Monday's pipeline review runs the same way it always does. The sales head reads down the forecast, calling out each deal by name. "Negotiation, eleven weeks." The rep who owns it says, "yeah, that one's basically there" — though nobody in the room, including the rep, can say when they last actually spoke to the customer. Two rows down, a smaller deal still sits in "Qualifying," even though the purchase order for it arrived last Thursday and is sitting, unread by the CRM, in someone's inbox. The forecast number gets read out at the end with a straight face, and everyone nods, and everyone in the room privately discounts it by some amount they never say aloud. Nobody lied. The stage field is just the last thing anyone had time to click, and clicking it has never been anybody's job.
Why we build it this way
The usual answer is more discipline: a rule that a stage gets touched after every call, a red flag on anything untouched for two weeks. That puts the fix on the person least able to give it the time — a rep mid-quarter is going to spend the free five minutes on the customer who might actually buy, not on a dropdown. Nag them enough and they stop being honest about it; they pick whichever stage is least likely to trigger the reminder, not the one that is true.
The stage was never really the rep's to know from memory anyway. It is a summary of things that already happened and are already written down somewhere: a quote that went out, a second call that got made, a purchase order that landed. None of those needs to be recalled. They need to be read.
So the fix is not to ask a person to translate what happened into a dropdown more often. It is to read what happened, propose what the dropdown should now say, and hand that proposal to a person who only has to check whether it is right — which takes seconds, not the ten minutes it takes to reconstruct an account from memory.
A pipeline a person has to remember to update is a wish list. A pipeline that reads what happened is a record — and a person only has to say whether the record read it correctly.
How it works, roughly
Setup is the same corpus the rest of the pipeline already draws on — the emails, quotes and calls already on file against each account. From there it watches for the kind of event that should move a deal, and reads:
- A quote or a revised quote going out or coming back
- A call or a meeting logged against the account
- A purchase order, a signed confirmation, or a message that reads like a decision
When one of those lands, it proposes a stage change or a next action, with the exact document or call it is pointing at attached to the proposal — never a stage moved on its own say-so. A person looks at the evidence and accepts it or rejects it. Accept it, and the pipeline moves. Reject it, and that correction is what the next proposal on that account reads from, not the same guess coming back a week later.
A year of these small confirmations and the Monday forecast stops being a number everyone quietly discounts. Every stage on the board points back to something that actually happened, not to whoever last had a spare five minutes — and the history of who moved what, and on what evidence, stays inside your own record, not scattered across memory.
See how the agents stay scoped →