Historic recurrence in search: Why AI feels familiar and what’s next

Historic recurrence is the idea that patterns repeat over time, even if the details differ.

In digital marketing, change is the only constant.

Over the last 30 years, we’ve seen nonstop shifts and transformations in platforms and tactics.

Search, social, and mobile have each gone through their own waves of evolution. 

But AI represents something bigger – not just another tactic, but a fundamental shift in how people research, evaluate, and buy products and services.

Estimates vary, but Gartner projects that AI-driven search could account for 25% of search volume by the end of 2026.

I suspect the true share will be much higher as Google weaves AI deeper into its results.

For digital marketers, it can feel like we need a crystal ball to predict what’s next. 

While we don’t have magical foresight, we do have the next best thing: lessons from the past.

This article looks back at the early days of search, how user behavior evolved alongside technology, and what those patterns can teach us as we navigate the AI era.

The early days: Wild and wonderful queries

If you remember the early web – AltaVista, Lycos, Yahoo, Hotbot – search was a free-for-all. 

People typed in long, rambling queries, sometimes entire sentences, other times just a few random words that “felt” right.

There were no search suggestions, no “people also ask,” and no autocorrect. 

It was a simpler time, often summed up as “10 blue links.”

Searchers had to experiment, refine, and iterate on their own, and the variance in query wording was huge.

For marketers, that meant opportunity. 

You could capture traffic in all sorts of unexpected ways simply by having relevant pages indexed.

Back then, SEO was, in large part, about one thing: existing in the index.

Dig deeper: A guide to Google: Origins, history and key moments in search

Google’s rise: From exploration to efficiency

Anyone working in digital marketing in the early 2000s will remember. 

From Day 1, Google felt different. The quality of its results was markedly better.

Then came Google Suggest in 2008, quietly changing the game. 

Suddenly, you didn’t have to finish typing your thought. Google would complete it for you, based on the most common searches.

Research from Moz and others at the time showed that autocomplete reduced query length and variance. 

People defaulted to Google’s suggestions because it was faster and easier.

This marked a significant shift in our behavior as searchers. We moved from sprawling, exploratory queries to shorter, more standardized ones.

It’s not surprising. When something can be achieved with less effort, human nature drives us toward the path of least resistance.

Once again, technology had changed how we search and find information.

Mobile, voice, and the second compression

The shift to mobile accelerated this compression.

Tiny keyboards and on-the-go contexts meant people typed as little as possible.

Autocomplete, voice input, and “search as you type” all encouraged brevity.

At the same time, Google kept rolling out features that answered questions directly, creating a blended, multi-contextual SERP.

The cumulative effect? Search behavior became more predictable and uniform.

For marketers running Google Ads or tracking performance in Google Analytics and Search Console, this shift came with another challenge: less data. 

Long-tail keywords shrank, while most traffic and budget concentrated on a smaller set of high-volume terms.

Once again, our search behavior – and the insights we could glean from it – had evolved.

Zero-click search and the walled garden

By the late 2010s, zero-click searches were on the rise. 

Google – and even social platforms – wanted to keep users inside their ecosystems.

More and more questions were answered directly in the search results. 

Search got smarter, and shorter queries could deliver more refined results thanks to personalization and past interactions.

Google started doing everything for us.

Search for a flight? You’d see Google Flights.

A restaurant? Google Maps. 

A product? Google Shopping. 

Information? YouTube. 

You get the picture.

For businesses built on organic traffic, this shift was disruptive. 

But for users, it felt seamless – arguably a better experience, even if it created new challenges for optimizers.

---

Quality vs. brevity

This shift worked – until it didn’t. 

One common complaint today is that search results feel worse. 

It’s a complicated issue to unpack. 

• Have search results actually gotten worse? 
• Or are the results as good as ever, but the underlying sites have declined in quality?

It’s tricky to call. 

What is certain is that as traffic declined, many sites got more aggressive – adding more ads, more pop-ups, and sneakier lead gen CTAs to squeeze more value from fewer clicks.

The search results themselves have also become a bewildering mix of ads, organic listings, and SERP features. 

To deliver better results from shorter queries, search engines have had to guess at intent while still sending enough clicks to advertisers and publishers to keep the ecosystem running.

And as traffic-starved publishers got more desperate, user experience took a nosedive. 

Anyone who has had to scroll through a food blogger’s life story – while dodging pop-ups and auto-playing ads – just to get to a recipe knows how painful this can be.

It’s this chaotic landscape that, in part, has driven the move to answer engines like ChatGPT and other large language models (LLMs). 

People are simply tired of panning for gold in the search results.

The AI era: From compression back to conversation

Up to this point, the pattern has been clear: the average query length kept getting shorter.

But AI is changing the game again, and the query-length pendulum is now swinging sharply in the opposite direction.

Tools like ChatGPT, Claude, Perplexity, and Google’s own AI Mode are making it normal to type or speak longer, more detailed questions again.

