d s simon - Spin Sucks

TL;DR

Our first Visibility Engineering™ webinar ran out of time long before it ran out of questions. They clustered around four sticking points: building anchor hubs, keeping the message consistent, knowing what AI can read, and measuring the work. The short version: one public, well-built anchor hub beats 30 hidden pages, consistency beats volume, and if AI can’t crawl it, it can’t cite it.

Key Insights

Comms teams already believe owned and earned media drive AI visibility. What holds them back is permission, workload, and old rules.
Anchor hubs should be public. Most organizations need three to five, not one per FAQ.
“Same language” means the same message, not the same sentences.
Paywalled coverage and anything trapped inside a video are invisible to AI.
Executive bios have to match everywhere, or AI blends them together.
Measure on the ladder (absent, mentioned, cited, featured) and expect signals in three to six months.

Our first free webinar on AI visibility, Beyond Media Hits: The Basics of Visibility Engineering™, ran out of time long before it ran out of questions. Gini Dietrich promised to answer the rest.

This is that.

But first, a pattern worth noticing.

We polled the room: Which media type matters most for showing up in AI answers? Seventy-six percent said earned media, which aligns with D S Simon’s survey of PR executives. Yet in both that survey and Sword and the Script’s, only 3% of respondents said PR owns AI visibility.

The questions explain the gap. Almost nobody asked, “Is this real?” They asked, “Am I allowed?”, “Who has time?”, and “Do the old rules still apply?”

Comms pros know owned and earned media drive AI visibility. They’re stuck on permission, workload, and habits built for a world that’s gone.

Visibility Engineering in 60 Seconds

The AI tools almost never agree with each other. Orbit Media tracked more than 13,000 citations and found ChatGPT, Claude, Gemini, and Perplexity cited the same source only 1.7% of the time.

You can’t rank, buy, or game your way into those answers.

As Gini put it, “The only thing that works across all of them is being the most consistent, most corroborated source on your topic, everywhere that people and machines go looking for proof.”

When AI surfaces you, it happens on a ladder: you’re absent, mentioned in a list, cited as a source, or featured as the answer, in your framework, language, and data.

The first step up is the anchor hub: a single, deep, defensible page that functions like “your own Wikipedia page on your website.”

A true anchor hub does five things:

It’s framed as the buyer’s question.
It defines the problem clearly.
It names your method or framework.
It shows evidence and outcomes.
It includes quotes and data that AI can cite word-for-word.

Miss two, and you have a page, not a hub.

The Anchor Hub Questions

Most of the questions landed here, which tells us the concept stuck.

Can it be hidden? Several attendees asked whether a hub could be a hidden page or back-end schema only. Technically, yes. Gini doesn’t recommend it, “If you are doing the five things I described in your anchor hub, it’s going to be as valuable to a human as it is to the AI.”

Make it public and add schema on the back end so it’s crawled accurately. The one exception is supporting material—it’s fine to put video and podcast transcripts on hidden pages just so AI can read them.

Do we need one for every FAQ? No, and nobody has time for that anyway. One hub can hold a whole set of FAQs.

The Spin Sucks PESO Model® page answers its number-one question and includes FAQs on implementation and certification on the same page. Most organizations need three to five hubs.

“Not 30, not 50, not a whole website of them,” Gini said. The same goes for big product catalogs: a hub isn’t a product page, so build hubs around categories or problems and link the product pages back to them.

Where do we start? With a page you already have. Find the blog post or landing page that answers a question buyers ask repeatedly, and run it through the five points. During the webinar, Gini scored HubSpot’s “What Is Inbound Marketing?” page four out of five. It nails the buyer’s question, the problem, the named framework, and the citable definition. What’s missing is evidence and outcomes, the most common gap, even on good pages.

Then build breadcrumbs. Every related post, transcript, or release that links back helps humans and robots recognize the hub as the authority. Where it sits in your navigation matters less than making sure it isn’t gated, buried, or orphaned.

The Consistency Questions

If the anchor hub is the foundation, consistency holds the building up. It’s also where old PR habits fight back hardest.

Does “same language” mean word for word? One attendee worried editors would reject bylines that echo website copy. It means the same message, not the same sentences. Keep the substance consistent: the same framework name, core claims, proof points, and expert. When those line up, AI treats coverage as corroboration, even without a link. “We used to have to ask for the link back,” Gini said. “We don’t have to worry about those things anymore.”

