For most marketers, 2025 was dominated by talk of Generative AI. We were told that traditional SERPs were dead; that AI mentions were everything, and that we all needed to rapidly re-skill or risk an inexorable slide towards irrelevance. 

Thing is, a lot of the practical advice on generative engine optimisation (or GEO) was both confusing and contradictory. 

In one breath, we were told that winning in AI search was contingent on our ability to completely rethink the way we wrote and published content. In the other, we were told that generative AI basically aped traditional search engines like Google, and that good GEO was basically just good search engine optimisation (or SEO) with a fresh coat of paint.

For marketing managers and CMOs with a limited budget, this sort of advice is beyond infuriating. There simply isn’t the time or resource needed to endlessly pivot without clear guidance – and a strong steer on the likely commercial impact. 

To make sure this year gets off to a better start, we’ve pulled together the latest research on GEO, with a view to helping you: 

• Model the likely impact of generative search on traffic, conversion, and brand visibility
• Understand how and where GEO diverges from traditional SEO best practice
• Build out a robust playbook for GEO, with actionable steps for your team

Before we dive in, we will say that this is an embryonic subject, and that best practice is evolving on a weekly basis. We’ll keep this article up to date, but don’t be surprised if some of the advice changes.

Why does GEO matter?

First things first, it’s important to cement the case for GEO. GenAI has always been a lightning rod for hype, but unlocking extra budget often demands more than big promises about the future of search behaviour.

At the tail end of 2025, McKinsey published research that suggested 50% of consumers used AI at some point in their search journey. Emphasis on ‘at some point’ because understanding GEOs place in the funnel is of vital importance.

A lot of thought leaders act as if traditional SEO is a complete waste of time, but in-depth analysis of McKinsey’s research paints a much more nuanced picture. 

Consumers make extensive use of AI when they’re ‘learning broadly about a category’, or about brands, products or services, but usage drops when they move into the consideration phase and start comparing specific products and services.

There’s a big divide across sectors too. 55% of respondents said they used AI tools when purchasing consumer electronics, but only 41% would admit to using AI when searching for financial or legal services, and that’s before we dive into B2B sales journeys.

Still, the fact remains that GEO is playing an important role in the decision making process, and savvy marketers will want to do everything they can to upregulate their chances of winning regular mentions.

How does GEO work?

The million dollar question. You can’t ‘rank’ in generative search. Not unless you count mentions in Google’s AI snippets, which we’ll talk about later.

What you can do is win branded mentions and/or product recommendations. The goal here is to ensure that you get in front of people when they ask Claude, Chat GPT or Gemini a question about your sector or offering. 

As an example, we provide event management services, and want to make sure we pop up every time someone looks for events management in the North East – or asks a question about the best event management company in Aberdeen. 

Depending on how much resource you can throw at your top of funnel search traffic, you may also want to win mentions for longer-tail, more informational prompts like ‘how to plan a corporate conference in Aberdeen’ or ‘what should I look for in a conference management company’ but that’s all dependent on your strategy.

GEO best practice in 2026

Actually winning mentions is all about credibility and relevance. We’ve written about how generative AI tools work before, but if you’re looking for simple, actionable guidance that you could use to quickly improve your site’s performance in generative search, the latest research suggests that you should focus on:

1. Producing clean and well structured content. 

We know that AI finds it easier to parse and use text that’s clear and unambiguous, substantive and confident. We also know that text structure matters, and that most models prefer to surface digestible content that’s broken up into simple chunks. 

Now, we’ll caveat this by pointing out that testing variables is notoriously difficult at the moment: Recent research carried out by SparkToro highlights an inconsistency in AI results that thwarts any effort to draw strong conclusions. Nevertheless, simple, direct and engaging content with no fluff and a heavily-segmented format does seem to do better overall. 

This is general advice about content structure though: Research published by Search Engine Roundtable suggests there’s no real merit to the idea that schema markup is used by AI.

2. Creating unique content that provides tangible value. 

Gen AI tries its best to supply credible information supported by a strong data set. It fails miserably, but it tries. As a result, most LLMs have a strong bias for content that (seems to) contain unique, verifiable research of some kind. 

Leverage this by polling your audience or producing some sort of first-party data and you’ll build credibility with users and AI. If you can’t get the budget to poll your audience or conduct some very basic research, tell your CFO they’re killing growth, then look for practical alternatives: Quotes from in-house experts or external partners still add value. As do meta studies that collate and analyse other people’s research.

As long as you’re actively contributing something useful to the conversation, you’re maximising your chances of an AI mention.

3. Build brand mentions and citations

A recent study by Search Engine Land found that AI is 3 times more likely to cite .gov domains (vs .co.uk or .com). Why? Credibility, which is still generally measured in backlinks. We can debate the unfairness of this until the cows come home, but the fact remains that backlinks (or citations) are still the best way to measure a website’s overall relevance and authority.

Another GenAI study published by SEMRush found a strong positive correlation between the size of a website’s backlink profile and the number of times it was mentioned by GenAI. Some good news though: the same study also found that nofollow links were just as important for GEO, and that image links slightly outperformed their text-based cousins.

In simple terms, link building for GEO might be slightly easier than it is for traditional SEO, and a lot of outdated link building tactics may suddenly become relevant again. Brand mentions matter too, so even if you don’t score a link, simply getting other websites to mention your brand may help to improve your GEO. 

4. Leveraging 3rd Party Platforms

AI loves to cite Youtube, Reddit and other outlets with a lot of (perceived) authority. While it’s not as effective as building organic mentions on other people’s websites, posting content on those 3rd party platforms and building out your brand’s footprint will have a tangible impact on your performance in AI search. 

GEO vs SEO

You may read the above and decide that GEO is, in fact, SEO wearing a slightly different moustache and trench coat. Don’t be fooled though; there are key differences. For example: keyword optimisation, the construction of elaborately manicured h1 tags or title elements are solely within the domain of SEO.

Structural considerations are different too. SEO content has to be structured in a way that Google chatbots will understand; think hierarchies, a clear flow of information, lots of semantically-linked keywords. For LLMs, the overarching structure is way less important – as long as the content you want it to ‘grab’ is clearly segmented and in a digestible chunk, you’re good to go. 

Conversely, LLMs are much more picky about your writing style: If you look at what Claude, Chat GPT and Gemini regurgitate, you’ll notice that it’s always very clear, direct and active. Reviewing your SEO content to sharpen the writing is a great first step for anyone who’s serious about improving their GEO. 

Link building is different too. Traditionally ‘pointless’ nofollow links and brand mentions have tangible value for GEO campaigns, and you’ll probably want to diversify your link building tactics to improve your AI mentions. All in all, we think a root and branch review of your pre-existing content, and a new strategy that incorporates GEO best practices, is the natural place to start for most brands.

The tweaks you’ll be making might be small, but they will significantly impact your digital footprint and should be a priority in Q1 and 2 2026.