Said Versus Did

Said Versus Did

Gartner gives a reader an unusually clean way to grade it. For five years management has repeated one concrete medium-term promise — that Research contract value grows 12% to 16% a year, and that double-digit revenue and modest margin expansion follow from it [1]. The number is specific, it is dated, and the transcript archive records both the claim and the outcome every quarter. That is the material of a credibility ledger: not whether the strategy is sound, but whether the word held.

The record is mixed in a way the headline revenue line hides. The operating promise — the growth algorithm behind the whole equity story — was missed by a widening margin every year and reaffirmed anyway, including at the November-2024 price peak. A risk framed as small and stable one quarter more than halved over the next four. A business once sold as a growth engine was divested as "not core." Against that sits one commitment delivered in full and then accelerated as the stock fell: the buyback. This act walks each in turn, then closes on how management is measured and paid around the exact segment that broke.

The Reaffirmed Model

The 12-to-16 model is not a slogan; it is the arithmetic the company guides to. At the February 2025 call the CFO put it plainly: "our financial model and expectations are unchanged. With 12% to 16% Research CV growth, we will deliver double-digit revenue growth" [2]. The Cash Machine established why contract value is the tell that matters — it is the forward-revenue balance the entire engine refills from. What the five-year series shows is that the tell moved one direction, monotonically, while the promise stayed fixed.

Realized total contract-value growth ran 16% / 12% / 8% / 8% / 1% (FY2021–FY2025); the reaffirmed medium-term objective was a 12–16% band throughout. Sources: Q4 FY2021 [3], Q4 FY2022 [4], Q4 FY2023 [5], Q4 FY2024 [6] and Q4 FY2025 [7] transcripts; revenue growth per reported financials.

Total contract value grew 16% into year-end 2021 [8], 12% in 2022 [9], 8% in 2023 [10], 8% again in 2024 [11], and 1% in 2025 [12]. The last print landed a full eleven points below the low end of the band management had reaffirmed twelve months earlier. Revenue growth tracked the same slope, from 15.5% in FY2021 to 3.7% in FY2025 — the deceleration was in the leading indicator and the reported line together, not a timing artifact between them.

The timing of the reaffirmation is what gives it weight. The February 2025 restatement of the unchanged model came with the shares near their November-2024 record of $551.80, and the CFO framed the year ahead as one in which contract-value growth "would continue to accelerate" [13]. It decelerated instead, to the 1% print, and the stock is down 74.6% from that high. By the February 2026 call an analyst put the pattern to management directly — that there had been "a number of challenging years where there's always something that disrupts your ability to hit that medium-term guidance" — and the 12-to-16 objective was maintained again, with the payoff placed "a couple of years" out. The evidence for the model is a decade of scale and cash; the evidence against it is five consecutive years of missing it by a widening gap. What would settle it is a single year of contract-value reacceleration back into the band — the test the closing act carries to the price.

The Federal Reassurance

The Engine Stalled sized the U.S.-federal cancellation wave and showed how much of the reported 1% it explains. The credibility question is separate and narrower: what management said about that exposure before it broke. Here the said-and-did sit one page apart in the same call.

Said, February 2025: "we ended 2024 with around $270 million of CV which is 5% of the total," spread "widely across agencies and departments," with almost all contracts running one year [14]. Asked on the same call whether the government book warranted a more conservative view, the CEO answered that the federal, state and local sectors "are very diversified, and that remains unchanged," with "no differences at this point" from the prior quarter [15]. The characterization was small, diversified, stable.

The federal book called "diversified" and "unchanged" in February 2025 more than halved over the next four quarters as one-year contracts came up for renewal. Sources: Q4 FY2024 [16], Q3 FY2025 [17], Q4 FY2025 [18] and Q1 FY2026 [19] transcripts.

Did: the $270 million became roughly $165 million by September 2025 [20], $126 million by December [21], and about $114 million by March 2026 [22] — more than halved in the fifteen months after it was called diversified and unchanged. The framing was not dishonest: the contracts were genuinely spread across agencies, and the "unchanged" observation was true of the run-rate on that February day. But it did not survive the renewal cycle, and it was offered at the moment a shareholder most needed the risk sized up rather than down. The useful read for later chapters is the discount to apply to contemporaneous management framing of a tail risk: outside the federal book, contract value was still growing 3.5% in the March-2026 quarter [23], so the reassurance about the isolable shock understated it precisely where it turned out to matter.

Digital Markets, Reclassified

The clearest instance of a said-then-unsaid promise is a whole business line. Gartner built and long promoted Digital Markets — the Capterra, GetApp and Software Advice software-review properties — as a growth engine adjacent to the research core. In 2025 it divested it. The CEO's explanation on the February 2026 call was that the business "didn't fit into that vision, so we made that decision after careful analysis" [24]; the CFO added that "we recognized that digital markets was not core to our business," to be sold "to a more natural owner" with proceeds redirected to the core [25].

