Over the past few years at OV, and specifically here on our blog, we’ve explored how shifts in consumer behaviour are reshaping growth opportunities for businesses.
As we approach 2026, we may be on the brink of another seismic change, prompting a crucial question for business leaders: do we commit to the AI and Large Language Model (LLM) race, or do we sit it out?
I recently had a conversation with a client looking ahead to 2026 and the marketing activities we wanted to focus on. It feels representative of a lot of the conversations happening, not just as we approach year end, but also as we move further towards an AI-dominated era of marketing and automation.
It began with reviewing the game plan of the last 12 months which had been “How do we drive more business from search?” and reflecting on whether it was still the right question to ask.
The discussion quickly evolved, first to, “How do we capture opportunities from LLMs and should our SEO focus now switch to a GEO – or Generative Engine Optimisation – one?” and then to a far more important question of “Where is most of our recent growth coming from, and does AI and LLM as a channel fit with our Go-to-market plan and acquisition strategy?”
This final question is exactly the one business leaders need to confront right now. It’s not just about experimentation – it’s about strategic clarity in a rapidly shifting landscape.
From Search to Answers
There is no doubt digital consumer behaviour is changing apace. In 2024 Deloitte reported 36% of UK adults had used a GenAI tool. That number, just 12 months on, is now 47% – and given how almost all the major platforms are adopting AI engines and LLM natively into their experiences (think everything from Google, Facebook, and WhatsApp to Reddit, Amazon and YouTube), it’s likely this number is much higher in reality (not everyone will be aware that the automated summaries they’re seeing and using are AI answers).
This wider adoption is also changing the way people search and consume information – prompts entered into ChatGPT or Perplexity are much longer and more precise than the traditional keywords people have used to search for results in Google.
Why? Because the expectations of these tools are different. We’re now searching for answers, bespoke and specific to our needs, evolving from the understanding that these new AI tools are powerful enough to do just that, rather than retrieve a page of links that might – if we’re lucky – provide relevant content.
Strategic Implications for 2026
Let’s be clear, blind faith and investment in what everyone else is doing is not a strategic decision and the next shiny marketing channel is not going to suit or deliver results for every business. Further, a strategic conversation about the allocation of time, budget and resource should never start with channel selection. Businesses need to remain clear on the fundamentals of where they win vs. their competitors, how (and where) their target audience is influenced, and stay laser focused on what their business is capable of – Return on Effort being key.
However, as we move towards the start of 2026, many businesses will be seeing their organic traffic disappear and its attributed revenue along with it (it’s a 61% reduction in organic CTR according to the latest findings). This is prompting lots of conversations about whether existing strategies still cut it.
Here’s our pick of strategic considerations for business leaders to reflect on:
1. Tom Roach, in his marketing week column, recently said “SEO isn’t dying, it’s having babies”.
The implications of this are clear – first, investments in SEO should continue. Traditional Search is still used by a much larger proportion of the UK population and this will remain the case in 2026. Second, the way in which brands and agencies assess SEO investment needs to become more nuanced, making room for both (leaning into where there’s cross over and broadening the remit where necessary).
2. Marketing to machines
Remember when everyone was obsessed with writing content and headlines just to satisfy Google…For those with long memories, you’d be forgiven for thinking we’d wound back the clock 25 years. The emergence of GEO (Generative Engine Optimisation) matters because it makes content easier for LLMs to find, interpret, and trust as a source when generating answers, and if (when) they become a core media channel for many brands, being cited and surfaced in AI answers will matter. However the same risks exist now as they did with the early days of SEO: don’t lose site of the user and eventual consumer. Whilst we need to consider LLM visibility, we aren’t yet at the point where machines are making all the decisions for us.
By using clean site architecture, structured data (like schema), fast, crawlable pages, and machine-readable signals such as clear headings and internal linking, you reduce ambiguity and help models correctly understand topics, relationships, and context. This technical clarity increases the likelihood that LLMs will surface and cite your content in conversational answers, rather than overlooking it in favour of better-structured competitors.
3. SEO and GEO are related but different.
-
- Shane O’Leary said recently that to have success and visibility in LLMs, you need to understand that they’ve been trained and built in different ways. He cited a massive research study that found Reddit, Quora, Wikipedia, Yelp & Trip Advisor were hugely dominant in the results for tools like ChatGPT and Google’s Gemini.
-
- GEO is focused more on answer visibility and citations and as a result, brand and content strategy is of greater importance. AI engines and LLMs want to cite authoritative sources in their answers, placing greater focus on authority, contextual relevance and trust. It’s a bid for inclusion, not clicks.
-
- If SEO evolved from a volume game into high trust, high quality content, GEO has turbo charged the table stakes. It’s now a high volume and high-quality game – the increased cadence and production efforts of this content (critically) distributed to the right channels is placing greater demands on marketers everywhere. In an LLM-mediated search environment (think Google AI Overviews, Perplexity, ChatGPT Search, etc.), content is evaluated differently. LLMs compress and summarise the dominant consensus view. If your content echoes that consensus, it becomes undifferentiated and is absorbed into the summary layer rather than linked or cited. Only content that introduces new information, whether perspective, data, methodology, or conceptual framing, has a chance of being surfaced or referenced.
-
- This eMarketer chart demonstrates the differences perfectly:
4. Measurement of LLM and AI engine visibility is hard, expensive and unreliable.
A recent study found that over the past 10 years Google has indexed 2 pages for every visitor it’s sent to a website. This has enabled Google to become the behemoth it is, and in turn provide a huge source of traffic and revenue for much of the planet. By comparison, Anthropic (the company behind the LLM Claude) index 10,000 pages on average to every visitor it sends to a website.
This has some massive implications.
-
- With so many LLM prompts, but with so few clicks, businesses can’t continue to use traffic as a KPI (if this was ever right in the first place).
-
- Businesses working with agencies should be asking how success is to be measured, if/when their strategies are switching to a balance of SEO and GEO focus.
-
- With brand visibility and authority being key determinants of AI engine and LLM inclusion, older metrics may well get a second lease of life. Share of voice, share of search, measures of brand awareness – these are all proxies for “mental availability” (the ability to be recalled) by your target customer / ICP. These are also the same signals that LLMs are looking for in their bid to cite the best sources and provide the most relevant, accurate results.
-
- There are lots of emergent tools to track LLM citations and traffic. None, as we enter 2026, are reliable given the unique and personalised responses that Answer Engines and LLMs provide. There are some simple GA4 hacks to see LLM traffic appear in source / medium and channel reports but this ignores the point raised above about diminishing traffic. Businesses’ drop in organic traffic will not simply reappear as equitable LLM traffic!
The bottom line
Our guidance to marketers and business leaders entering 2026 is to first ask where your existing business comes from, and how disruptive AI engines and LLMs are to the decision making of their customers of the future.
For those that do need to jump in, marketing success will hinge on whether brands are trusted, cited, and contextually relevant within AI ecosystems. The winners will treat GEO not as a technical fix, but as a strategic transformation – uniting content, PR, and brand authority to earn visibility in the conversational era of search.
