Discovery Optimization for AI - A Simple Explainer
AIO? GEO? AEO? REO Speedwagon? No one really knows what to call it yet, but here's how to work with what we've got...I think.
As a business or content creator, there’s a decent chance you’ve spent real money on content this year. Blog posts, LinkedIn carousels, whitepapers, maybe a podcast nobody but you and Alexa (secretly) listened to past episode three. You’ve done the keyword research and ticked the classic SEO boxes, but…
Try and ask ChatGPT (or your platform of choice) who/what the best option is in your category. There’s a good chance your brand isn’t mentioned and some competitor you’ve never heard of probably is, because they wrote one very specific, very well-structured article 18 months ago and forgot about it. Said article is now being cited by AI engines to millions of buyers while you’re still optimising meta descriptions.
This is the situation for lots of folks, so I’ve been hearing. It’s not a crisis, but it probably will be if you leave it be.
How We Got Here (very brief)
Search used to be simple. You typed “best CRM software” into Google and it showed you ten blue links. You clicked the second one because the first was always an ad. Life made sense.
Then, in November 2022, OpenAI launched ChatGPT. Within two months it had 100 million users. Within 18 months, people weren’t just using it to write cover letters and free therapy anymore. They were using it to research vendors, compare products, and make real purchasing decisions. Today, 89% of B2B buyers now use AI as a primary source of self-guided research throughout their buying journey.
The search bar didn’t die, but it moved somewhere else.
The numbers are still small relative to the traditional Google search volume, but the trajectory really matters. AI referral traffic to websites grew 527% year-over-year through mid-2025, according to Previsible’s research. Here’s the kicker: that traffic converts at 14.2% compared to Google organic’s 2.8%, based on an analysis of 12 million website visits published by Pixelmojo in February 2026.
The volume is modest but the intent is extraordinary.
Like what you’re seeing?
What All These Acronyms Actually Mean
The marketing industry, being the marketing industry, has given this shift approximately four different names and is currently having a very online argument about which one is correct. You can ignore the argument. What matters is that they all describe roughly the same thing.
GEO (Generative Engine Optimization)
The term that comes from academia. Researchers at Princeton University, Georgia Tech, the Allen Institute for AI, and IIT Delhi published the first peer-reviewed study on the subject at ACM KDD 2024. Their paper introduced a benchmark of 10,000 queries and tested nine specific content optimization techniques to see which ones made AI engines more likely to cite your content.
AEO (Answer Engine Optimization)
Preferred by practitioners who want to emphasise that these platforms answer questions rather than listing links. Profound, one of the leading tools in this space, uses AEO. It describes the same underlying goal: get your brand cited in the AI-generated answer, not just indexed somewhere in the web.
AIO (AI Optimization)
This is the broadest term, and not just cause it has ‘IO’ in it. It covers both of the above and also includes optimizing how AI understands your brand across all surfaces, including your own products.
Thing is, they’re not really competing frameworks. They’re different ways of naming the same shift, which is AI engines don’t rank pages, they synthesize answers. Traditional SEO got you on a list of ten links whereas this gets you into the answer itself or leaves you out of it entirely.
The difference is brutal. LLMs cite an average of only 2 to 7 domains per response while Google’s ten blue links gave you ten chances. More alarmingly, approximately just 12% of sources cited by ChatGPT and Perplexity overlap, meaning visibility on one doesn’t automatically give you visibility on the other.
You’re either in the answer or you’re not in the conversation.
REO Speedwagon
Has nothing to do with this. Just good vibes.
What (we all) Can Do About This
Here are the specific things leading research says can ‘move the needle’ for AI search discovery.
Quote Real People by Name
An AI engine synthesizing an answer wants to quote someone authoritative, so give it something to quote. This also applies to your own content: if you’re writing a B2B article and you want AI to cite it, include a named expert saying something specific and true.
Structure Content To Answer Questions Directly
AI queries average 23 words, according to research cited by Mersel AI. “Best CRM” is a keyword. “What CRM would work for a 20-person B2B company with heavy Salesforce integration needs and a budget under £500 per month” is an AI query. These produce different results from different domains. Write content that answers the long, specific version of the question, not just the short one.
Keep Content Fresh
Approximately 85% of AI Overview citations were published within the last two years, and recently updated content appears 4.3 times more often in AI answers than stale content. That blog post you published in 2021 and haven’t touched since isn’t being cited regardless of how good it was.
Get On The Platforms That AI Engines Already Trust
For better or for worse - the most cited platforms across AI engines are Wikipedia, Reddit, YouTube, Quora, and LinkedIn (God help us all). This doesn’t necessarily mean you should spam Reddit, but it means your brand needs to exist and be discussed in places that AI systems have learned to trust. Reddit citations in AI responses increased 450% between March and June 2025, partly because it’s constantly fresh and perceived as ‘authentic’ by LLMs.
Fix Your Schema Markup
This is technical but not complicated. Structured data tells AI systems what your content is about in a format they can parse without inference. If you’re not sure whether your site has it, ask your developer or run a free schema test on Google’s Rich Results Tool. It takes maybe an afternoon to fix and improves your chances of being understood correctly by every AI engine simultaneously.
Awesome, you made it this far!
What To Watch For
My honest summary of where this is going, based on what I can scrounge up. Gartner predicted in 2024 that traditional search volume would drop 25% by 2026. Separately, they projected up to 50% decline in traditional organic traffic by 2028 for brands that don’t adapt. Whether those numbers land exactly on schedule is debatable, but the trend isn’t.
Google traffic actually increased slightly in 2025 by 0.8%, so reports of its death were premature and largely written by people who had something to sell you. However, informational queries, specifically the type that B2B buyers use to research decisions, have migrated significantly. Publisher sites have seen search decline from 44% to 37% of their traffic between 2022 and 2025. The erosion is real, and it’s happening fastest in exactly the content categories that B2B marketers have invested in most heavily.
The window to get ahead of this is not closing tomorrow, but the first-mover advantage is already visible. The brands that establish consistent AI citation authority in their categories now will have it compounded over them by brands that arrive later.
Something To Try This Week
Run that test, the one from the top of this article.
Open ChatGPT and Perplexity
Type the questions your buyers/target audience actually ask before they buy from you.
Write down who shows up. If it’s you, understand why and do more of it.
If it’s not, pick one piece of content, add three named statistics with sources, add one attributed expert quote, update it with current data, and test again in maybe thirty days.
The terminology will keep changing but the underlying principle won’t - AI engines cite sources they trust, understand, and can extract clear answers from. Write to be understood by a very intelligent system that has no time for vague language and a very strong preference for evidence.