Lead story – GEO or AEO – where do brands go from here?
Reaching buyers matters – always has, always will. But increasingly, we need to reach the buyers and the bots, or reach the bots to reach some of the buyers, etc.
Sound fun? Barb has the story in Generative Engine Optimization isn’t the new SEO. Here’s what brands need to do instead… And it’s a good one.
First off, re: GEO or AEO, can we just pick a term and be done with it? Barb defines AEO as a subset of GEO, which makes sense, but I’m not convinced most of us need two wonky new phrases for this. Let’s just go with GEO…
Okay, glad that’s settled. Let’s move on to Barb’s core point: the AI bots as brand/content gatekeepers (and, I’d argue, attention gatekeepers, at least to a point). Barb writes:
AI answers are becoming today’s new first impression, and brands risk losing visibility before a potential customer ever reaches the website. What to do? They turn to Generative Engine Optimization (GEO)… The problem is that many are treating AI search like Search Engine Optimization (SEO) when it first came out. But the smart ones are focused on building real authority and credibility, making them the answer customers seek. But it’s not an easy road.
Sounds like the right overall pursuit. But Barb throws the caution flag: don’t try to game the system, SEO style.
When brands started working with SEO, we saw a lot of keyword stuffing in content and metadata, invisible text on the page for Google to read, and real backlink farms. Today, we’re seeing the same behavior, just different technology.
With GEO, we get fake citations, AI-generated recommendation spam, and fake expert ecosystems. And that’s just to get surfaced in the Large Language Models (LLMs). The website still matters, and what’s happening, says Wheeler, is that smart brands are trying to feed both masters.
This is still an unfolding game/obstacle/opportunity, but one thing is clear so far: repeated mentions of your brand in an authoritative context are a big key. Barb extracts this lesson from her talk with Andrew Wheeler, CEO of Skyword:
Wheeler believes that AI systems will increasingly evaluate patterns of credibility. They are going to learn which brands are repeatedly cited, such as how trusted third-party publications reference them, which experts consistently appear in authoritative contexts, and whether they are real. What brands demonstrate expertise across different channels and conversations? Then the AI and its prediction engines will rely more on the sources that are repeatedly cited to know who’s most authoritative.
So we have another fork in the road – pursue credibility and authority, or obsess about gimmicks and mentions. Barb quotes Wheeler:
The competitive advantage in AI search is not content volume. It’s really about whether the AI system recognizes your brand as a credible authority…That should be the full point, not that if I just create more content for what people are prompting and searching for, I’ll be okay. It’s really important to think about it more from an authority ecosystem.
This gels with a piece I wrote off the back of Salesforce Connections: Salesforce Connections 2026 – SEO is out and GEO is here. But can commerce brands up their game? In that piece, I did a stats reveal that even the practices of trying to make an LLM-ready version of your web site doesn’t seem to be much of a differentiator right now. With SEO, there were always tactics that made sense (e.g. smart titles for articles, answering reader questions), and gaming tactics (keyword stuffing, shady link swaps) that worked but risked punishment.
However, I don’t agree that black hat gaming for SEO – and now GEO – doesn’t work. I wish it didn’t, but for example: back in another era, Yahoo Answers showed up way too prominently in Google search as as supposed authority site, even though a lot of the “Yahoo Answers” were user-generated junk, on the sub-basement level that makes modern Reddit look like Wikipedia. We’re seeing the same now, to a lesser extent, with Reddit, where good insights surface alongside craptastic blathering.
Elevating Reddit as an authority site in a blanket way is a mistake; many AI search bots are making that mistake now (and yes, Reddit can be gamed for AI, as 404 Media just reported – and once an LLM is told/”determines” that a site is an authority to be trusted, the LLM can be very gullible within that scope of blanket trust).
I badly wish these prince of SEO/AEO darkness pseudo-authority tactics didn’t work, but they work more than we want to admit. And: if you are striving to be an authority site in an area that Reddit covers, good luck with competing for those citations!
Putting that rant aside, what Barb and Wheeler laid out here is clearly the proper long game for brands, based all that we know at this point.
Diginomica picks – my top stories on diginomica this week
Vendor analysis, diginomica style. Here’s my three top choices from our vendor coverage:
Event season (will it ever stop???) winds down –
A couple more events before we call it a season… Alyx was with Dynatrace in London: Why Dynatrace says AI changes what your logs are for. Also see: Alyx’s How Storio rebuilt its observability to move at AI speed with Dynatrace – and why the hard part was cultural: “After Dynatrace’s Davis AI caught a multi-million-euro Black Friday failure before it reached checkout, Storio’s Alex Hibbitt explains what came next.”
