Work With Us

Search Has Changed - But What Exactly Does AI Want From Us?

by Sarah Main
9 min read

Just a few years ago, reaching organic position #1 felt like the ultimate SEO win.

Not only was position #1 excellent bragging rights for us SEOs (“Look, look, our strategy worked”), but it also meant you’d be pretty much guaranteed the lion’s share of clicks, and ecommerce sites would likely be raking in that organic revenue…

Fast forward to 2026, and the search landscape looks very different. Paid ads, sponsored content, AI Overviews, video results and chat assistants are all competing for attention, before users even reach the traditional organic results. Position #1 simply doesn’t hold the same real estate in Google it once did.

For businesses of all sizes, visibility now means more than high rankings. Appearing in AI Overviews and being cited by chat assistants is becoming increasingly important for getting discovered – and, ultimately, driving conversions.

Luckily, for now, AI has levelled the playing field somewhat. It’s no longer the biggest brands with the biggest budgets that dominate the SERPs.

Our Senior Content Strategist, Sarah Main, has almost 15 years’ experience in content marketing and SEO, and has spent the last year and a half researching and testing the things that work, and don’t work, when it comes to AI search. Which is why she’s perfectly positioned to answer the question on everyone’s lips – how do I improve my AI visibility?

By the end of this article, you’ll understand how search has fundamentally changed, what LLMs actually look for when choosing sources, and one practical thing you can change tomorrow (!) to give your site a better chance of being cited.


Key Takeaways:

Search isn’t broken, it’s just different 

Over the last 20 years, there’s been a big change in search - we’ve moved from a ‘discovery’ engine to an ‘answer’ engine.

Cast your minds back to 2006 – the Nintendo 360 had just come out, ‘I Bet You Look Good on the Dancefloor’ was in the charts, and Google was very much a discovery engine.

Back then, when you searched for something on Google, you’d scan the 10 blue links on page 1 and click through to find the answer yourself. Google would help you by ranking what it thought were the most relevant websites. The goal was discovery: finding the right page that had the information you needed.

We’ve then moved through various iterations of finding the ‘best’ answers; including Featured Snippets, and from around 2024, AI overviews, where users were able to reach their answers without even laying eyes on a traditional organic result.

2025 and beyond has seen AI mode roll out in most countries, where you’re able to ask direct follow up questions. Instead of typing a few keywords and choosing from a list of links, users can ask detailed questions and have a back-and-forth conversation to refine what they’re looking for, with a seemingly friendly counterpart.  

These shifts have led to organic clicks tanking (studies suggest the declines are up to 38%). So even if your site holds the highly coveted position #1 spot, it’s likely you aren’t getting the traffic and conversions you used to from this alone.


Today, search engines are much better at understanding our intent and what we actually mean, especially as we are able to give more context in our searches.

For example, take the difference between searching “fertility supplements” and “I’m in my early 40s and trying to conceive for the first time – what supplements would you recommend?”

The second query gives Google much more context about the searcher's intent. And increasingly, search engines can use that context to answer the question directly, rather than simply pointing you towards a list of webpages you’d have to hunt through, as it would have done in the mid 00s.


Put simply, ranking highly is no longer the only way to get discovered. The question is becoming: how do we create content that AI actually wants to use? And if you’re looking for the answer, you’re in the right place.


Do people really trust AI?


Before we go any further into the content changes you can make to give your site the best chances of being cited in AI overviews, or recommended by chat assistants, it’s worth mentioning a key theme: trust.

Whilst running an AI workshop a few months ago, a client came to me and asked, “It’s all good and well us changing up our strategy to include optimising for AI, but does anyone even trust it, really?”.

It was a very good question, particularly as his business was in the healthcare space – where trust is key. In the moment, I didn’t have any exact data for him on whether it is actually being trusted or not, and he had a point. So, I did some research.

The most recent figures show:


Research also shows niche matters in search

The latest research shows that Google remains the preferred option for:

How does AI choose who to trust? 

Okay so let’s get to the point, and discuss what you’re actually here for. How does AI choose the content it cites and brands it recommends? And what can you do to make sure your website is appearing in overviews and chat assistants? 

I want to be very clear here: Writing user-friendly, trusted content, that’s genuinely helpful to users, will always win over writing solely to tick boxes for AI.

And luckily for SEO teams, all those E-E-A-T optimisations you’ve been implementing for SEO, will also be beneficial to LLMs. There’s nothing you’ll have to ‘undo’, so to speak. Which is great – but there are some small tweaks to content that we can make to give ourselves the best chances of being noticed.


How do AI overviews and chat assistants choose what they rank and cite?

Generally speaking, your content ranking in AI comes down to three core signals:

With this in mind, what types of content does AI love?


A successful content strategy, should cover the following: 

1. Trustworthy content
Not just on your site, but the wider web.


A journalist recently told me that many journalists, especially younger ones, are now using AI chat assistants to source their experts for pieces they’re writing – so making sure your experts in your niche appear is a great opportunity to build that organic coverage for your brand, that’s going to impact visibility.

2. Helpful content
Is it genuinely useful to a reader? If not, don’t write it.


3. Answer-first content
Conversational intent is increasingly important.

And finally, something I have personally tested and found has made a real difference to the performance of guides or blogs in particular, is front-loading content with key takeaways.

What I mean by that, is simply putting the most important information right at the top of the page, before you get into the details. Bullet points are fine, two to three at most. Scroll up to the top of this blog for an example – it’s unintrusive, but it gives users a clear idea of what the article is about, if it’s worth reading, and is a signal to AI crawlers too.

I’ve tested this across a number of guides, in different verticals from healthcare, to property and ecommerce – and found that it’s one of the simplest changes I’ve made that has had a noticeable impact on how content performs in AI search.

So, my challenge to you – why not take one of your existing pieces of content on your site, add a few key takeaway bullet points and test for yourself?

Written by Sarah Main
Senior Content Strategist

Evoluted partners