There is a moment, somewhere in the fourth quarter, when most companies find out whether the year worked. The finding-out is treated as natural, the way weather is natural. It is not natural. In most misses, the deciding information was in the data by spring: the discount overrides accumulating in one region, the pipeline thinning behind a healthy headline number, the cost line drifting a fraction each month.
The standard version of the story runs like this. The plan's headline health indicators stay green until the December close, when the miss finally becomes undeniable. Reconstructed afterwards, the driver, a pattern of discount overrides in one sales region, was statistically visible in March. Nine months of warning existed. The operating rhythm just never asked the question that would have surfaced it.
Nine months is the difference between managing the year and explaining it.
The wrong tense
The standard operating rhythm interrogates the past. Monthly reviews ask what happened and why; quarterly reviews ask it more formally; the annual review asks it with consequences. Every question is in the past tense, which guarantees the answers arrive after the outcomes.
The tense that changes results is the future one. Will we make it, and at what odds? Asked continuously, that question converts a plan from a document into an instrument. The target is not a number to be compared against in December but a probability tracked every week: 74% last month, 68% now, and falling because of a named driver in a named region.
Gartner's research on finance AI explains why this is rarer than it should be: 84% of finance organisations have started with AI, 7% report high impact, and the investment leans heavily toward productivity rather than decision quality. Accelerating the past tense produces the same December, sooner.
What continuous odds actually change
The objection writes itself. We already forecast, and reforecast. But a periodic reforecast is a snapshot with the same weaknesses as the plan it revises, and it usually reports a point estimate with no odds and no attribution. Three properties separate an instrument from a snapshot:
- Odds, not points. "We forecast 102" invites debate about the number. "Odds of hitting the target fell from 74% to 68%" invites a decision, because probability moving is unambiguous even when the point estimate barely shifts.
- Attribution, not narrative. The odds moved because of measurable drivers, ranked by contribution, not because of the story the loudest region told in the review.
- Interventions tested before commitment. If the response is a price action or a reallocation, it is run against scenarios first, so the meeting compares outcomes rather than opinions.
A plan is not a promise about December. It is a standing question: what are the odds now, and what moves them?
The cultural consequence is larger than the technical one. When odds are visible all year, the December meeting loses its drama, and that is the point. "Did we make it?" becomes a formality because "will we make it?" was answered, and acted on, forty times already.
Keeping the year's odds continuously visible, attributed and testable is what Prophesee's Horizon engine does. Ask it about your own year. Start here.