As AI search eats into conventional search engine traffic, the tyranny of keywords could be at an end, writes 1440’s Robin Grainger. That can only be a good thing.
One could argue that the launch of search engines marked the real start of the internet age. Instead of memorising whole URLs to find things online, all you needed to do was search for them.
That sparked a whole industry focused on pushing pages to the top of search rankings.
Now, that era may be passing, as large-language models (LLMs), which underlie AI tools like ChatGPT and Perplexity, grab ever-greater shares of search activity. According to one prediction, these LLMs will overtake traditional search as soon as 2028 – just in time for Google’s 30th birthday (see chart below).
That raises the possibility that the days of keywords as the means for businesses to raise their profile are numbered. Indeed, history may well come to view their dominance as an unusual blip in time.
Google, whacked
Brands strive to put themselves at the top of search ranks by gaming their algorithms using keywords and links. Tools to seduce algorithms they may be. The heartbeat of an entire digital industry, they certainly are. Common sense they are not.
So long as success is based on the ability to game an algorithm, and not the intrinsic content of a site, the user experience can be lousy. A site which looks helpful enough on the search page might turn out to be drivel. Pages and posts that are little more than word soup are common, as are counterintuitive navigation and overzealous linking.
Google Search and other technologies may love that. But to humans, it just makes building a clear picture harder. Ironically, the growth of AI search might make things better.
A triumph of common sense… with caveats
Against that background, the increasing use of LLMs looks to be a welcome return to common sense.
Where search engines look for specific words and phrases, LLMs discern meaning. They look for clarity and consistency, rather than word-for-word matches.
Where search engines look for links to and from sites it ranks highly, LLMs look for media coverage which reflects meaning (as much as 95% of all AI citations come from earned media).
Unlike search, however, LLMs will try to cover gaps in their source data with information that looks right. If they can’t find anything, they’ll often make it up, sometimes without saying so – so-called “hallucinations”.
Nonetheless, LLMs seem much closer to reflecting how humans understand the world, than search engines. That‘s a good thing, which explains why they are growing so fast.
Moving the needle on AI depends on changing (human) minds
The growth of LLMs for search will mean a world where real ideas, coherence, and credibility matter more than superficial keywords and backlinks.
And that means many organisations which have geared their marketing towards supporting search results, may need to rethink things.
With the goal of shifting minds, they may need to focus more on building the arguments rather than chasing short-term digital outcomes alone.
LLMs assess information in much the same way people do. So it makes sense to prioritise human things that achieve that: clarity on audience and messaging, and strong proof, supported by influential sources.
Talk to us about how we help companies change minds (artificial and otherwise) and shape futures.