Search Without Search Engines
You’ve probably noticed by now that whenever you search for anything on Google you have an AI response, summarising the findings and giving a pretty decent response. We’re at the beginning of a major shift in how people find information online, and with it, a transformation in how businesses will need to think about visibility, advertising, and SEO.
LLMs are changing how people search
The rise of large language models (LLMs) like ChatGPT, Claude, Gemini, and others is already reshaping how users interact with the web. Instead of typing keywords into a search engine, more and more people are asking AI for recommendations, whether it’s for restaurants, travel tips, local services, or even software or product suggestions.
Personally, I find myself using ChatGPT for a wide range of questions I would have previously Googled. Things like “what temperature should I cook salmon until?” or “what second-hand SUV can fit two sets of golf clubs in the boot?” I even use it for researching software and comparing solutions based on my use case.
These days, I only use Google for finding places local to me, like the address of a restaurant, or to locate specific websites I need to visit. (Agents will eventually remove the need for that too… but that’s another topic altogether.)
Paying to be at the top of a Google search or sponsoring results will likely become less relevant over time. It’s likely that we’ll eventually see ads embedded directly within LLM responses, though it’s not yet clear when or how that shift will happen.
Buyers are already turning to LLMs instead of Google for things like software purchase decision making. Traditionally, buying software meant researching on Google, reading comparison sites like G2 or looking at Gartner, downloading whitepapers, and booking demos. But now, more people are asking LLMs directly: “What’s the best CRM for a large sales team?” or “Which analytics platforms integrate with Snowflake?”
This changes the game completely. Instead of clicking through ads or landing pages, buyers might now receive a condensed shortlist from an LLM, removing the vendor’s chance to pitch altogether. If your brand doesn’t appear in that initial response, you’re already out of the running.
For software companies, this means the battle for visibility is shifting from SEO to LLM fluency. Your product needs to be well-documented, widely mentioned in reputable sources, and present in the data that LLMs are trained or grounded on. Otherwise, you’re invisible at the moment of decision-making.
The big question for businesses is: how do you ensure your brand shows up in those AI-generated answers?
A disruption to $175B in ad revenue
It’s no coincidence that Google is cautiously rolling out its Search Generative Experience (SGE) while still clinging to the traditional search interface. Search ads generated around $175 billion in revenue for Google last year. The stakes are massive. Google is likely sitting on a next-gen experience but reluctant to flip the switch too quickly, since doing so could cannibalise its own business model.
Meanwhile, the pressure is building. Users are discovering that asking an LLM is often faster and more intuitive than clicking through pages of links. If AI models can deliver equally (or more) relevant results, that’s a serious challenge to the traditional discovery model.
Will LLM responses include ads?
It seems inevitable that we’ll see ads embedded directly into LLM outputs at some point. But the challenge is how to include paid content without breaking user trust. One of the reasons LLMs are compelling is because their responses feel helpful and neutral. As soon as monetisation starts skewing outputs, users may become skeptical.
What would it look like? Would it be like a sponsored Google result clearly indicated as an “AD”, or would you just include your brand in responses to certain questions without letting people know.
Still, someone will crack this, and once they do, businesses will rush to participate.
Could brands exploit LLM outputs?
This also raises an uncomfortable but important possibility: what if brands start trying to manipulate LLMs to ensure inclusion in responses?
Not through prompt injection (which affects model inputs), but by subtly shaping the model’s outputs via data pollution, SEO-style link tactics, or pattern injection into publicly available training data.
This could be the next wave of “black-hat SEO”: figuring out how to influence what LLMs say without any formal advertising model in place. If someone figures out how to reliably steer LLM answers, that’s the new equivalent of ranking #1 on Google, and potentially even more powerful, since users don’t see multiple options - just one synthesised response.
Toward “LLM Optimisation” - a new kind of SEO?
If people are bypassing search engines and asking AI assistants instead, then traditional SEO becomes far less relevant. What replaces it?
We may see the rise of LLM Optimisation, or LLO. This would involve:
- Publishing high-quality, structured content LLMs are likely to reference
- Ensuring brand presence in authoritative sources, databases, and reviews
- Making your content and data AI-readable and easily crawlable
- Using structured metadata to provide context
In this new world, owning your presence across structured and semi-structured data sources becomes critical.
In short
LLMs are changing the rules of online discovery. The old model - keyword search plus paid ranking - is already starting to fade for certain use cases. Instead, people will ask their AI assistant, and that assistant will decide what to surface.
For businesses, the challenge is clear: how do you stay visible in a world where LLMs filter the web on your behalf?
The shift is already underway. What comes next will reshape digital marketing and advertising as we know it.