AI in search will kill SEO as we know it

For the last 17 years, I have supported myself and my family with an income directly related to my experiences in SEO, and it pained me to write the headline of this newsletter.

I believe SEO in some form or another will always exist as long as users use any sort of search engine to request answers and information; however, what we consider to be SEO today will cease to exist in very short order if the planned changes that Google announced in their recent I/O conference become the new standard.

If I take an objective high-level view of what SEO really is in today’s paradigm of search engines, I would be forced to acknowledge that the need for SEO highlights an inefficiency in the way search engines work. This is a long way from agreeing with the myriads of people that accuse anyone that practices SEO to be just a spammer who is an Internet parasite; however, in a perfect world where search engines do everything they are advertised as doing, SEO should not really have to exist.

SEO is a middle layer that helps translate search engine requirements into the best-practice-infused implementation of web projects. The need for SEO is born out of an inefficiency that requires a broker to help the parties meet in the middle. While neither a user nor the search engine is negatively impacted by the middleman, it is still an efficiency that is only required because search engine technology was not good enough to guide users to precisely what they sought without SEO.

In the early days of search, there were advantages available for the websites that optimized for search engine best practices. Today, even prior to the launch of the generative-AI focused search experiences, search engines used to do a really good job of understanding a website’s content and marry that with the intent of a user’s search query. While SEO adds marginal returns, for most sites it has nowhere the impact that it did just a few years ago.

Generative AI propels the Internet to the next level of search where there are no best practices and at present no way to even optimize for this search. There are no rules because the rules are written on the fly. Websites can’t optimize for the query because there is no query. ( A future update will delve into the death of the search query and keyword research.)

Search engines operate by bucketing websites into database entries that can be recalled by queries that are repetitive over many people and long tail queries can be aggregated up to more popular queries. When a user queries something, the engine parses that query and does a lookup like any database lookup. Even those long tail queries that the engines have never seen before are translated into queries that are more common. Never does the engine (at least to the best of my knowledge) do a real-time query into the corpus of the entire web to do an on-the-fly ranking. It is always a query in a pre-rendered and standardized database.

With AI this isn’t a pure database query since first the AI needs to “understand” the query and then it answers the query by writing a new answer that is processed from information contained in its database. Two people asking similar questions but with different phrasing will receive different long-form answers even if the answers themselves are similar.

For example, "Explain to me the ideal exercise heart rate when running” will receive a different answer than “What is the best heart rate for a 35-year old running a marathon.” In the old search paradigm, the results would be nearly identical while in the new experience, the results are also similar but the answer text is not.

There is no query to be optimized for because it is unique for every user every time.

I am certain that not all searches will be overtaken by generative AI; however, enough of search will change that the core efforts behind SEO will be forever different.

The success of generative AI is far from a sure thing as there are still hurdles that need to be surpassed:

• Content providers might cease providing their knowledge to generative AI engines to be used as learning tools for free

• Users might be repelled by this new form of search and refuse to use it. Keep in mind that much of the adoption to date has been by typical tech users and not the general searching public.

• Google and Bing might fail to find a way to monetize this new experience and therefore generative AI would have to rely on new tools like ChatGPT and others which would keep it from becoming mainstream for some time.

Even with these hurdles, I am convinced that the roles of SEO practitioners will change, and while change might be uncomfortable, it will still be necessary and might even be in more demand. I will write a post in the next few weeks on what this role looks like as soon as I get access to Google’s new experience. In the meantime, all of this is a work in progress and I look forward to continuing to learn and share my thoughts. Thank you for reading!