Designing a Sales Pipeline That Reflects Reality
A bad pipeline is easy to spot: every deal piles up in one stage, and nobody can explain the difference between stage three and four.
One rule that solves most problems
Every stage must be marked by a customer action, not a sales feeling. "Has received the quote" can be verified. "Interested" cannot.
Five to seven stages is enough
A pattern that usually works: new lead, need confirmed, quote sent, negotiation, won or lost. If you have ten stages, several are probably internal steps rather than sales stages.
Write definitions and exit criteria
For each stage: what it means, what must exist before moving on, and how long it normally takes. Without this, two sales people will judge the same deal differently and the forecast becomes guesswork.
Record the reason for losing
A short list of five to seven options: price, timing, features, chose a competitor, decided not to buy, no response. After six months this becomes the most useful market information you own.
Do not trust default probabilities
Win probability per stage should be calculated from your own data after a few months, not taken from a default. See sales dashboards.
Review every quarter
A pipeline is a living document. If many deals skip a stage, that stage may not need to exist. See Zoho CRM.