Five readings sit beside each name on the earnings calendar, under the heading Forward demand. They all answer versions of one question — what does this company's own paperwork say about demand that has not shown up in revenue yet? — and they are deliberately not combined.
Each column carries its own evidence label, because each rests on a different amount of evidence. One has been confirmed in a single out-of-sample window. Two are observations from a small hand-built study and are being tested right now. One is labelled suggestive and always has been. The last is simply a quote from a filing.
There is no composite, no ranking and no ordering by any of this. A score would assert that five findings of five different strengths add up to something, which is exactly what our own research declines to say. The absence of that number is a deliberate product decision, not an omission.
Coverage — delivered growth minus the growth the price requires. We reverse-engineer the revenue growth rate that would justify the company's current enterprise value under a deliberately small, fully-specified model, and subtract it from the year-over-year revenue growth the company actually just reported. A positive number means the last quarter delivered more than the price is asking for; a negative one means it delivered less. Almost every company shows a negative gap, so the useful reading is the comparison between names and against a name's own history, never the sign on its own.
RPO acceleration — is the contracted backlog growing faster than it was? Remaining performance obligation is contracted revenue a company has not yet recognised: the best forward-demand figure a subscription business publishes for free. This column is the change between the latest annual growth rate of that backlog and the previous one, in percentage points, with a marker when the backlog is both growing and growing faster than before.
Deferred-revenue gap — is the prepaid book outgrowing the top line? The shipped bookings-ahead reading, re-served here unchanged: growth in current deferred revenue minus growth in quarterly revenue, on the latest filed quarter against the same quarter a year earlier. Its own page explains it in full.
Segment acceleration — is one part of the business outrunning the whole? From the latest filing that breaks revenue out by segment: the fastest-growing reported segment's year-over-year growth, minus the company's consolidated growth over the same period, in percentage points. "Fastest" means fastest-growing, not largest — a small segment posting a big rate wins the column, which is the intended reading and not a defect.
Net revenue retention — the filer's own words. Where a company states a retention rate in its annual report, we show the rate together with the sentence it came from. Where it does not, we show nothing. Filers differ on the base, the window and the customer population, so a rate without its sentence is not comparable to anything — which is why this column is quote-or-nothing and is never ranked.
Coverage comes from an event study of roughly seventeen thousand earnings announcements between 2023 and mid-2026. In the validation window it separated the highest and lowest coverage groups by about three and a half percentage points of subsequent three-month drift, with a confidence interval that excluded zero after correcting for the four tests we had declared in advance. In the two-year discovery window that ran first, the same statistic was indistinguishable from zero. The evidence is one window wide. A result that is null across twenty-four months and significant across sixteen is the shape both a genuine regime-dependent effect and a chance finding produce, so this ships as context and never as a signal. Its label says so.
RPO acceleration and segment acceleration come from a hand-built study of thirty-six companies that produced very large multi-year returns, against seventy-two that did not. Backlog acceleration appeared about five times as often in the winners; a segment outrunning the consolidated business about one and a half times as often. That is a hypothesis, not a finding: a hundred-odd companies chosen with hindsight cannot establish a signal. A confirmation study over the full event panel is running as this ships, pre-registered before any of its data was assembled, and it will either confirm these two, kill them, or leave them as permanently labelled cohort observations. Whichever way it lands, these labels change with it.
The deferred-revenue gap is labelled suggestive and stays that way. It ranked first among sixty-five candidate measures in our own point-in-time study — and failed every multiple-testing correction for having looked at sixty-five things. The bar it must clear to be relabelled was written down before any live data existed and cannot be reached before the 2030s.
Net revenue retention is not evidence of anything on its own. It is a disclosure.
Every column is computed from the latest stored filings, and each carries the period it was built from. This matters: the research behind these columns measured them at a filing date, using only what was public then. The calendar is a reading aid on today's screen, not an event study, and a fresh price sitting beside an older filing is the normal case rather than an error.
The coverage model is not our fair-value engine. The engine's inputs cannot be reconstructed as they stood at a 2023 announcement, so the study committed to a much smaller model whose inputs can be: one constant revenue growth rate over five explicit years, a flat sector-median operating margin, the US statutory tax rate, a constant reinvestment charge, a terminal growth rate capped at the risk-free yield, and a single market discount rate with no per-company beta. It ignores leverage and company-specific risk by construction. No number in this block is a house fair value, and none of it feeds one.
That model demands more growth than the median company delivers, for reasons the research names: it charges no extra reinvestment for faster growth, and the sector margin is a median rather than the company's own. Read the column as a relative measure.
The research directories behind these columns — the requirement-coverage event study, the winner-anatomy study, the bookings-ahead pre-registration, and the pre-registration of the confirmation study now running — are cited on the monitor itself. Exact parameter values for the coverage model are documented internally and are not enumerated here; that withholding is disclosed rather than hidden, as it is on every other page.
