Gartner's latest annual report describes its main sales force in one plain sentence: Global Technology Sales sells to "users and providers of technology". One organization, selling the same subscription research to the companies that appear in the rankings and to the companies that buy software because of them.
That is the dual-customer problem, stated by the firm itself in an SEC filing. Nothing hidden, nothing alleged. The interesting question is what the arrangement does to the product, and answering it honestly requires holding two ideas at once: the courts have cleared Gartner of the worst accusation, and the structure still deserves more scrutiny than most buyers give it.
Who actually pays for analyst research?
Gartner reported total revenue of $6.5 billion for fiscal 2025, up 4 percent on the prior year. The engine is the segment it renamed in mid-2025 from Research to Business and Technology Insights: $5.1 billion of that total, growing 5 percent, at a 77 percent gross contribution margin. Conferences contributed $644.7 million and consulting $552.5 million. Subscription research is not a side business. It is nearly four of every five revenue dollars, and it is spectacularly profitable.
Both sides of every deal that research covers are paying customers. The filing assigns technology users and technology providers to the same sales organization, and it does not disclose how revenue splits between them. Commercially, a vendor being placed on a market ranking and a CIO consulting that ranking before a shortlist are the same kind of client: a contract-value number in the same $5.2 billion pool.
Sit with how unusual that is. An equity analyst holding positions in the companies they rate files disclosures. A judge with a consulting arrangement with one litigant recuses. In technology research, the evaluated and the evaluating funding the same evaluator is not a scandal. It is the published business model.
What did the courts actually find?
Two vendors tested the arrangement in litigation, and the record deserves telling in full because it is so often half-told.
ZL Technologies, an email-archiving vendor unhappy with its niche placement, sued in 2009. A federal court in California dismissed all five of its claims, holding that a Magic Quadrant position is an expression of opinion, not a provable statement of fact.
NetScout went further, alleging in 2014 that its challenger placement was retaliation for not buying enough Gartner services. The theory failed completely. Gartner won at the trial court in 2017, and the Connecticut Supreme Court affirmed in January 2020, finding both that the rankings were protected opinion and that there was no evidence quadrant placement correlated with how much consulting a vendor bought.
Let me state that as plainly as fairness demands: pay-to-play was alleged, litigated for years, and rejected on the evidence. On the records before those courts, nobody bought a ranking, and repeating the accusation without the outcome misleads the reader.
But look closely at what won. Gartner's successful defense was never that its ratings are accurate measurements. It was that they are subjective opinions, not susceptible of being proved true or false, and therefore protected speech. The courts agreed, twice. Which sets up the one question worth carrying out of this section: if the winning legal argument is that a ranking is opinion rather than fact, what exactly is a procurement shortlist built on that ranking standing on?
Why the structure still matters even if nobody cheated
Because incentives do their work upstream of misconduct. Neil Pollock and Robin Williams, two University of Edinburgh scholars, spent years studying the analyst industry for How Industry Analysts Shape the Digital Future (Oxford University Press, 2016). Their central finding is that firms like Gartner do not simply describe technology markets. They help constitute them. Category definitions, inclusion criteria, and rankings produce the shortlists vendors compete for and buyers choose from. In their account, analyst firms are part of the machinery that makes IT markets exist at all.
Take that seriously and the dual-customer question changes shape. What follows is my reading of the structure, an opinion resting on the facts above. When one firm defines a market category, decides which vendors qualify for it, ranks them within it, and sells advice to those same vendors about competing in it, the feedback loop never leaves the building. No analyst needs to shade a score for the system to tilt. The tilt lives in quieter choices: which markets get a quadrant at all, which capabilities harden into inclusion criteria, whose briefing shapes the category's vocabulary.
A hypothetical shows the mechanics without accusing anyone real. Picture a category report on exposure-management platforms featuring two fictional vendors: Meridian, an incumbent with a deep advisory relationship with the research firm, and Sentrix, a newer entrant with none. Nobody bribes anybody. Meridian briefs the analysts quarterly, though, so its roadmap language seeps into how the category is described. When inclusion criteria get drafted, they happen to mirror an architecture Meridian shipped and Sentrix deliberately avoided. Sentrix lands low for reasons that are defensible line by line and structural in aggregate. Every step honest, and the output still carries the fingerprints of whoever was in the room most often.
Vendors are not confused about this. Analyst relations is a standing line item in enterprise software marketing budgets precisely because access is believed to matter. Rational actors do not fund a function that buys nothing.
What a single-customer model looks like
OmniAxis is built as that single-customer model, and the fix is architectural rather than attitudinal, since attitude is the one asset the dual-customer model already holds in abundance.
The design rule: the party being graded has no commercial path to the grade. In OmniAxis, a research layer grades every vendor capability claim on an ordered scale of evidentiary strength, and it grades on independent evidence, never on a briefing. OmniAxis refreshes that evidence on a managed schedule, weekly on the top plan, so a grade tracks the current documented record instead of an annual research cycle, and every profile carries its last-refresh date. The platform then puts each vendor's marketing-claim score directly beside its evidence score, because the distance between what a vendor says and what independent evidence supports is intelligence in itself. A claim with nothing independent behind it stays at the bottom of that scale, and no meeting, sponsorship, or subscription changes that. Only new independent evidence does.
None of this makes ranking opinion illegitimate. Experienced analyst judgment has value, and the courts were right to protect it. But protection for the speaker is not assurance for the reader. Before your next shortlist leans on a ranking, read the filing and the book, then apply the oldest test in the trade to the research you pay for: follow the money, and notice that it arrives from both directions.