Forecast accuracy
Stage is not probability
By Starvik Team · · 2 min read
Mapping a pipeline stage to a fixed percentage records what your team has done. It says almost nothing about what the buyer will do.
Nearly every pipeline attaches a probability to a stage. Discovery is ten percent, demonstration is thirty, proposal is sixty, and so on. The numbers are usually inherited, rarely revisited, and almost never checked against outcomes.
What a stage actually records
A stage records a seller side event. A demonstration happened. A proposal was sent. These are facts about your activity, and they are genuinely useful for knowing what to do next.
What a stage does not record is anything about the buyer's intent. Two deals can both be at proposal: one because the buyer asked for a proposal to start a procurement process, the other because the seller sent one unprompted to create momentum. Treating those as the same sixty percent is the error.
How the error compounds
The error is not symmetrical. Stages only ever move forward in practice, because moving one backward requires admitting something, and the system makes that awkward. So probabilities ratchet up and rarely come down, even as the underlying situation deteriorates.
Across a full pipeline, that produces a number that is biased high in a way that no amount of averaging corrects.
A better input
The useful question is not which stage a deal is in. It is when the buyer last did something, what it was, and whether it can be verified.
| Deal | Stage | Last buyer action | Honest read |
|---|---|---|---|
| A | Proposal | Requested pricing nine days ago | Live |
| B | Proposal | Attended a demonstration in March | Stalled |
| C | Discovery | Posted a role for this exact function | Live and early |
Deal B and deal A carry the same stage probability and are not remotely the same deal. Deal C is earlier by stage and healthier by evidence.
What to change
You do not need to abandon stages. They are useful for process. The change is to stop letting stage be the only input to probability, and to add one more field that is harder to fudge: the date and nature of the last verifiable buyer action.
Sorting by that field tends to reorder a pipeline substantially, which is itself the evidence that stage was carrying more weight than it earned.
Find the decision-makers in your market
Three free searches, no card required.