney).

If the query is research related, the user will, more likely, use a tool like Perplexity.ai. Search phrases on Perplexity tend to be multi-word. They can run to several sentences in length. If your potential customer is likely to search that way, then Perplexity.ai links should (we suggest) be a concern.

Try it for yourself. Type an extended search phrase into Google and note the organic links it presents. Now use the same search phrase in Perplexity and note its links. Our research shows that often there is a close correlation, but not always. Some search text produces very different results.

Note these experiments are based on small sample sizes and a tight context window. You should not assume they represent the general case. The general case is? We don’t know.

At the moment we surmise (with limited evidence) that if your content ranks high in Google SERPS it has a reasonable chance of appearing in Perplexity.ai links. If your content is good but has little to no chance of appearing high in the Google SERPS there is a small chance Perplexity.ai will pick it up.

If your content is great, but you have no authority your chances of showing up on Google (we suggest) are non-existent. Perplexity seems (again, very small sample) to improve your chances a little.

General LLM-type applications (e.g. ChatGPT)

If a business is looking for a reference as the source of content presented by a LLM, then we suggest they will be disappointed. The situation is worse than the AI search engine example, presented above.

With ChatGPT there is at least some chance of a citation. Currently, with Claude, there is none.

If a customer or prospect uses an extended search phrase to research a topic and the response is pulled in part (less likely – in full) from your content, you are unlikely to be referenced. You don’t exist.

Again, you may not care, it depends on the type of search and many other factors – see above. If your current audience looks to the WWW for information, but they are likely to migrate to using LLMs then you have a problem.

The Response

Much has been written about how to deal with the new reality of AI-driven search engines and LLMs. Most of this advice comes from SEO experts and digital marketers.

Their advice? Produce unique content, with deep insight. We discussed the problem with that advice in a recent LinkedIn post .

The only solution to the problem, we suggest, is a root and branch re-assessment of your marketing strategy. Find other ways to reach your audience.

This website uses cookies. By continuing to use this site you are agreeing to our cookie policy

Privacy policy