A sociology degree, a difficult job hunt, and a friend who needed someone to run Google Ads. That’s how Aaron Wilson ended up specializing in SEO.
Years later, he made the case to Tenable that his discipline deserved a far bigger investment. He’s been there just over a year full-time, and his thinking on search is sharper than anyone I’ve spoken to in this series.
In this conversation, we covered the parts of the traditional SEO playbook that still hold, what cybersecurity’s technical nature demands of content teams, why gated content is undermining buyer discovery, and the shift that most practitioners, if they’re being completely honest, aren’t prepared for.
You didn’t come up through a traditional marketing path. How did you end up specializing in SEO?
Right place, right time. But the skills compounded. After college, a good friend saw me struggling to find meaningful work with a sociology degree and offered to teach me Google Ads while he was growing a boutique agency. That statistical training turned out to be genuinely useful for the analytics side of this work.
After that, I spent the next five or six years managing digital advertising at different organizations. A mentor at one of those employers started teaching me SEO after their SEO person left, and it really snowballed. I worked at multiple organizations, and my specialization deepened with each one.
Eventually, Tenable brought me on as a contractor. After a while I made the case for a more substantial, full-time investment, (partly because of how important AI was making the discipline). They agreed, and I’ve been there full-time for just over a year now.
How much of the old SEO playbook still holds in a world where LLMs are doing much of the research?
Yeah, read that again.
The underlying infrastructure behind how AI platforms train their models and crawl the internet is built on the same foundations search engines created. Authority matters more now than it ever has. Authority comes from external entities recognizing valuable content, not from gaming the system.
None of the technical best practices around web infrastructure, metadata, and content structure have changed. What has changed is the stakes. The weight put on AI has made those fundamentals more important than ever. The playbook hasn’t become obsolete, but the new system doesn’t tolerate the same shortcuts and hacks the old one did. Black hat tactics have a very short runway left.
What does a piece of content created to be cited by an LLM look like?
Research firms have done studies on which content types get surfaced most in LLM responses, and the pattern is consistent. From an internal point of view, pillar pages – the “what is X” style – get the most visibility, then blogs. Product, solutions, and marketing pages fall towards the bottom, and very often aren’t used at all.
Content has to be genuinely informative, rooted in research, and backed by data and statistics. It can’t be one person’s opinions loosely assembled into paragraphs. The more authoritative it is, the more likely an LLM will surface your brand when a buyer is doing their research.
Informative content is now a brand exercise in a way it wasn’t before. You’re not just informing readers, you’re establishing your organization as a recognized voice on a topic for models buyers are now talking to first. That means going deep, not just producing one flagship piece. Build content clusters around a topic so the LLM understands you have answers to many related questions, not just one.
How does gated content factor into all of this?
AI has effectively destroyed the traditional attribution model, and gated content is one of the clearest casualties. If your content is behind a form, LLMs can’t crawl it, can’t read it, and can’t surface it when buyers are doing research. And buyers are increasingly making their decisions in LLMs before they ever reach your website.
The old logic was: gate the asset, capture the lead, follow up. But if your ideal customer is asking ChatGPT or Claude a detailed question about your category, and the only content you’ve published that answers it is behind a form, you don’t exist in that conversation. Someone else does.
My recommendation is to ungate everything and shift how you measure success. Form fills and lead counts are going down across the board. That’s a structural change in how buyers research – however good your content is. It means little if an LLM can’t crawl it, and buyers don’t see it. The metric worth tracking now is whether your content is showing up in the places where decisions are actually being made. I.e., the LLM answers.
If form fills are declining, what story do you tell to a board that still wants ‘proof’?
This is the critical conversation happening right now between marketing leaders and boards, so you’re not alone. The honest answer is it requires closer alignment between marketing and sales. Revenue and pipeline have to become marketing KPIs, not just sales metrics.
Beyond that, two signals are worth watching closely. Direct traffic is rising for organizations doing GEO well. People see a brand mentioned in an LLM response and open a new tab to go directly to the site. AI referral traffic, the clicks coming from citations in LLM responses, is on a consistent upward trend for the same reason. Organic traffic may be declining, but a meaningful portion of it has shifted into these two buckets.
It’s not a perfect one-for-one. But it’s a coherent story: here’s what we lost in traditional organic, here’s where it moved, and here’s why revenue is holding or growing. That’s a conversation worth having. The alternative doesn’t work. Defending lead volume while ignoring what’s driving the business isn’t a strategy.
Cybersecurity content is notoriously technical and jargon-heavy. What is the demand of content teams other functions don’t face?
Some marketing functions can operate in a silo and do fine. Content doesn’t work that way because the accuracy bar is just higher. If you get something wrong, the audience notices immediately, and they’re not forgiving about it.
That means GEO, in a cybersecurity context, is fundamentally an organizational challenge. It’s not an SEO person sitting in a corner doing technical optimizations. It requires real collaboration: research teams, product teams, sales engineers, SDRs having conversations with customers about their problems. That’s where the intelligence comes from to make content authoritative enough to rank and be cited.
Silos organizations have built over the years are actively working against us in this respect. Marketing in one lane, research in another, sales in another. The companies getting this right are the ones that have made it easy for those teams to communicate. Quarterly briefings don’t cut it, you need real, ongoing dialogue.
How do you think about the balance between writing for humans vs writing for machines?
Serve humans first. Without humans, there are no LLMs. ChatGPT wouldn’t exist if no one is prompting it. So the content itself has to be useful to a person with a question.
You can structure human-centric content so machines can also consume it. That’s where your relationship with the web team matters. Schema markup, proper meta tags, and content structured around how people search, ask questions, and read. This is all ‘traditional’ SEO practice. Nothing has changed in that respect.
Recency also still matters, as it always has in search. Content kept current signals to models your organization is actively engaged in the space, not just producing content and abandoning it.
What’s the uncomfortable truth GEO practitioners aren’t ready to accept?
Search isn’t dead. That framing is lazy and it does marketers a disservice. Google isn’t going anywhere. But, the search experience as most of us have known it since the early days of Yahoo and Google is changing. The simple keyword box is giving way to conversational, generative AI platforms. That’s happening now, and it’s going to accelerate.
The practitioners who grew up with keyword-based search must adapt. The adaptation means shifting from tracking keywords to tracking prompts. Right now, we don’t have the real-world data to do that cleanly. There are tools that will manufacture prompts and test how you appear in LLM responses, but none of that reflects how real buyers are actually behaving. We’re in the gap.
The hard pill to swallow is you have to start preparing for prompt-based thinking before you have the data infrastructure to support it fully. A younger generation is already using LLMs the way their predecessors used Google. That behavior is going to become more common. Waiting until the tools catch up means being behind when they do.
Aaron Wilson is an SEO specialist at Tenable, where he leads the organization’s search and generative engine optimization strategy. He can be found on LinkedIn.
ABOUT BORA
We’re Bora. We work with security companies to turn complex technical capabilities into clear, credible market narratives.
Get in touch for a free 30 minute consultation. If we’re not the right fit, we’ll help find someone who is.
