AI Search Isn’t About Optimisation.
It’s About Consistency, Clarity, and Relevance
There’s a lot of chat going on at the moment about the role of AI search in B2B marketing. Indeed, a whole new micro-industry seems to have appeared overnight, full of people promising to get every LLM known to man to recommend your product on command.
I’ve been reading quite a bit on this over the last few weeks, and I’m not entirely convinced the direction of travel is right. A lot of it is being treated like a brand new channel to optimise, with its own checklist and set of tactics layered on top of everything else marketing teams are already doing.
But I think this approach could be missing the point.
We all know that tools like ChatGPT, Claude, and Perplexity don’t behave like traditional search engines. They don’t return a list of options for someone to click through, instead they give you an actual answer. And, like it or not, that answer shapes what a buyer thinks about you, way before they ever land on your website, speak to sales, or even know who you are.
Which means the main question around AI search isn’t about “how do we rank?”, or “how do we get the LLM to mention us?” It’s “what does the AI actually say about us when we’re not a direct part of the conversation?”
This is where I see a slight problem with all this ‘AI optimisation’ stuff that is floating around at the moment. I am not sure optimisation is the answer. I think it is more about fixing some long-term problems that most teams have been ignoring because they’ve been getting away with it. Until now.
Luckily for you, dear bang average reader, I have been doing a bit of reading around the subject, and giving it a bit of thought over the last few weeks, and I have prepared a handy ‘cut-out-and-keep’ guide to help you towards the goal of AI search nirvana.
Content consistency
From what I can see, there are a couple of themes that keep cropping up when you look at what’s actually working. The first thing is consistency. Or more accurately, the lack of it.
One complaint I see a lot is that LLMs aren’t doing a good job of describing a company or their products. While it’s easy to blame LLMs for this, the reality is it’s not really their fault. This is a classic case of ‘crap in, crap out’ - the LLM can only give answers based on the information it can glean from the internet. If you tell your story one way on your homepage, a slightly different way on your product pages, and different again on blog pages and in sales decks, then there should be no surprise that the LLM struggles to articulate what you do accurately.
And this is not just about your own website. You need to factor in off-site mentions as well. AI answers reflect your brand’s reputation in the market, which isn’t limited to your own website and needs to span review sites like G2, social media, as well as trade publications and analyst reviews.
Organisations like Phantombuster have talked about this openly. By tightening and aligning how they describe themselves across their website, review platforms, and external mentions, they’ve seen a noticeable improvement in how they are referenced and summarised by AI systems.
Be clear & get to the point
Another way to positively influence LLMs in your favour is to make it as easy as possible for them to find the information they need. As I have already said, these machines don’t just give a user a passive list of options - they give actual answers - so it goes without saying that the content that shows up in AI answers tends to be the content that actually answers the question. Not the content that rambles for three paragraphs, introduces a framework, and slowly works its way towards a point. Just the content that says, clearly and directly, ‘this is the answer.’
A good example of this is Replit’s Vibe Coding 101 page. Not only is it linked directly from their website footer - making it easy for LLMs to find - it’s designed and structured in a way that it gets straight to the point and answers all the main pertinent questions.
Focus on specific questions for relevancy
The third opportunity is around what people are asking in the first place. A lot of the advice here still leans on the idea of ‘long-tail keywords,’ I don’t really like this term - it’s old language which is being shoehorned into this new situation. What’s actually happening is that people are asking questions that are specific to their actual situation. In their recent ‘State of Gen AI’ report, Similarweb suggests that the average query length of Google searches was 3.4 words. In ChatGPT, it’s 60 words.
That’s a seismic shift in behaviour. If your content only covers broad, top-level topics, you won’t show up in those conversations because those aren’t the questions people are asking anymore.
The good news is that most organisations already have access to these questions. They come up every day in sales calls, support conversations, and customer discussions. The difference is that most marketing teams don’t treat these areas as a primary source of content. They default to keyword tools because they’re easier to quantify.
Use AI to analyse sales call transcripts and highlight recurring questions, pain points and trending topics of interest, then build out dedicated content to show up when the next buyer starts asking the same thing. You can then expand this to any data source with customer insights: support conversations, reviews, user associations, community chats, Reddit threads, etc.
When you step back and look at all of this together, the pattern for LLM success is fairly clear. AI search isn’t rewarding better optimisation, it’s rewarding consistency, clarity, and relevance. Three things most marketing teams already know they should be doing, but often struggle to deliver.
My Marketing Books:
My Marketing Courses (via Udemy):