Chapter 1
The Subscription Engine
Gartner sells research by subscription: enterprises pay an annual fee for advice on their most important decisions, and the same clients have renewed and spent more, year after year, for most of the last decade. That flywheel is measured by contract value and retention. In 2025 it stopped turning at the top end — technology-sales contract value went flat and existing clients, for the first time in years, renewed for less than the year before. The stock has fallen roughly 71% from its late-2024 peak. This chapter establishes what Gartner is and frames the question the rest of the report tests.
FY2025 Revenue
Insights Contract Value
Insights Gross Contribution
Operating Margin
Sources: FY2025 Annual Report (Form 10-K), Results of Operations [1] and Reportable Segments [2].
What Gartner does
Gartner (NYSE: IT) is a research and advisory business serving over 13,000 enterprises across roughly 90 countries [3]. It reports in three segments. Insights — the segment renamed from "Research" in the second quarter of 2025 — is the core: subscription access to published research, data and benchmarks plus direct access to more than 2,400 experts [4]. Conferences runs destination events; Consulting delivers project-based advisory work. A fourth reporting line, "Other," holds Gartner Digital Markets, which the company sold in February 2026 for about $110 million [5].
Insights is about four-fifths of revenue and nearly all of the profit. Within it, two sales forces carry the business: Global Technology Sales (GTS) sells to buyers and providers of technology, and Global Business Sales (GBS) sells to leaders in finance, HR, supply chain, legal and other functions [6].
Source: FY2025 Annual Report (Form 10-K), revenues by segment (post-recast; Digital Markets in "Other") [7].
The economics are unusually attractive. Insights carries a 77% gross contribution margin, effectively unchanged year over year [8]. A subscriber pays up front, Gartner recognizes the revenue ratably over the contract, and the research produced for one client is reused across thousands — the classic syndicated-publishing model with software-like incremental margins.
How the flywheel is measured
The number that matters most for this business is contract value (CV): the annualized dollar value of every subscription in force at a point in time. Management calls it the signal of "the long-term health of our Insights subscription business, since it measures revenue that is highly likely to recur over a multi-year period" [9]. Two retention rates sit alongside it. Client retention is the share of last year's clients still on the books. Wallet retention is the share of last year's dollars retained — so a figure above 100% means the surviving base spent more this year, and below 100% means it spent less [10].
For most of the last decade the flywheel compounded. Combined contract value roughly doubled from about $2.5 billion in 2016 to $5.2 billion at the end of 2025, with GTS the larger of the two engines.
Source: FY2025 Annual Report (Form 10-K), Reportable Segments — GTS and GBS contract value (FX-neutral) [11]; prior years as reported in successive 10-Ks.
Where the engine stalled
The stack above hides the break, because GBS kept growing while GTS did not. On a currency-neutral basis, GTS contract value was $3,910 million at the end of 2025 against $3,911 million a year earlier — flat — after years of high-single to double-digit growth. GBS added 3%. More telling than the level is the flow: GTS wallet retention fell to 96%, down six points from 102%, and GBS wallet retention fell to 99%, down seven points from 106% [12]. Both crossed below 100% — the installed base renewed for fewer dollars than it held a year earlier, a first in this dataset for GTS since the pandemic year of 2020.
Source: FY2025 Annual Report (Form 10-K), Reportable Segments [13]; prior years from successive 10-Ks. The 100% line separates net expansion from net contraction within the retained base.
Gartner names the largest single cause. Its Insights contract value with the US federal government was about $126 million at the end of 2025, and "less than half" of the December 2024 federal contract value was retained during the year — with further termination-for-convenience notices covering roughly $3 million of contracts set to lapse in early 2026 [14]. A federal book that halved in a year is a concrete, cyclical shock. What the filing does not settle is whether the weakness stops there, or whether cautious enterprise technology budgets and the arrival of generative-AI research tools are also thinning renewals in the commercial base. That distinction is the spine of this report.
The reported earnings are noisier than the business
Headline results overstate the deterioration. Net income fell from $1,253.7 million in 2024 to $729.2 million in 2025, and diluted EPS from $16.00 to $9.65 — a 42% drop. But 2024 carried a $300.0 million pre-tax gain on event-cancellation insurance claims that did not recur, and 2025 absorbed a $150.0 million goodwill impairment on the Digital Markets unit it was preparing to sell [15]. Strip both, and the earnings step-down is far shallower than the chart of net income suggests.
Source: reported financials, FY2021–FY2025 10-Ks; FY2025 figures per Results of Operations [16].
The cleaner read sits in operating income, which peaked at $1,237 million in 2023 and slipped to $1,026 million in 2025 even as revenue rose. Some of that is the $150 million impairment; the rest is real margin compression — SG&A grew 6%, including a $56.6 million workforce-reduction charge, while the quota-bearing sales force shrank about 2% [17]. Operating margin fell from about 21% in 2023 to 16% in 2025. The franchise is still highly profitable; it is no longer expanding its margins.
Capital allocation and the stock
Gartner returns essentially all its free cash flow through buybacks — no dividend. Boards have authorized roughly $7.5 billion of repurchases since 2015, plus a further $500 million in January 2026 [18]. In 2025 alone the company spent $2.0 billion on its own shares — more than the $1.3 billion it generated from operations, funded in part by $800 million of new senior notes issued in November 2025 [19]. Share count has fallen from about 92 million in 2018 to 75.6 million.
Source: derived from reported financials, FY2018–FY2025 10-Ks (shares outstanding).
That buyback met a collapsing tape. Gartner compounded from a $160 year-end close in 2020 to a $484 close at the end of 2024, then round-tripped: $252 at the end of 2025 and about $142 in July 2026, roughly 71% below the peak.
Source: daily price history (NYSE: IT), year-end closes; 2026 as of July 7, 2026.
At about $142, the market values Gartner's roughly 75.6 million shares near $10.7 billion — close to 15 times trailing (impairment-depressed) earnings, against the roughly 30 times it commanded at the 2024 peak. The de-rating and the retention break arrived together.
The evidence points to a genuine slowdown in the core Insights franchise, concentrated first in GTS, with a documented federal-government shock as the largest identified cause. The strongest fact on the other side: GBS contract value still grew 3% and client retention held near 85%, so this reads more as decelerating expansion than client flight. What would settle it is whether wallet retention stabilizes above 100% once the federal comparison laps in 2026, or keeps sliding in the commercial base.
The question this report answers
For a decade Gartner was a subscription compounder: rising contract value, wallet retention comfortably above 100%, expanding margins, and a shrinking share count that levered all of it into per-share growth. In 2025 the leading indicator broke. The central question of this report is whether the 2025 stall in Gartner's core research franchise — flat technology-sales contract value and wallet retention falling below 100% for the first time in years — is a cyclical air-pocket driven by a US federal pullback and cautious technology budgets, or the leading edge of structural erosion as generative AI commoditizes syndicated advice. Everything that follows is built to test that.