If you weren't already aware, we’re on the leading edge of a colossal shock wave. McKinsey’s seminal Winning in the age of AI search points out that over 55% of consumers now use AI search to augment product or vendor discovery; whether that’s asking Claude for a recommendation, prompting ChatGPT to narrow down their shortlist, or consuming the AI snippets positioned at the top of a Google SERP. 

If generative AI is going to cannibalise our traditional search traffic, we need to get in front of it and find proven ways of influencing these tools. That means maximising mentions, or guiding LLMs to talk about our brand in an attractive way. 

We’ve talked about the mechanics of GEO before, but to expand on this we wanted to share even more insight, to make it actionable, and take the opportunity to surface original insights from some of the SEO industry’s brightest sparks. To accomplish this, we approached six industry luminaries and asked them to share their top tips for winning at GEO. 

Now, I think it’s important to acknowledge that GEO is an embryonic subject. Just like SEO before it, best practice can shift incredibly quickly, but these tips are the bleeding edge as of April 2026, and can be fed directly to teams looking to enhance your GEO. 

Here's what six of the best in the business had to share.

Chris Shelbourn - IDHL’s Organic Search Project Delivery Director

"Whilst the growing user shift towards AI engines has certainly prompted learnings and change within our team, we still structure our approach to GEO around our core delivery pillars: Technical, Content and Off Site / Authority.

From a technical perspective, if AI can’t easily access your pages, understand what each page is about, or see how content connects, you’re unlikely to be surfaced. Technical fundamentals like a clean crawlable navigation, strong internal linking, fast load speeds, clear heading structure and relevant page titles still matter. Factors such as schema have become a stronger focus, and we are exploring newer developments such as the LLMS.txt file with our clients. 

In terms of content, our aim is to research and address real user questions, and structure content in formats AI engines can more easily digest (such as short paragraphs, FAQs, TLDR, tables, bullets, numbered lists etc). LLMs rely heavily on entities and relationships, so consistency really matters - what you do, who you’re for, and why you’re credible should be obvious throughout your content. As with Google’s historic EEAT principles, surfacing trust signals, expertise and your brands reputation via content also remains a core focus for LLMs. 

Perhaps however, the biggest change we’re seeing is for our off site teams. LLMs cite a vast range of sources from blogs, media, advertorials, affiliates, social channels, forums, business directories, Wikipedia, review sites - as well as our client’s sites and their competitors - when formulating responses. We’ve therefore worked closely with specialists across the group to find ways of growing brand authority and positive sentiment coverage across these mediums."

Tristan James - SEO/GEO for EE

"One of the biggest challenges we've had as an organisation is both communicating and controlling the narrative around GEO. To remedy this we've focused on the idea of aligning GEO requirements with existing fundamentals that, by and large, teams across the three brands in the group are aware of. Brand, Content and Technical. 

With these three core pillars we've been able to support existing SEO initiatives with new value, piggyback to some extent on an appetite for something 'new', educate on the nuances of GEO, bringing the SEO team internally to the forefront as experts of another acronym, and build all of this into processes and project plans offering a robust but understandable approach that allows for stakeholder buy-in.

With this pillared view of GEO we've been able to create really interesting growth initiatives in brand visibility, build out pilots for new multi channel content strategies on platforms like Reddit, successfully start a whole process of content creation across all areas of the business to feed the more educational, top funnel content requirements, something we previously had with BT Sport, and even start preliminary discussions around Agentic readiness and commerce, something that in short order will likely be critical and for a large organisation with a complex tech stake may take some time to implement. 

So for me, success in GEO is currently more in control and narrative than specific tactics that differ from good SEO in the broad sense."

Omi Sido - Senior Technical SEO & AI Search at Canon 

"Technically speaking, GEO is not about ‘optimising for LLMs’. Base models do not crawl or index the Internet - they generate probabilistically. The visibility layer in generative search relies on information retrieval systems (search indexes, vector databases, RAG pipelines).

That means I remain hyperfocused on what I have always been: crawlability, indexability, structured clarity and now chunk-level extractability.

First of all, retrievability is foundational (nothing works without reliable retrieval). In other words, if crawlers can’t crawl, render and index the content of a webpage reliably enough, that page will never get into retrieval systems to begin with. From an SEO point of view, this means:

• Always use clean, semantic HTML (proper <h1>, <h2> and <h3> hierarchy, structured lists, accessible tables).
• Do not over-rely on client-side JS (avoid JS-dependent critical content). Use server-side rendering or fully rendered HTML snapshots.
• Make sure internal linking is logical and that it reinforces entity relationships and topical clusters. Internal linking is for entities, not just ‘SEO juice’.
• Apply consistent URL structures and canonicalisation.
• Use structured data that is aligned to intent (Article, FAQPage, HowTo, Product, Organisation) in order to disambiguate entities. Explicitly tell the system ‘this is what this page is about’.

Secondly, we now have to think at the passage level, not just the page level. Many retrieval pipelines chunk documents into token-limited segments. If a section of a web page is extracted in isolation, will it still make sense?

