A written gate gets marked satisfied on management's adjective instead of its own arithmetic

Claim

A gate exists to stop a rating from moving on price action or narrative. It only does that job if it is re-read verbatim at the moment it is claimed to be satisfied.

The instance

On 2026-06-11 the Gartner file gated any upgrade on:

"positive net CV additions 2 consecutive quarters"

On 2026-08-04 Gartner reported, the CEO said "Contract Value growth accelerated again", the stock rose +19.0% in a session, and Watchlist.md was updated to "🟢 GATE SATISFIED — needs /analyze."

The gate was not satisfied. Q1 CY2026's net CV addition was −$56M — negative — and that figure was recorded in the June file itself. A streak of two consecutive positive quarters therefore cannot begin before Q2. Q2 was, at best, quarter one of two. The earliest the gate can clear is the Q3 print, roughly three months after it was declared cleared.

Two supporting distortions made the adjective more persuasive than the arithmetic:

1. The quotable growth number was "ex-Fed" (+3.3%). Total CV growth was ~+1.7%. The qualifier carried ~160bps of shrinking federal business and travelled with the headline into the summary.
2. A rate improvement was read as a level. "Growth improved 70bps" is true and is still the weakest CV growth in the company's public record outside 2020.

Why the failure is systematic, not careless

The gate is written once, months earlier, in a file nobody re-opens. The print arrives as a headline, a beat/miss, and a price move — all of which are same-day, high-salience, and none of which contain the gate's text. The default information flow never puts the test and the result on the same page. Marking it from memory means marking it from the adjective.

Note the asymmetry: this failure only ever runs one direction. Nobody mis-remembers a gate as unsatisfied after a +19% day.

How to apply

1. Re-open the prior file and copy the gate's exact sentence into the new analysis before evaluating anything. If it cannot be quoted, it cannot be marked.
2. Count the required quarters backwards from the print, not forwards from today. A 2-quarter test needs the prior quarter's value, which lives in the old file.
3. Strip every qualifier from the reported metric — "ex-Fed", "constant currency", "same-store", "organic", "adjusted" — and ask whether the gate was written on the qualified or unqualified number. Gartner's was written on total CV.
4. A rate-of-change improvement never satisfies a level test, and vice versa.
5. When a gate is partially met, write "quarter N of M" in the watchlist rather than a binary. It preserves the progress without spending the upgrade.

What this is not

Not an argument for ignoring management commentary — the CEO's remark was directionally true and the CV acceleration is the first encouraging Gartner datapoint in four quarters. The error is letting true commentary substitute for a specific test it does not answer.

Related: [[principle-primary-source-beats-vendor]] · [[pattern-shortlist-runs-when-unexecuted]] · [[pattern-buyback-manufactured-eps-screens-as-growth]]

History

• 2026-08-04 — Filed after the Gartner Q2 CY2026 print. static: this is a property of how gates and earnings coverage interact, not of any market regime.