Nobody Should Be Charging You Extra for AEO

I had Romana Kuts on Barely Shipping last week, and a line she dropped halfway through is going on a t-shirt. She runs SaaStorm, a B2B SaaS agency out of Europe, and she took one climate tech client from near-zero search visibility to roughly £3M in pipeline inside twelve months. We were deep in a conversation about LLM optimization when she said the thing that explains the fastest-growing grift in our industry:

“I was adding FAQs before LLMs were a thing. It just keeps coming back to the same good old SEO.”

That line is funny on its own, but it’s also the entire argument against the new acronym everyone is busy monetizing.

The add-on

You’ve seen the move if you’ve shopped for an agency in the last year. The pricing page looks clean and transparent, SEO is included, content is included, and then near the bottom there’s a fresh block that looks like it was always there: an AEO add-on for another two grand a month. Romana’s reaction to this is the correct one, and she didn’t soften it for the podcast.

“I hate seeing when agencies have everything transparent, and then out of nowhere they add this new block on the pricing page. They’ll do SEO for you, everything included, but if you want to also be in LLMs, you need to pay two grand a month more. What do you mean, guys?”

What’s actually happening is that they’re billing you twice for one piece of work, because the thing they’re calling AEO is the thing good SEO already was. They just found a newer acronym to attach a markup to. Romana watched this happen in real time across her own book of clients: she was doing genuine SEO for them before any of the LLM stuff kicked off, and when it kicked off, those same clients started surfacing in the answer engines without her changing a thing. The structured, well-organized pages that earned rankings in Google got pulled straight into ChatGPT and Perplexity, because the models are reading the same content Google was already rewarding. That’s not a downgrade of AEO. That’s the whole point of it.

What we were actually being lazy about

The harder thread in that conversation, and the reason I reached out to her in the first place, was a blog post she’d written with a title I couldn’t argue with: your B2B blog posts weren’t even that good before AI.

For years you could rank without doing the work, and so most of us didn’t. You could skip the original video, skip the custom images, skip the interactive calculator, skip the FAQ that actually answered the question, publish a wall of text, and still land in position one because the bar was low enough to walk over. Romana admits she coasted on that, and I’ll admit it too. Some of the content I shipped at Copy.ai was built for the crawler rather than the reader, the page ranked anyway, and the job was technically “done” even though nobody asked whether a human who landed there learned a single thing.

The extras everyone is now repackaging as AEO were always part of good content. They were just optional, because mediocre work still won. As Romana put it, the difference now is that there’s nowhere left to hide:

“We were allowing ourselves to be a bit lazy, because we knew that even if you don’t add that video or those images, you can still hack the system and get to the top. But you always needed them. Let’s be honest. We were just like, whatever, I can rank without this.”

AI didn’t invent a new requirement. It removed the spots where lazy content used to be invisible.

The quality bar was already on the floor

This is the part where I lose a few people, so let me be direct about it. When someone calls a piece of content “AI slop,” the honest question back is: compared to what?

The average B2B blog post in 2023 was already generic, keyword-stuffed, and written by a freelancer who had never once opened the product they were describing. AI didn’t lower the quality bar, because the quality bar was already lying on the floor. What AI did was make it cheap and fast to produce content at that same floor, which is a different problem and a much more honest one to name. A lot of the companies loudly worried about AI slop diluting search spent the previous decade publishing the human version of exactly that. The mediocrity isn’t new, the speed is, and the fix is the thing we should have been doing the entire time: make it accurate, make it clear, make it genuinely helpful. Hit those three and you can argue about style all day. Miss any one of them and it isn’t good content, no matter which acronym you slap on the invoice.

The extra mile is the moat

Romana described what her agency does differently now, and it’s the most practical version of the “human moat” idea I’ve heard anyone articulate. The article still gets built so the algorithms and the LLMs can read it cleanly, because that’s table stakes. But then they layer in the thing a person notices and a crawler doesn’t: a Loom right under the H1 where a real subject matter expert answers the question off the cuff, ums and all, or an interactive chart that moves when you hover over it. The reader might still get their actual answer from ChatGPT, but if they click through to the page, they find something that was obviously made for a human and not just for an index.

“Even if the entire article is written for algorithms and search, I want humans to find something that was actually made for humans. It’s normally humans for humans. That’s why unpolished Looms work so well.”

This is one of the four principles of Systems-Led Growth stated in plain language: lived experience is the only moat left. A model can generate any wall of text you ask it for, but it can’t record the thirty-second Loom of your expert saying the one thing only they would know to say. The structure is what gets you found. The human layer is the only reason anyone stays once they arrive.

How to tell an operator from an upsell artist

If you’re evaluating an agency right now, three questions will sort the people doing the work from the people inventing line items. Run all three before you sign anything.

Ask whether AEO is a separate charge

If it shows up as its own block on the pricing page, ask them to tell you exactly what’s inside it that competent SEO wouldn’t already cover. A real operator will explain the genuine extras (structured data, entity coverage, a content refresh cadence) and fold them into one number. An upsell artist will get vague, because the honest answer is that they’re charging you a second time for FAQs and schema you were already paying for.

Ask what they add that a model couldn’t

Anyone can export a rankings dashboard, so don’t let that be the demo. Ask what goes on the page that Claude or ChatGPT couldn’t have generated on its own. If the answer is “better keywords,” keep looking. If the answer involves capturing your team’s actual expertise (the Looms, the real numbers, the opinions your competitors won’t say out loud), you’ve found someone who understands where the moat actually sits.

Ask whether the content would survive being read

Not skimmed by a crawler. Read, start to finish, by the specific person you’re trying to sell to. Pull up a piece they’ve published and read it as that buyer. If it answers the question directly and tells you something you couldn’t have gotten from a generic prompt, that’s trust being built. If it’s a wall of throat-clearing with the answer buried at the bottom, you’re paying for traffic, and traffic that doesn’t convert is vanity.

The agencies that make it through the next two years won’t be the ones who coined a new acronym to bill against. They’ll be the ones who quietly did the work everyone else skipped back when skipping it still worked. Romana’s been doing that since 2014, while the rest of the industry is only now getting the invoice.

---

This piece came out of my conversation with Romana Kuts on Barely Shipping. The full episode is on YouTube. If you want this thinking applied to your own GTM system, that’s what I build.