In the first week of 2026, digital marketers still working through the holiday period were treated to a particularly interesting report from the Microsoft Advertising team.
A report was published titled From Discovery to Influence: A Guide to AEO and GEO, written by internal Principal Product Manager, Jen Myers.
Now as 2025 has closed out, and another year passed where SEO’s had the chance to trial and tinker with their LLM strategies, GEO reports could be found in every fourth LinkedIn post, each author claiming they’d finally figured out the secret sauce to ranking in AI search.
What makes this report so interesting, is that at the time of writing, Microsoft holds an investment equivalent to roughly 27% of OpenAI on an as converted basis, the creator of the most popular AI LLM in the world, ChatGPT.
This positioning gives Microsoft an unusually close view of how large language models are trained, deployed and surfaced at scale, making its perspective incredibly valuable.
As a digital marketing agency with a keen interest in growing visibility in LLMs, we were keen to dig deeper into this report and see what guidance Microsoft had to provide.
Here is a collection of our key takeaways:
“AI-driven shopping is transforming discovery and purchase journeys. Traditional SEO focused on clicks; now Answer Engine Optimisation (AEO) and Generative Engine Optimisation (GEO) determine visibility in LLM-powerd ecosystems.”
This is the first quote pulled out of the executive summary, and this tracks with the slowly moving mindset of the SEO industry.
We aren’t in the clicks game anymore; we have moved into the visibility era.
The days of writing a top, middle or bottom of funnel blog post related to your industry, then getting the chance to introduce curious searchers to your business are quickly evaporating in front of our eyes.
While SEO content will always remain important, the goal is no longer traffic alone, but being cited and recognised as a credible source within AI generated summaries.
In October 2025, I was sent by nimbl to Ahrefs Evolve in San Diego, and this topic was already being discussed at length.
As search behaviours change, so to must our thinking behind what SEO success looks like. Ultimately, we are all driven by conversions, but reverse engineering the path to those conversions is increasingly harder and harder to do.
For those eager to hear the thoughts of those presenting at Ahrefs Evolve 2025, click here if you’re interested in a comprehensive speaker-by-speaker breakdown from the world-class event.
“Should we call optimising for LLMs AEO or GEO?.”
Trick question: the author contends that they aren’t the same.
While LinkedIn thought leaders have been debating for months (perhaps years at this point?) whether or not to label this new school of optimisation as either AEO (Answer Engine Optimisation) or GEO (Generative Engine Optimisation), the author of this piece states early on that they’re two different disciplines.
As taken directly from the report:
Answer/Agentic Engine Optimisation (AEO):
Optimises content for AI agents and assistants (like Copilot or ChatGPT) so they can find, understand and present answers effectively.
Generative Engine Optimisation (GEO):
Optimises content for generative AI search environments (like LLM-powered engines) to make it discoverable, trustworthy, and authoritative.
The author is making the distinction: AEO makes content accessible by AI assistants, GEO makes it visible in AI-driven search.
“AI discovery depends on how content is interpreted and used..”
AI driven discovery is not a single system but an interconnected ecosystem of browsers, assistants and agents, all interpreting content in different but overlapping ways.
The key takeaway her being that understanding how each layer reads, contextualises and uses data is now critical to visibility.
Assistants translate user intent into guidance, while agents go further by taking actions such as navigating sites, completing forms and making purchases.
This shifts optimisation away from channel-specific thinking and toward ensuring content is usable across multiple AI behaviours.
The practical focus is no longer where content appears, but whether it is accessible, accurate and trustworthy across product feeds, structured data, reviews and real time signals, as this determines whether AI systems can confidently surface and act on it.
“...the industry is talking about shifting from SEO to AEO and GEO. At Microsoft, we believe SEO and catalogue investments built a foundation to expand upon in LLM-based search”
We agree.
In internal conversations within the agency and on calls with clients, we maintain that SEO fundamentals should be a key part of any AI SEO strategy.
While there is certainly no shortage of extra work you can be doing to improve your visibility in LLMs, SEO fundamentals (well-optimised landing pages, easy to crawl site and value-driven content) should be the key pillars of any approach.
This also means treating product catalogues, feeds, pricing and availability data as core SEO assets, not supporting inputs.
With that in mind, SEO’s influence isn’t shrivelling, it’s growing.
The report goes on further to say that while SEO has historically been focused on driving clicks, the focus of AEO is on driving clarity with real-time data. GEO, by extension, aids in establishing industry credibility through a brands authoratitve voice.
This is where SEO agencies need to think beyond the tried and true tactics, and make sure they are tangibly growing their brands voices.
“How do AI Assistants answer an ecom query?”
This detailed, yet simply laid out image breaks down the rational of an AI assistant when posed with the question Hey Copilot, what’s a good waterproof jacket for a three day hike?.
There are three phases that break down the response:
- Knowledge sources: Pre-trained base level knowledge, real time web search results, existing product data base
- Page level data: Dynamic content (meaning pricing, USP features e.g. free shipping), on page structure data, rendered page content (what does the AI assistant see on the page)
- Use info: Brand familiarity, location context and practical constraints such as sizing or availability
With all that information considered, your AI Assistant moves onto the reasoning phase, looking at things like commercial signals, how fresh is your content, contextual relevance between what you searched, and what it’s considering recommending.
Mould that all together and a response will follow, factoring in natural language, providing examples and citing trusted sources.
And as easily as that, ChatGPT has provided me with three jacket options for my impending hike.
I actually find this graphic very digestible and easy to understand.
“How AI weighs brand, data and availability in product recommendations?”
When an AI assistant evaluates a product recommendation, it enters a reasoning phase that combines crawled web data with structured product feeds.
General knowledge, category expectations and brand positioning are weighed alongside real time inputs such as price, stock availability and key specifications.
The outcome is not driven by rankings or content alone, but by how competitive, current and complete a brand’s data is at the moment of the query.
Brands surface not because they rank first, but because their information makes them a credible, available and commercially viable option.
“What happens after your customers click through to your site from an LLM?”
One of the more interesting points in the report is what happens after a user clicks through to a product page.
Once on site, an AI agent can pull in additional context in real time, such as detailed reviews, supporting video content, live promotions and delivery timelines.
It then translates that information back to the user in plain language, helping them understand availability, incentives and urgency without needing to manually piece it together.
The key takeaway here is simple: even if your feeds and catalogue data are perfect, the experience still depends on a functional, accurate and well structured website.
If pricing breaks, stock is wrong or checkout fails, the agent cannot complete the task and the sale falls over. It’s not an unreasonable assumption to guess that this will likely have an impact on further citation from AI agents.
Keen to get your brand AI-ready for the new frontier of search?
The team at nimbl are at the forefront of adapting SEO strategy to the rapidly shifting waters of AI search. Explore our AI SEO solutions for more today, alongside our other SEO services for a comprehensive, versatile digital presence that doesn’t miss on opportunities traditional or emerging.
