97th Floor CEO, Paxton Gray, and WordPress VIP CMO, Jodi Cerretani, teamed up on a three-part webinar series about AI discoverability. The second session covers how to turn your baseline into a strategy AI can understand and recommend.
Knowing how AI sees your brand is step one. Changing what it sees is step two. That's where most brands stall.
The goal is to show up in front of the right people, saying the right things, in a way that grows your bottom line. Paxton called that "the real trick these days." He breaks the work into four parts: brand, architecture, content, and third-party validation.
Enterprise AI Discoverability Series
Investing in AI Discoverability: Measuring Impact. Learn how to measure progress and make smarter investment decisions.
Your Technical Foundation Works Like a Library
Jodi explains the technical basics with a library. Information architecture is the floor plan. It shows where everything lives. Structured content is the label on each book. It says what each page is. Governance is the librarian who keeps it all accurate over time.
Most content systems were built before AI search. Her three fixes:
- Name things the same way everywhere. Mixed names for the same product mean "AI can see it as two different things."
- Build content once and reuse it. Product names, pricing tiers, and executive names should live in fields you update once. "Use fields, not free text blobs."
- Bake machine-readable tags into templates. Don't rely on someone to remember.
If your website software can't do this automatically, "you might be in trouble."
Being Readable Isn't Being Recommended
As Jodi put it, "being readable isn't the same thing as being recommended." That's the job of content.
SEO meant getting one pillar page to rank. Paxton explained that "AI is much more holistic and it should be focused on topics and establishing authority around certain topics."
The biggest mistake he sees is treating prompts like keywords. Prompts are longer and more detailed, and AI runs several searches off each one. He calls this the query fan-out. "The key is answering these secondary questions through multiple pieces of content on our site, not simply on the single page."
That starts with audience research. Scour Reddit for the words and problems your audience shares. Map your personas, their pain points, and the questions they ask. Then build a pillar page with focused supporting pages around it.
Your Old Content Is Teaching AI the Wrong Things
Enterprise brands carry decades of content. Much of it mentions products they no longer sell or positioning they've moved past.
Paxton's team runs clients through a content analysis matrix: Is the page pulling traffic? How does it perform? Does it align with the business? A page can perform well and still point the wrong way. "That page should not exist and it is steering AI in the wrong direction."
He shared the example of a cruise line client that wanted to own Alaska. His team cut irrelevant content, then built focused pages on ports, wildlife, excursions, and timing, all mapped to the customer journey. Now, "They're the most cited cruise line in the world for anything related to Alaskan cruises."
Off-Site Validation Shapes What AI Believes
Paxton was clear that "Your site is one of the biggest factors that AI uses to determine who you are and when they should bring you up." But AI learns from the rest of the web too.
A brand sentiment analysis shows what AI thinks about you and where it learned it. Key sources vary by industry. For one client, Jazz HR was the third most cited site in its space. Enterprise brands also have partner marketplaces, which AI search currently treats as third-party validation.
Reflecting on his point about pruning, Jodi added, "more is not more when it comes to content. It's about clarity."
Quick Wins Come Before the Rebuild
Jodi splits the work into two buckets.
Quick wins: fix wrong tags on key pages, unblock AI crawlers locked out by accident, patch sitemap holes, and clean up conflicting messaging.
Foundational work: rebuild your tagging system, make sure your site technology automates what it should, build authority on your core topic, and write real governance rules.
AI Discoverability Needs One Captain
Marketing, content, SEO, web, engineering, security, and governance all play a part. What's usually missing, Jodi said, is one directly responsible individual to quarterback the work. A live poll showed just 5% of attendees had a clear owner in place.
The pressure is coming from the top. Boards see organic traffic falling, and their message to CMOs is simple: "You got to win your traffic back."
Paxton sees measurement as the industry's biggest bottleneck. "It's one thing to know that you need to do something and it's another thing to know whether or not what you're doing is having an impact."
Infrastructure and Content Only Work Together
Jodi closed with the core idea. Strong infrastructure without content makes you readable but not recommended. Strong content without infrastructure is "a great story that no machine can read."
You need both to get cited and recommended. Watch the full recording now.
