The Impact of AI on Search & SEO

One of the questions I hear often is, "Now that we have AI search, isn't it a matter of time before SEO is dead?".

It's an interesting question, and as someone who has been utilizing content and SEO as key parts of the marketing playbook for B2B lead generation and awareness over the years, it's something I've been paying close attention to. The answer also significantly impacts the marketing strategies and marketing spend decisions for companies of all sizes.

In The Short Term, Traditional Search is Still Winning

With the global dominance of Google, it will take time to chip away at its lead in search by AI search competitors. The biggest portion of AI search activity might actually come from Google itself, as it positions itself as an AI search player (and includes more AI search results in the Google search engine results pages (SERPs)) to head off the competition.

According to a recent report by Datos (An Semrush Company) and Sonata Insights (h/t Rand Fishkin ), upstart search engines haven't made a significant dent in Google's search dominance yet. But there is strong growth of usage of Perplexity, the largest pure play AI player (with 500% growth in usage between May '23 and May '24). The current situation with traditional search vs. AI search is well explained in the recent SparkToro post and I encourage you to read the post and the study itself . The chart below from the study shows Google dominating in searches per user (vertical axis) and Monthly Active Users (horizontal axis and bubble size).

What's Holding Back AI Search Growth? Trust

In a recent survey by Digital Marketing tech company, Scorpion, only 14% of respondents would completely trust AI search results and 33% would trust depending on the source. A full 39% would always double-check the results (with 14% not trusting the results at all). Some potential issues with the results could include inaccurate information, hallucinations, lack of transparency of sources, and lack of the most recent data/news. In one project I was working on, for a newer topic, the only results that came back were from vendors pushing a particular new sub-category of enterprise software. This meant that AI summarized the vendors own very positive outlook regarding this new sub-category (without any critical content available), making it seem like it was superior to the more established, dominant original category.

Trust, Access to Sources and Recency of Data Will Drive Adoption

With higher quality results, transparency of sources, and more recent data available in the LLMs, I see the usage of AI for search increasing significantly, including Google's own AI search results taking over more of the search engine results page. This won't have a major short term impact on overall search metrics, but over time, the influence of AI search will grow significantly. In another study, viewable on Statista , by 2027 it's estimated that 90 million adults will use AI for the first online search (versus 13 million in 2023).

Perplexity's approach, with easy access to sources to judge the quality of the results, or to dig deeper, goes a long way to building more trust in the results. See below for a typical search and how the sources are identified in the search results.

Is AI Optimization the New SEO? What does that entail?

As AI Search becomes grows, brands will increasingly want to see mentions of their companies and products, services or content highlighted in the AI search results. Some of the reasons for this will include:

• Getting clicks to their sites by being highlighted as a key source for a result
• Getting direct traffic to their sites because their product, service or content was featured in a result positively
• To see if mentions of the brand are accurate or comprehensive
• To see how our brand compares to competitors in AI search

So will Rank Tracking in LLMs become an industry like Rank Tracking in Search Engines?

In order to find out how a company is going in AI search, will there be tools that measure this? Companies like Semrush , Ahrefs , Moz and many others track keyword ranks in the search engines, visibility versus competitors, backlinks, and other useful metrics related to search with a directional degree of accuracy. So can we port that model over to LLMs and measure brand strength in AI search?

Christopher Penn believes this will be a fool's errand (my words). You can click the image or here to hear his take: Almost Timely News: You Can't Reasonably Measure AI Brand Strength

I'm optimistic that clever companies will get some rudimentary (or better) ways to measure how brands show up in AI and how they compare to competitors. One thing they'll have to be careful about is the regular updates to models, which could require ongoing adjustments.

Traditional content and SEO are still crucial

The LLMs will be ingesting content in order to create their results. And, as mentioned earlier, one of the keys to establishing trust will be to cite sources and allow searchers to review them. Chris Penn and I completely agree that generating high quality content will be the key to showing up in LLMs (basically the same content and SEO playbook that works today, if done well and with high quality). I will add that I believe optimizing that content for findability (e.g. on page, off page and technical SEO optimization) is also something important to keep doing. And while you're at it, targeting high volume, high intent keywords still make sense as search volumes are typically a window into how people think about and search for certain topics or items. For both traditional SEO and LLM SEO, I recommend creating "content clusters" that are clusters of keywords, including high intent ("buyer") keywords that you target with your content and SEO optimizations. Below is a snapshot of a targeted keyword cluster and the growth from zero visbility in search to roughly 38% compared to a set of competitors, through content creation and SEO.

Traditional SEO for LLMs

Traditional SEO is the key to getting your content to show up in the search engines and I, Chris Penn and others believe that it will be key to having your content show up in LLMs as well. Some of the things involved with SEO include:

• Selecting keywords based on intent (e.g. using "web personalization platform" vs. just "web personalization"). People searching for a platform are typically looking for a tool to help them.
• Selecting keywords based on volume. If there are similar keywords, but one has higher volume, optimize for that.
• Utilizing basic metadata - page title, page description.
• Cross-linking with anchor text from other pages on your site.
• Getting backlinks from authoritative sites in your industry - often just by creating great content or getting mentioned in the media.
• Structuring your site in a way to highlight key pages for search engine priority.
• Getting indexed in search engines (e.g. using sitemaps that make it easy for them to see all of your pages).

And so on and so forth.

Structured vs. Unstructured Data

Despite this, content on the web is fairly unstructured data. It's not typically formatted in neat tables with labeled columns and rows. It's often has minimal metadata and doesn't have a clear schema.

Schema Markup is a way to place information in the code of your pages that allow search engines to treat that like structured data. It's been around for a long time, but is not always used. Rosemary Brisco, AI Consultant has been teaching ho she believes that schema markup can be a game-changer for AI/LLM optimization. Read her detailed post about AI Search Optimization , but also check out her FAQ Schema Markup tool as well.

So follow the playbook with high quality content and SEO for both traditional and LLM optimization. Lean in more into adding structured data into your content and you should be well covered (and ahead of many of your competitors who are just wrapping their heads around this).

Conclusions: AI For Search Is Coming; Watch This Space!

• AI for Search will steadily grow - both from existing search players and new entrants
• The results will become better and more transparent
• There is risk that more of the underlying content will be AI-generated and that this introduces unintended inaccuracies into the world
• Basic searches - e.g. how tall is Mt. Everest, what is CRM software - will rarely result in a click to a webpage and will likely be covered by an AI-generated answer
• High quality content creation and SEO optimization will still be essential for your company/products to show up. It may or may not lead to a click to your website.
• Consider adding more structured data (via Schema Markup) to your content to improve search engine findability
• Higher quality, creative content will be the best way to combat the coming wave of AI-generated content

I'll be posting more about this topic (and other data and AI-related topics) on the Truly Data Driven Newsletter . Subscribe and let me know your thoughts in the comments!

Image credit: Image by analogicus on Pixabay