Meanwhile, I was at Workday Elevate New York City, where I heard some of the most incisive/revealing AI customer stories I’ve heard from the keynote stage this year: Workday Elevate – what kind of AI do you want? Customers share AI success tactics – and how employees bought in.
A few more vendor picks, without the quotables:
Jon’s grab bag – Stuart has the news we missed in The long and the short of it – the week in digibytes. I’m not a fan of the Hiroshima analogy, but then again, better to get out in front of such issues than chase them from behind… Katy has the story in Avoiding an AI Hiroshima – can a Kore.AI and Atos partnership secure sovereign agentic AI? Whenever you can sneak ABBA onto diginomica, that’s fine by me. Stuart found a way in Something for the weekend – AI ABBA-style. Don’t just say thank you for the music, pay for it, says Björn Ulvaeus.
Best of the enterprise web
My top eight
Cisco’s 90,000 Employees Will Soon Have Personalized AI Agents – This theme, of large organizations countering tokenomics by building their own stacks, and ‘rightsizing’ their model use, is going to be a big theme for the rest of the year. As per Entrepreneur.com:
Cisco’s system automatically chooses the right AI model for each task. ‘It knows which tool is most effective and most efficient,’ Patterson said. Much of the AI’s infrastructure is built on-site, giving Cisco more say over costs. ‘We feel like that’s the most efficient way is to build our own AI stacks, which will go out and query the different models based on the particular use case,’ Patterson said.”
The burning question, of course: where does this leave enterprises without the internal resources to build/manage their own AI stacks?
69% of enterprises share AI agent credentials – I hope Louis Columbus isn’t planning any long vacations away from the AI security beat this summer…
DeepSeek’s real impact happening now – Larry Dignan builds on the ‘bring your own model’ theme: “Enterprises are managing costs and that means model routing. The rise of good enough models has started to commoditize LLMs.” Indeed – and that’s the good news part of the story. But here’s the (macro-economic) pit of this particular avocado: “The economics of LLMs aren’t going to hold up just as Anthropic and OpenAI chase IPOs and valuations topping $1 trillion. Something will give, but it’s a question of when.”
The real AI advantage: “The next wave of value from AI will flow to those leaders who redesign business models, eliminate friction, and build organizations that learn faster than the competition.” McKinsey might be stating the obvious here, but it’s probably worth repeating.
APIs aren’t dead. Here’s where MCP fits alongside them – different approaches, both still relevant. Let’s get our geek on…
Is Europe Getting AI Wrong? – Via Tech Policy Press, one of the best AI-related podcasts I’ve heard in a long time.
Analyst Relations gut check – live feedback sessions on AI-created AR “solutions” – the video replay of a super-experimental live session I did with Bonnie Tinder and Josh Greenbaum, where they both gave honest live reactions to my AI-generated content on the future of analyst relations. But was the content slop or not? What did the audience think? Why did I do it? All is revealed…
The Loop Trap: Why Cheap Tokens Don’t Mean Cheap Tasks – Vijay Vijayasankar has been blogging up a storm, with a series of must-read AI posts. I take a bit of an issue with the premise: token costs (from frontier models) certainly aren’t stable, nor do I think they are cheap – even when agentic loops are not heavily used (some of the token overruns at big companies are caused by things like PDF-to-slide conversions, believe it or not). See: Claude and ChatGPT Are Getting Too Expensive, Even for Microsoft, or Arora’s CEO, who believes token costs need to fall by 90 percent. Still, Vijayasankar’s central point stands: agentic loops (and other guardrail-like refinements such as agent cross-checking) are where the extreme token cost pileup happens. Where do we go from here? Model right-sizing and open source models are an important piece of the puzzle, as well as the right systems around those models, and operational discipline as well. Vijayasankar issues his take on that (and more) in recent posts, including The Real AI Race Isn’t for Better Models. It’s for Pricing Power. Let’s follow along…
Whiffs
Meta: so many desperate? vision-deficit? forward-thinking AI “ideas” these days:
Oh, and this one: don’t just make a device that creeps like, but actively cater to them by reducing the ability for others to detect their spying:
Meta’s new AI creep glasses will record 24-7 without a light pivot-to-ai.com/2026/07/09/m…
“The real use case is the market of creeps and weirdos who film people out and about without their consent.”
-> Metaworse?
— Jon Reed (@jon.diginomica.com) 2026-07-12T21:39:12.268Z
I could have done an entire Meta whiffs section this week (Bob Turek found another one), but I don’t want to leave LinkedIn out:
Maybe things are looking up after all… Improve your social life while virtue-signaling and doing brand peacock things? What’s not to like?
See you next time… If you find an #ensw piece that qualifies for hits and misses – in a good or bad way – let me know in the comments as Clive (almost) always does. Most Enterprise hits and misses articles are selected from my curated @jonerpnewsfeed.