A sales pipeline that tells you something

Most pipeline reports are built from two fields a salesperson fills in optimistically: the stage and the expected close date. Neither is verified, both affect the forecast, and the person entering them is measured on the result. That is not a data problem to be solved with a better chart.

Give each stage an entry criterion, not a name

A stage called Negotiation means whatever the person dragging the card believes it means. A stage that requires something observable - a written proposal sent, a decision maker identified, a quotation issued from the system - means the same thing to everyone, which is the only property that makes a total addable.

The criteria do not need to be elaborate. One verifiable fact per stage is enough, and it should be something that exists in the system rather than something asserted in a note. A stage that advances when a quotation document exists cannot be advanced by enthusiasm.

Forbid the transitions that make a funnel meaningless

If an opportunity can move from first contact straight to won, you cannot measure conversion between stages, which is the one thing a funnel is for. Define which transitions are allowed, allow backwards movement explicitly - deals do go cold and pretending otherwise is how a pipeline fills with the undead - and record who moved it and when.

The history is what makes the pipeline analytical rather than decorative. Average time in stage, the stage where deals stall, the difference between two salespeople at the same stage: all of these need the transitions, not the current state.

An expected close date that is not a wish

The single most useful discipline is that a close date in the past is an exception that must be resolved, not a value that silently ages. Every pipeline with no such rule accumulates deals whose dates passed months ago, and their value is still in the forecast total.

Two habits fix most of it: a weekly review of overdue dates where each one is either re-dated with a reason or closed as lost, and a rule that moving the date is recorded. A deal whose date has moved five times is not a forecast, it is a pattern.

Lost is information, not failure

A deal closed as lost with a reason is worth more than one left open forever. Reasons need to be a short, fixed list - price, timing, competitor, no decision, wrong fit - because free text cannot be counted, and the count is the point.

No decision deserves its own category and is usually the largest. It is also the one most responsive to process changes on your side, which you cannot discover if it is recorded as price.

Connect it to what actually happened

A pipeline that ends at won and never meets the invoice tells you about intentions. Tie the opportunity to the quotation, the quotation to the order, and the order to the invoice, and you can finally compare what was forecast to what was delivered and at what margin.

That comparison is usually uncomfortable the first time, which is a good sign: a pipeline that always agreed with the outcome was not measuring anything.