The GEO field is young, and it shows. It is full of confident claims about what "works on AI" and remarkably thin on evidence. We wanted to do something about that. So Synlig.ai collaborates with BI Norwegian Business School on research into AI visibility: How the content of a website actually affects the way AI models describe and recommend a brand.
Why a research collaboration?
Because our customers deserve better answers than "the industry believes that". When we recommend an initiative, we want to be able to say why it works, how large the effect typically is and under which conditions. That standard is hard to uphold alone in a field changing this fast. Academic method, with systematic experiments and verifiable findings, is what separates knowledge from anecdotes.
For BI this is a living research field with access to real data. For us it is a professional backbone in the product. For customers it means the recommendations in the platform are built on tested knowledge, not opinion.
What we investigate
The core question is which properties of content make AI models pick it up, quote it accurately and use it as the basis for recommendations. That spans topics like:
- How structure and phrasing affect whether content gets cited in AI answers
- How changes to a website show up in visibility over time, and how quickly
- What separates brands that get described consistently and correctly from those that get confused or left out
The findings flow directly into how we build the product: Which initiatives we recommend, how we prioritise them, and how we measure effect. Much of what we write in our guides rests on exactly this combination of measurement data and systematic testing.
Why we talk about it
Because openness about method is part of credibility. We ask our customers to trust the numbers and recommendations in the platform, so they should know how the knowledge behind them is produced. And because we believe the business community benefits from real, documented expertise in AI visibility being built here, not just imported claims.
We will share findings on this blog as they are ready. If you want to know how this applies to your business already, the simplest step is to book a demo and we will show you what the measurements say about your brand.