How our research is made — and how to cite it.
We don't write opinions; we publish data. This page is the standard every report in the Visigent library is held to: where the numbers come from, who is accountable for them, how they're dated, what we disclose, and what happens when we're wrong.
What we research from
Visigent's research corpus was built over eighteen months from academic papers, platform documentation, industry studies and our own citation analysis — more than 40,000 websites and 1,500 sources across eleven categories of GEO research, plus 12,500 cross-platform query responses gathered from five LLMs. The corpus informs both our published research and the GEO Score methodology.
One claim, one primary source
Every published statistic ties to a primary source, an experimental dataset, or a reproducible methodology. When the finding is ours, the report states the sample, the prompt set, the models queried, the dates of the run, and the limits of the method. When the finding is someone else's — Ahrefs, SE Ranking, Profound, Muck Rack, Pew, McKinsey, Aggarwal et al. — we name them in the sentence, not in a footnote, and we don't present their result as our own. Where two credible sources disagree, we publish both and say so.
Who is accountable
Every report and guide carries a named author and a named reviewer. Tom Howell, Co-Founder & CEO, is author of record for the research library; members of the Visigent team review for method, product accuracy and legal exposure, and are named on the page. "Visigent Research" is the publisher, never the author.
Dates you can trust
Each piece shows its publication date and its last-updated date. Statistics carry the date of the underlying source or run. We refresh cornerstone pieces on a schedule — because AI systems weight recency, and because the systems we study change monthly — and we never show a publication date later than the day the piece went live.
What we disclose
Visigent sells a GEO platform and a managed program. Where a report compares vendors, including us, it says so at the top, uses the same criteria for every vendor, and labels our own position as our opinion. Where a report cites a Visigent customer result, the case study it links to holds the metrics, the window and the tool that measured them. Practical implications for buyers sit in a labelled module at the end of each report, separate from the findings.
When we're wrong
Send corrections to help@visigent.com with "Research correction" in the subject. We review within five business days. Material errors are corrected in the piece with a dated note explaining what changed; minor errors are corrected silently and logged. We do not remove pieces to hide errors.
Citing our work
All published research is openly shareable and fully sourced. Teams, analysts, journalists and marketers may reference our findings with attribution to Visigent and a link to the report. Please cite the figure's date; if you need the underlying data for a Visigent-original finding, ask — where we can share it, we will.
Start with the score.
Enter your website. In 24 hours: your GEO Score, how each model describes you, where you're missing, and the first three fixes.