Pipeline stage

The stages that work are exit criteria written as facts: a meeting happened, a technical evaluation was scheduled, a quote was sent, a security review started. Anyone can look at a deal and agree on which stage it is in.

The stages that fail are named after feelings — interested, engaged, hot. Two reps will place the same deal differently, the forecast built on top becomes noise, and the standard response is to add more stages, which makes it worse.

Five or six is usually enough. Every additional stage divides the same deal count into smaller cohorts, and conversion rates computed on small cohorts move around for no reason. A pipeline with twelve stages produces twelve statistics nobody can act on.

The useful test: if moving a deal between two adjacent stages does not change what you would do next, the two stages are one stage.

Attaching a win probability to the stage rather than to the individual deal is what stops forecasting from being a poll of the sales team. It says: historically, deals that reached this point closed this often. Reps can then be responsible for the stage, which they observe, rather than for the probability, which they guess.

Those numbers should come from your own closed history and be revisited, not inherited from a template.

SalesShift pipelines are made of stages, each carrying a win probability, and deals move between them on the same record that holds the signal, the sequence and the contract. The forecast reads the stage probability rather than a per-deal guess.