In practice, this means that from now on, we need to change how we write web page copy:

• Self-contained sections are key. Every sub-header should lead into a section that fully answers a specific question (the goal is not to artificially ‘chunk’ your content for AI, but to make sure each section is semantically complete and understandable if retrieved independently).
• Front-load your definitions. Put the ‘What’ and the ‘Why’ at the very start of your sections.
• Kill the vague pronouns. Stop saying ‘this approach’, or ‘it works by...’ halfway through a page. If that sentence gets chunked and separated from the intro, the AI won't know what it is. Re-anchor the subject every time. I hope it makes sense.
• Narrow the scope. Don't let a section drift into three different sub-topics. Keep it tight to avoid ‘semantic dilution’.

From an SEO point of view, the shift with GEO isn't abandoning the traditional SEO fundamentals - it's extending them. We are no longer optimising just for ranking positions - we are also engineering pages to be reliably retrieved, cleanly parsed, and safely extracted at the passage level.

In a generative environment, structured clarity isn't just good practice - it's a visibility requirement."

Dan Taylor - Head of Innovation (Search & AI) at SALT.agency

"GEO isn't just a tactical shift; it's a strategic overhaul of how we approach core traditional SEO tasks like keyword research and content planning. Rather than fixating on high-volume terms, we're now prioritising comprehensive topic coverage to thrive in AI-driven search. 

The Infinite Tail, that vast array of hyper-specific, often zero-volume queries surfacing in AI interactions, demands we tie targeting directly to audience needs and intents. This ensures our content emerges across diverse discovery paths.

It perfectly tackles Query Fan-Out, where AI unpacks a single prompt into multiple sub-queries, drawing from related intents and entities to craft responses. 

By leaning into the Infinite Tail through GEO, brands align with these expanded query clusters, mirroring real user behaviours and the psychology of information foraging. 

And the facts we highlight about the brand and website hold true when measured against Grounding Queries, the precise searches AI triggers in fan-out, affirming our relevance and authority."

Alex Bussey - Digital Strategist at Mearns & Gill

"GEO is certainly important, but I don’t see it upending pre-existing processes. If anything, it simply deepens and enriches standard SEO best-practice. 

Link building is a great example: Winning high-value links from authoritative websites is still incredibly important, but there’s now a very pressing need to broaden our horizons and look at ways of winning branded mentions too.

Keyword research and on-page optimisation are also changing: We used to optimise for keywords that struck a good balance between volume, intent and specificity. Now, as Dan notes, AI tools use something called retrieval augmented generation (or RAG) to improve their responses, and most RAG models use something called a query fan-out to boost the accuracy and comprehensiveness of answers.

Practically, this means we now try to optimise for a bunch of different search terms, and finding those search terms isn’t as simple as it used to be. 

To explain, traditional search behaviour is self-reinforcing. If searching using a specific format (‘dog groomers near me’, ‘best dog groomer for exotic breeds’) people will learn to replicate that behaviour; uniting around fairly-predictable inputs that we can use a tool like Ahrefs to profile & exploit.

If people are now using more natural language - and we are relying on AI tools to turn those inputs into a fan of relevant queries - we can’t really trust (or use) search data to guide search marketing strategies. Instead, we need to think more holistically about user behaviour and intent; spending more time profiling our audience and their use of various research tools. 

This means more interviews, and real-life conversations. It also means shifting towards a content creation model that heavily prioritises discoverability. Everything is chunked up neatly, and we try to focus content instead of trying to cover every base."

Sarah Ricketts - SEO lead at Busuu

"Due to the exponential growth of AI/LLMs/GEO, it's more important than ever to adapt - or be left behind in the blur of the ever-changing search landscape. AI has surpassed technological advances of the past, however it’s key that site integrity is not compromised amongst the opportunities LLMs present. We still need to exercise discernment, adhere to compliance and data security, whilst harnessing the power these systems provide. 

As GEO is an iterative field, it requires an adaptability mindset, a willingness to responsibly test and being okay with pivoting, using a data led approach. 

Understanding the mechanics behind how LLMs surface and interpret requests is essential. It is important to understand how these models operate and what signals they prioritise, such as query fan-out data and cosine similarity. Developing an iterative prompt engineering and testing framework is also key, as there is no one-size-fits-all approach; analysing outputs across different LLMs helps build a clearer picture of what effective results look like. 

Technical SEO remains fundamental, particularly through the use of structured data, which provides LLMs with a clearer structural understanding of your website. Organisations should also embrace agentic workflows that responsibly combine AI and data to generate content formats such as FAQs and tables, whilst adapting content so it can be more easily interpreted and resurfaced by LLMs. All the while ensuring a human touch remains integrated throughout the process, as the reader is, and always will be the priority.

Reporting frameworks should also evolve to measure where efforts are delivering ROI in terms of reach and visibility, for example through AIO citations and LLM-driven referral traffic and conversions. Finally, it is important to approach LLM usage consciously, recognising that every query and automated workflow consumes real-world resources such as energy, water and data, so these tools should be used thoughtfully and with purpose.

Essentially, GEO is an extension of excellent SEO, continuously elevating the standard will develop the foundation for LLM growth to follow." 

Roundup:

As you can see, there’s no ‘right’ approach to GEO, but a consensus is emerging. Frameworks - and the specific actions taken to influence generative ‘answer engines’ - differ from company to company, but most experts agree that the growing importance of AI search demands a behavioural shift.

Chunking content, examining query fans, thinking about the way we write content and finessing the technical SEO that underpins page performance are critical. This activity has to be undertaken holistically too, and while it’s tempting to oversimplify for brevity, the fact of the matter is that investing time, and thinking carefully about the way your website’s understood by AI tools will be key to succeeding in the age of GenAI.