Who Gets Picked When You Ask AI for a Recommendation?
Ask ChatGPT to recommend a project-management tool, a cybersecurity firm or a payroll provider, and you’ll get a tidy, confident list in seconds. It feels like advice from a well-read friend. A new book argues it’s often closer to a sales pitch, and that plenty of good companies never make the list at all.
In The Invisible Audience: Why AI Leaves You Off the List, and What Actually Matters, Maria Dykstra, AI Visibility Architect, examines how tools like ChatGPT, Perplexity and Google’s AI Overviews decide which businesses to recommend when buyers ask who they should consider. Her conclusion: by the time a customer contacts a company, an AI system may already have quietly decided whether that company is in the running.
Who’s really behind the “best of” lists
For the book, Dykstra traced 1,855 citations across 160 AI answers to buyer questions about which companies to choose. She then took a close look at 40 of the web pages those answers relied on.
What she found matters whether you run a business or just use AI to shop. Thirty of those 40 pages, three out of four, came from companies selling in the very category they were writing about. More than half (22) ranked or included the publisher’s own products.
In other words, many “Top 10 tools for X” articles that read like neutral reviews were written by one of the contenders. The AI summarizes them, and the result lands in your chat window looking impartial.
Found, then misunderstood
You might assume the companies left out are simply too small to notice. Dykstra says that’s usually not the problem. More often, the AI finds a company and misreads it, getting wrong what it does, who it serves or why it’s credible. As she puts it, the machine “finds them and gets them wrong.”
The book breaks the problem into seven layers, a framework Dykstra calls the Algorithmic Authority Stack. In plain terms, it asks:
- Can the AI tell what the company actually is?
- Can it verify what the company knows?
- Does the company use the same words its customers do?
- Can an AI lift a clear answer straight off its website?
- Does it show up where buyers are searching?
- What do outside sources say about it?
- Do its AI visibility reports measure what matters?
She also draws a useful distinction: being mentioned, cited or ranked isn’t the same as being recommended, and only a recommendation puts a company on the shortlist. The book closes with a first-30-days plan for fixing the biggest gaps.
Why it matters beyond marketing
Dykstra wrote the book for founders and marketing teams. She has applied her framework to more than 50 B2B companies in industrial, healthcare, fintech and cybersecurity markets. Before that, she spent 13 years at Microsoft working on advertising and search systems, then 13 more leading her own firm, TreDigital.
But the takeaway applies to anyone who has asked a chatbot for a recommendation. AI answers are only as neutral as the sources behind them. Before acting on that tidy list, it’s worth asking who wrote the pages it came from, and who might be missing.
About the book
The Invisible Audience is available on Amazon in paperback ($14.99) and Kindle ($9.99) editions. Learn more at mariadykstra.com.
Source: announcement via EIN Presswire, October 22, 2026.
