This is the replay of our first Raleigh–Durham HubSpot User Group session, Answer Engine Optimization: Search changed. Did you? The core reframe: AEO isn't about earning the click anymore — it's about earning the citation. AI engines increasingly answer your buyer's question outright, without ever sending a click your way, so the job becomes being the passage they quote and credit, not just the page they rank. And the biggest lever isn't even on your own site — branded mentions and earned media correlate with AI visibility roughly 3x stronger than backlinks.
Watch the full session above, or read the transcript below.
Want the data companion — what we found testing hundreds of NC B2B sites for AI citation? That's in the full study: What 471 NC B2B Sites Reveal About Getting Cited.
Watch the Replay
Full Transcript
Welcome to the first in a series of hopefully very successful events for the new Raleigh–Durham HubSpot User Group. This first session — Answer Engine Optimization: Search changed. Did you? — is focused on answer engine optimization: things we're seeing in our own data sets, things we're implementing for our own clients, and things you can put to work in your own marketing.
Before we get into that, a bit of context. The Triangle — specifically Raleigh — had a HubSpot User Group several years ago that effectively went defunct. About a year and a half ago we reached out to HubSpot to ask if we could help facilitate a new group and foster some new relationships in the community. So hopefully this is the first in a series of events covering HubSpot, HubSpot-adjacent skills, marketing, sales, customer service — the range of skills that come into play when you're working with a CRM. The intention of this group is to educate, connect, and produce real examples for the people in it.
My name is Matthew Deal. I'm the Chief Marketing Officer and one of the co-founders of Vaulted — we were a merger of two organizations about three and a half years ago. I work frequently with marketing CMOs and sales leaders, and I have the pleasure of helping foster this group. Today my goal is to take some of the abstract out of the theory and put it into practice.
Today we're covering AI search and what to do about it, in four sections: why some of your buyers have already decided on you by the time they reach you through an AI query; how AI engines pick who they cite; what's oversold — and what I'd argue is a lot of noise — versus what actually works; and what to do Monday, your most immediate first actions.
First, a definition: AEO is earning the citation
I'd argue AEO is earning the citation — earning the ability to rank and be seen for the terms that matter. You'll sometimes see it called GEO (generative engine optimization), and sometimes people use it interchangeably with SEO. Even the term itself is a little fluid and fast-moving.
One big distinction: in SEO we're used to optimizing pages. The page is the unit of analysis and the point where we focus most of our effort. AEO is much more about optimizing passages within pages — creating quotable opportunities, building authority, and ultimately winning the ability to be the answer to a question. That last part is an old SEO concept — being the answer to a user query — carried into a new context.
Why your buyers already decided
Buyers are, in a lot of ways, already decided by the time they use AI tools to reach your site. Around 61% of the buying journey is done by the time a buyer contacts sales, and about 94% are using AI as part of that research — a deep-research query in ChatGPT, a prompt asking for the best version of a product or SaaS tool, and so on. We already knew a lot of B2B buyers were doing proactive research long before they reached out to sales. That trend continues, now with AI as part of the process.
The big shift is what happens at the query itself. When AI answers the question, your link might not appear at all. Roughly 68% of Google searches now end without a click — the user gets their answer on the results page or straight from the generative tool they're using. The "visit" you used to earn is now a question you're either the answer to, or invisible to.
That can sound gloomy — less traffic, shifting from SEO to AEO. But there's evidence that while the total traffic from AI is smaller, the traffic that does come through tends to be more qualified. These are educated buyers, which tracks: these tools aggregate a lot of information from a lot of sources. If I don't know a topic, I can run a deep-research report and get a solid synopsis quickly — and we see that same behavior in B2B buyers.
How AI engines pick who they cite
A few core principles here. Again, AEO is less about pages and more about passages — the specific parts of a page an LLM can scrape and cite. We'll get to the myth around "chunkable content" in a moment, but in this context chunkability really means having passages that feel quotable and citable. So when we optimize a page, we're looking for opportunities to create citable, succinct answers for LLMs.
The other component is a paradigm shift from keywords to prompts. Often one question is really made up of several composite questions or prompts. "The best CRM for us" contains a lot of smaller terms used along the buyer's journey — and we need to speak to each of them in a way that shows up in LLMs. The idea of writing a single blog post for "best CRM for us" and suddenly appearing in results next week isn't realistic. It's more useful to think about ranking for the smaller terms that make up that primary one — similar to a long-tail keyword strategy.
There are effectively two ways to appear in an AI answer. The first is that your brand shows up in the answer — the model uses you as a source but doesn't name you. That usually doesn't produce clicks; it's the first rung of the ladder. The higher tier is being cited — your page is named as the source. That's the goal, and that's where traffic and conversions come from. Being mentioned is part of the journey, but we want to be named.
Here's what that looks like. Same question, two very different outcomes. Mentioned: "For best results, segment your lists before automating outreach — a step recommended by most implementation guides." No link, no credit, zero clicks. Cited: "For best results, segment before automating — per RDU HUG's implementation framework, rduhug.com." (That domain isn't real — please, nobody go register it.) The difference is that the cited version names an authoritative source and links to it, which lends the content more authority. It takes the answer out of the amorphous "most guides recommend this" and attaches it to a specific entity.