Every byline should map back to a hub topic and be written fresh for that publication’s readers. The editor gets an original piece, and AI gets one more credible source saying the same thing about you.

Another attendee argued hubs are promotional by nature, which makes them hard to echo in editorial pitches.

Gini disagreed. A pricing page is promotional, but a page answering “What is inbound marketing?” is educational. “I would think about your anchor hubs as educating and not promoting.”

Do executive bios matter? Yes, and they have to match everywhere. “If you have a bio that’s on the website that’s different from your bio on LinkedIn, AI doesn’t know which is real,” Gini said. “So either it will conflate and make things up, or it just ignores the bio entirely.”

Case in point, AI recently told her she was an editor at PR Daily. She isn’t. The source was a speaker bio from a July PR Daily workshop that didn’t match the bios Spin Sucks uses everywhere else.

If it can happen to the person who teaches this, it can happen to your CEO.

How often should hubs be updated? At least quarterly, and immediately when a fact changes. According to Muck Rack’s “What Is AI Reading?” report, more than half of the journalism AI cites was published in the last 12 months.

“This is a system you have to feed,” Gini said.

When she asked Claude about the PESO Model, it said the model was last revised in 2020 (it was revised this February). Nothing was invented. It was just old. Older sources tell an older story.

The “Can AI Read This?” Questions

The short answer, in Gini’s words: “AI reads text, period.”

Is paywalled coverage useless to AI? Gini’s answer was one word: “Yes.”

If AI can’t crawl it, it can’t cite it. The placement still builds credibility with subscribers, but make sure the same message lives somewhere AI can read. And no, you can’t just upload the full article from your subscription—that’s a copyright violation. If the publication sells you reprint rights, go for it.

Otherwise, write about it in your own words and link to it.

What about YouTube? AI can’t watch video, so captions burned into the video don’t count. It reads the transcript and description, so upload a cleaned-up transcript and add chapters titled as the questions buyers ask. Then give every video and podcast a text companion on your own site that links back to your hub.

Why doesn’t AI cite LinkedIn? It’s starting to read LinkedIn newsletters, but in our experience, it isn’t citing them yet. The workaround is to publish on your own site first, then republish on LinkedIn.

And as Gini warned, “Six months from now, I could do this exact same webinar, and it’ll be completely different.”

The Ownership and Measurement Questions

This last group is really about closing that 3% gap: who owns the work, and how to prove it’s working.

Does SEO still matter? Yes, but it’s a different game. In Orbit Media’s study, only 10% to 30% of the sites AI cited also ranked in Google’s top 10.

So, you have to do both.

How do you measure it? No vendor has nailed this yet, because nobody can see what real people type into AI.

Gini’s alternative takes a spreadsheet and an hour a week.

List the 20 questions people ask most about your business. Once a week, ask all 20 in at least four AI tools in incognito mode, and note where you land on the ladder.

Then set expectations: signals take three to six months, and compounding takes six to 12.

“It is a long game,” Gini said, “but it’s the only game that works across any of these tools.”

What if PR doesn’t own the website? Partner with whoever does. And if they aren’t open to it, Gini’s offer stands: send them the webinar recording and let her be the bad guy.

Can a startup compete with established brands? Yes. A startup can’t out-age the competition, but it can out-consistent it. AI rewards clear, corroborated, recent sources, not longevity.

Start with one hub, make your founder the visible expert with one bio everywhere, and go after the trade publications your buyers read before chasing national outlets.

The Hub Is the Easy Part

Building one good anchor hub takes a few weeks. What teams underestimate is the next 12 months.

Every byline, release, bio, and podcast has to tell the same story, across teams that don’t share a calendar, while someone in leadership asks, “Okay, what’s next?” about results that take six months to show up.

That’s where Visibility Engineering breaks down. Not in the hub, but in the handoffs between comms, the web team, and leadership, and in the measurement that’s supposed to prove it’s working.

Find Out Where Your Brand Sits in AI Answers

Want to do this hands-on?

Join Gini for Where Your Brand Sits in AI Discovery on Thursday, Oct. 22, from noon to 2 p.m. ET. You’ll leave with an AI visibility audit, an anchor hub outline, a short media list, and a measurement plan.