Turn your quick AEO wins into a real AEO system
On-Demand | Recorded September 21, 2026
.png)
You got a citation. Maybe a mention in an AI Overview. Maybe a contact who said they found you through ChatGPT. But did that result change what you tested next, or did you just wait to see if it happens again? Without a learning loop that turns wins into direction, AEO stays a one-off experiment.
In this session, Nick Lafferty from Profound and Vivian Hoang from Webflow join Iron Horse to break down how to build a 90-day AEO experimentation framework. Nick shares Profound's own approach to accuracy tracking (see how National Geographic holds 99.3% accuracy across AI-generated claims about the brand). Vivian walks through how Webflow used citation data to catch and correct outdated positioning across hundreds of pages. And we cover what's actually driving the results our enterprise clients are seeing, and where most of them are still stuck.
If you've already started your AEO work and the results are inconsistent, this session gives you a way to build a system around what's working, instead of another list of tactics to try next.
Speakers
- How to read your early AEO results correctly, and why a single citation doesn't mean your work is done
- How to turn what worked into a repeatable system, not a one-off win you have to chase again next quarter
- Which AEO experiments to keep running, which to cut, and how to decide within 90 days
- Where AI visibility actually breaks down, and why only 5-10% of your citations come from your own website
Key takeaways
- Only 5-10% of citations in a given category come from your own .com property. The other 90% happen off-site, on Reddit, Wikipedia, competitor content, affiliates, wherever your buyers actually go looking.
- A third of the room hasn't started AEO work at all, and another third are just running their first experiments. Only 20% have a system in place they're actively scaling, and even those teams are struggling to prove pipeline impact.
- Webflow didn't roll out FAQ formatting and structured tables company-wide on a hunch. They A/B tested headers-as-questions and table formats first, then built the winners into their CMS templates once the data backed it up.