What's oversold — and what actually works
Alongside the fundamentals, there's a lot being oversold right now — products and esoteric, unproven methods that promise to rank you fast. A lot of it sounds better than it works.
One rare, useful thing: in May of this year, Google actually published an official guide to AI search — which is unusual, since most SEO knowledge comes from large-scale statistical studies rather than Google saying things outright. In it, Google effectively said AEO is still SEO, and debunked several heavily promoted tactics:
- llms.txt. The idea was a file at the root of your domain that LLMs would read to understand your organization. Most LLMs ignore it. It doesn't appear to be a factor.
- Special schema / microdata. The claim that schema markup gives you an AEO boost. To be clear, schema has real benefits for SEO and user experience — we literally build schema plugins ourselves — but I'm not sure even we'd claim it's an AEO lever.
- Chunking and AI-rewriting. The idea that your paragraphs are "too long" and content needs to be rewritten smaller and more parsable for LLMs. There's not much evidence for it. This is different from what I said earlier about chunking: being a citable source means creating passages that can win an answer — not wholesale rewriting your content to be smaller.
Important caveat: those rules are what Google says apply to Google. Other tools — ChatGPT, Perplexity, Copilot — retrieve differently. What's dead for one can still matter for another.
So what does get cited? A few specific measures. First, adding original statistics and data. A lot of organizations are hesitant to publish their own numbers, even small data sets — "29% of our customers enjoy a callback after their appointment," whatever it is. It's authoritative, it builds data, and it leads with data first. Second, naming experts — quoting named authorities the way a journalist would. Third, citing your own sources. One of the best ways to lift lower-level pages is to consistently reference your sources and build your own authority on a topic.
What works, per a recent 2026 meta-analysis: answer first — lead each section with a direct, quotable answer, then expand. Self-contained passages. Question-style headings. It all comes down to establishing yourself with confidence and authority so LLMs read you as credible. These engines are hunting for credibility, trying to decide who should be authoritative — and we want to make that answer obvious.
One weird exception worth the tip: if you're not in Bing's index, you won't exist in ChatGPT or Perplexity. Unblock the AI crawlers and submit your sitemap there. That said, I rarely see a site with a catastrophic technical block — if you've already seen yourself in AI results, you probably don't have that kind of issue.
But the biggest lever isn't on your website. Polishing your own pages only gets you so far — it's the floor, not what moves citations. Branded mentions correlate with AI visibility about three times stronger than backlinks. Where we see this play out: Reddit, YouTube, Wikipedia, and review sites like G2 and Capterra if you're on the software side. Most citations trace back to earned media. This is classic Google, really — putting the emphasis back on producing high-quality content people actually want to consume. That hasn't changed. What's renewed is the emphasis on where those trusted mentions live.
What to do Monday — and what to ignore
So, what's realistic in the short term? First, review your top pages and make sure each has a quotable answer that's succinct, authoritative, and backed by data and references that would hold up elsewhere. If you ever wrote a paper in grade school, the logic for finding a source is the same. Second, lean into your original data — it's far easier now to run statistics with AI tools than it was a few years ago, so use that to build original data and sources. Third, make sure you're reachable in AI search engines: nothing blocking the crawlers, Bing indexing you. Again, if you've seen search traffic, catastrophic issues are rare.
The tool in your stack: HubSpot's AEO tools
Since this is a HubSpot User Group, we should talk about the HubSpot part. Every HubSpot account now includes their new AEO tool for managing and tracking your AEO — and I have strong opinions about it.
First, it's useful, and I'm glad they built it. Prompt tracking is useful. Brand-visibility monitoring is useful. If you don't know where you stand, you don't know which direction to go — running AEO without a tracking tool is like walking around blindfolded. The inverse: the generative side is pretty weak. It tends to push listicle-style content, and the AI copy isn't as robust as what you'd get from higher-tier models or more capable agentic setups. So we recommend pairing it with an external SEO tool — Semrush, Ahrefs, something like that. We tend to source a lot of our original AEO data out of an SEO tool, then track it on the HubSpot side. Don't lean on the generation alone.
A couple of things to keep in mind. There's no such thing as a real "AI visibility score." HubSpot hasn't cracked it, and neither has anyone else — Semrush, Ahrefs, every one of them has a proprietary model, and none of them match. Treat that number with a grain of salt; it's not even comparable account to account. Watch the trend, not the decimal.
And don't let the tool turn you into a listicle machine, or into someone publishing content that doesn't fit your brand. Not every AEO technique fits every brand. Buyer's guides and comparison frameworks aren't good fits for high-end B2B brands, and often aren't good fits at all. The brands that win build content that sounds like them — and happens to be extractable.
The bottom line
If you've worked in SEO and ever wanted to rank before, search may have changed — but you still can. A lot of this is old SEO fundamentals: produce genuinely good content, translate it into the world of AI, and think about intelligent ways to scale it up and be seen in the answers.
To close: the RDU HUG is back. This was the first event, and I'm grateful to everyone who came out for this first session. We're doing a new event in July — in person this time — with details coming soon. And this isn't a Vaulted thing; we're just the facilitators. It's not a pitch fest. It's a group where we can all network and learn, so we're looking for speakers, topics, and ideas. If you've got one, reach out and let us know. Thanks, everyone.
