Most pipelines measure what the seller did. Open your CRM and read the stage names: Proposal Sent, Demo Completed, Follow-Up Scheduled. Every one of those describes your effort. None of them describes a decision.
A pipeline stage is a claim about the buyer. If the only thing that changed is something you did, the deal did not move.
Activity-named stages produce a forecast nobody trusts
Run the math on a quarter. Say 22 open deals sit in Proposal Sent, average contract value $24,000, and your CRM weights that stage at 60 percent. That is $316,800 in your commit number.
Now redefine the stage. A deal only counts as Proposal Sent if the buyer acknowledged the proposal, named a decision date, and pulled a second stakeholder onto the thread. Apply that to the same 22 deals and maybe 7 clear it. The other 15 fall back to an earlier stage weighted at 20 percent. Your commit number becomes $172,800.
The $144,000 difference was never pipeline. It existed because a PDF left your outbox.
That is not a stage. It is a timestamp.
Exit criteria are the design
Every stage needs one observable buyer action required to leave it, written down, stored in the CRM rather than in a rep’s head. HubSpot and Pipedrive both give you a description field on every stage. Almost nobody fills it in.
Useful evidence costs the buyer something. A calendar hold they accepted. A redline on your MSA. A security questionnaire routed to you by someone in IT. A procurement contact introduced by name. A budget owner stated out loud. Anything that took their time or their political capital.
What does not count is enthusiasm. A buyer who says the timing looks good has spent nothing.
Fewer stages, harder gates
Firms under 50 people usually run too many stages, which hides the problem by spreading it thin. Four is enough for most professional services sales. Qualified, Scoped, Decision, Closed. Each with a written gate, and no card moving on feel.
Then add a decay rule. Any deal with no buyer action in 21 days moves backward or comes off the board. Not “nurture.” Off. Stale deals are why your average cycle reads 90 days when your real cycle is 34 and your corpses are dragging the mean.
The gate matters more than the tooling, though a gate is hard to install when the CRM was configured by whoever had a free afternoon. That is a separate repair job, and it comes first: the object model has to hold a real pipeline before stage discipline has anywhere to live.
Hard gates also surface dead deals early, which feels like losing pipeline and is actually reclaiming hours. Disqualifying faster is the cheapest growth lever a small firm has, and stage criteria are how you make it automatic instead of a judgment call at 4pm on a Friday.
Behavior-based stages tell you where you actually lose, because the drop-off finally has a location. For most firms the leak sits past the proposal, in the follow-up nobody owns.
Run this before Monday
Open your CRM, go to your most crowded stage, and pull the five oldest deals in it. For each one write down the last thing the buyer did and the date they did it. Not the last thing you did. If you cannot name a buyer action inside the last three weeks on at least three of those five, your stage definitions are decorative and your forecast is a guess with a dollar sign in front of it.
Fixing it takes an afternoon of writing gates and a quarter of holding the line on them. The second part is where most firms quit.
Building revenue and growing it both depend on a pipeline that tells the truth about where deals stand. That is the work I do at Florida Man Innovations: design the system, then hand it over so it runs without me in the room. Book time at meet.brettfl.com or write to [email protected].
The buyer moves the card.