A divestiture is not a broken promise on its own — pruning a non-core asset is ordinary capital discipline, and selling to a natural owner is the right move if the fit was wrong. What makes it a ledger entry is the reframing. A line once carried as evidence of the growth story was retired as "not core" in the same window the growth algorithm stalled, and management now describes the surrounding effort — a "BTI transformation" of the research product plus staff reductions — as the most significant change of the CEO's two-decade tenure. Whether that reset is a genuine sharpening of the core or a relabeling of a demand problem the company could not otherwise name is not yet decidable from the filings; the honest note is that a previously touted thread was quietly dropped rather than delivered.

The Kept Promise

Credibility runs both ways, and on capital return the record is unambiguous. Diluted share count fell every year, from 86.18 million in FY2021 to 75.61 million in FY2025 — about 12% of the company retired in five years — funded by cash repurchases of $1.66B, $1.04B, $606M, $735M and, in FY2025, a record roughly $2.0 billion [26]. Cash spent on acquisitions across those five years was zero. Where the operating model was serially missed, the buyback commitment was met and then leaned into: as the stock fell through the drawdown, the pace increased rather than paused.

Share count fell from 86.18M (FY2021) to 75.61M (FY2025), ~12% cumulative; FY2025 repurchases of ~$2.0B were the largest of the five years. Sources: FY2025 10-K repurchase disclosure [27]; annual amounts per reported financials.

The buying continued into 2026. In the first quarter management "reduced our share count by about 4%, buying back $535 million of stock" [28] — a quarter's pace that, annualized, would retire shares faster than any full year in the table. That the record repurchase was struck into a 75%-off drawdown, part debt-funded, is the substance of The Buyback Bet; the point for the credibility ledger is narrower and firmer. On the promise it could control directly — returning cash and shrinking the count — management did exactly what it said, without interruption, for five years.

Pay and the Carve-Out

How management is measured completes the ledger, because the incentive design routes around the segment that broke. Both 2025 plans exclude the U.S.-federal public-sector business by construction. The annual bonus — half EBITDA, half revenue — paid at 119.6% of target, on certified EBITDA of $1,537 million against a $1,463 million target [29]. The long-term performance shares, earned on a one-year contract-value goal, vested at 82.1% of target [30]. Both results, the proxy states, were struck "excluding our U.S. federal public sector Insights business" on an FX-neutral basis; the Compensation Committee set the goals that way because of the "unusual amount of volatility and uncertainty" in that book [31].

Bonus and PSU payouts, measured on segment-adjusted metrics that exclude the federal book, ran above and near target through a 74.6% peak-to-trough decline; shareholders ratified the pay program with 93% support. Sources: 2026 proxy CD&A [32], PSU results [33] and say-on-pay result [34]; drawdown per reported market data.

The mechanism is worth stating in plain terms. The federal cancellations are the piece of the decline management points to most often; they are also the piece the bonus and equity metrics are defined to leave out. So the payout formula measures performance around the exact segment whose collapse the company cites for the stock's fall — executives earned above-target cash and near-target equity in the same year the shares dropped by three-quarters [35]. The design is defensible in isolation: carving out a genuinely volatile public-sector book from a manager's scorecard is common practice, and the "long-term" plan resets its goal annually rather than over a multi-year window. But the two choices compound — a one-year performance window on the equity, and the exclusion of the segment that broke — so realized pay tracked the healthy core while the price tracked the whole.

Two structural facts sit alongside the pay outcome. Shareholders ratified the program with 93% say-on-pay support [36], so this is not a governance revolt in the making. And the person at the center of it is the same one throughout: Eugene Hall, CEO since August 2004 and Chairman since July 2024 [37], the only non-independent director on a thirteen-member board and the only executive whose employment agreement "obligates the Company to include him on the slate of nominees to be elected to our Board" for the agreement's term [38]. The oversight scaffolding around him is conventional — a lead independent director, all-independent committees, no related-party items — but the authority to set and reaffirm the medium-term model, and the board seat from which to defend it, are concentrated in one twenty-two-year incumbent.

The State of the Record

The ledger reads clearly on both sides. Broken, or at least serially unmet: the 12-to-16 contract-value model, missed five years running and reaffirmed at the peak; the "diversified and unchanged" federal reassurance, offered one quarter before that book more than halved; and Digital Markets, promoted as growth and retired as "not core." Kept: the buyback, delivered every year and accelerated into the drawdown. Measured around the break: an incentive design that excluded the failing segment and paid above target while the equity fell 74.6%.

Two live commitments a reader can hold management to, both dated and checkable. First, the reaffirmed medium-term model — 12% to 16% contract-value growth, with the payoff again placed "a couple of years" out on the February 2026 call; the falsifiable test is a single year back inside the band, and the next print is the FY2026 disclosure. Second, the buyback discipline the company has actually earned trust on — whether the record, partly debt-funded repurchase proves accretive or merely concentrates the bet is the question the capital-allocation act takes up next, where the cash the machine throws off meets the price the market is paying for it.