We can now:

• Ask questions instead of searching for keywords. 
• Refine queries conversationally. 
• Ask follow-ups without starting over. 

And as users, we can finally skip the over-optimized lead gen traps that have made the web a worse place overall.

Here’s the key point: we’ve gone from mid-length, varied queries in the early days, to short, refined queries over the last 12 years or so, and now to full, detailed questions in the AI era.

The way we seek information has changed once more.

We’re no longer just searching for sources of information. We’re asking detailed questions to get clear, direct answers.

And as AI becomes more tightly integrated into Google over the coming months and years, this shift will continue to reshape how we search – or, more accurately, how we question – Google.

Dig deeper: SEO in an AI-powered world: What changed in just a year

AI and search: Google playing catch-up

Google was a little behind the AI curve.

ChatGPT launched in late 2022 to massive buzz and unprecedented adoption.

Google’s AI Overviews – frankly underwhelming by comparison – didn’t roll out until mid-2024. 

After launching in the U.S. in mid-June and the U.K. in late July 2025, Google’s full AI Mode is now available in 180 countries and territories around the world.

Now, we can ask more detailed, multi-part questions and get thorough answers – without battling through the lead gen traps that clutter so many websites.

The reality is simple: this is a better system.

This is progress.

Want to know the best way to boil an egg – and whether the process changes for eggs stored in the fridge versus at room temperature? Just ask.

Google will often decide if an AI Overview is helpful and generate it on the fly, considering both parts of your question.

• What is the best way to boil an egg?
• Does it differ if they are from the fridge?

The AI Overview answers the question directly. 

And if you want to keep going, you can click the bold “Dive deeper in AI Mode” button to continue the conversation.

Inside AI Mode, you get streamlined, conversational answers to questions that traditional search could answer – just without the manual trawling or the painfully over-optimized, pop-up-heavy recipe sites.

From shorter queries to shorter journeys

Stepping back, we can see how behavior is shifting – and how it ties to human nature’s tendency to seek the path of least resistance.

The “easy” option used to be entering short queries and wading through an increasingly complex mix of results to find what you needed.

Now, the path of least resistance is to put in a bit more effort upfront – asking a longer, more refined question – and let the AI do the heavy lifting.

A search for the best steak restaurant nearby once meant seven separate queries and reviewing over 100 sites. That’s a lot of donkey work you can now skip.

It’s a subtle shift: slightly more work up front, but a far smoother journey in return.

This change also aligns with a classic computing principle: GIGO – garbage in, garbage out. 

A more refined, context-rich question gives the system better input, which produces a more useful, accurate output.

Historic recurrence: The pattern revealed

Looking back, it’s clear there’s a repeating cycle in how technology shapes search behavior.

The early web (1990s)

• Behavior: Long, experimental, often clumsy queries.
• Why: No guidance, poor relevance, and lots of trial-and-error.
• Marketing lesson: Simply having relevant content was often enough to capture traffic.

Google + Autocomplete (2000s)

• Behavior: Queries got shorter and more standardized.
• Why: Google Suggest and smarter algorithms nudged users toward the most common phrases.
• Marketing lesson: Keyword targeting became more focused, with heavier competition around fewer, high-volume terms.

Mobile and voice era (2010s–early 2020s)

• Behavior: Even shorter, highly predictable queries.
• Why: Tiny keyboards, voice assistants, and SERP features that answered questions directly.
• Marketing lesson: The long tail collapsed into clusters. Zero-click searches rose. Winning visibility meant optimizing for snippets and structured data.

AI conversation era (2023–present)

• Behavior: Longer, natural-language queries return – now in back-and-forth conversations.
• Why: Generative AI tools like ChatGPT, Gemini, and Perplexity encourage refinement, context, and multi-step questions.
• Marketing lesson: It’s no longer about just showing up. It’s about being the best answer – authoritative, helpful, and easy for AI to surface.

Technology drives change

The key takeaway is that technology drives changes in how people ask questions.

And tactically, we’ve come full circle – closer to the early days of search than we’ve been in years.

Despite all the doom and gloom around SEO, there’s real opportunity in the AI era for those who adapt.

What this means for SEO, AEO, LLMO, GEO – and beyond

The environment is changing.

Technology is reshaping how we seek information – and how we expect answers to be delivered.

Traditional search engine results are still important. Don’t abandon conventional SEO.

But now, we also need to optimize for answer engines like ChatGPT, Perplexity, and Google’s AI Mode.

That means developing deeper insight into your customer segments and fully understanding the journey from awareness to interest to conversion. 

• Talk to your customers. 
• Run surveys. 
• Reach out to those who didn’t convert and ask why. 

Then weave those insights into genuinely helpful content that can be found, indexed, and surfaced by the large language models powering these new platforms.

It’s a brave new world – but an incredibly exciting one to be part of.

Read more at Read More