Every sales manager knows the moment: you open the pipeline at quarter end and a third of the deals have been sitting in the same stage for two months. The usual response is to ask the reps what's happening with them, get a vague answer, and move on.
That response assumes the problem is the rep. Usually it isn't. A deal stalls when nobody defined, in advance, what has to be true for it to move — and without that definition there is no such thing as "stuck". There is only "still open".
In most portals we're brought into, the stages are named after what the seller did: Proposal sent. Call completed. Meeting booked. It sounds sensible, and it is exactly why deals stall.
A stage named after a seller action is finished the moment that action happens. It has no exit criterion, because there is nothing the buyer has to do for the deal to progress. "Proposal sent" is still true two months after you sent the proposal and nobody replied. The system reports the deal as correctly placed, so it doesn't look stuck — it looks active.
A stage named after what the buyer did behaves differently. Proposal reviewed with the decision maker. Budget approved. These are verifiable states, and it is obvious when they haven't happened. The moment a stage has an exit criterion, the word "stuck" means something.
This is not a renaming exercise. It is the question of whether your pipeline describes the buying process or the selling process. Only the first one can tell you when something has stopped.
The distinction that matters: time in stage is not the same as inactivity. A deal can be full of activity — emails, calls, meetings — and still not have moved a single step forward. Reporting built on "last touched" misses precisely those deals, and they are the worst ones, because they look healthy.
Open deals, grouped by stage, showing time in the current stage alongside the median time that won deals spent in that same stage. The comparison is the diagnosis. A deal sitting at three times the median for its stage doesn't need a discussion; it needs a decision.
An expected median, and a threshold past which action is required. These aren't board-report metrics — they are the threshold values your automation will read. Without them every alert is arbitrary, and the team learns to ignore alerts inside a fortnight.
The most common cause and the easiest to fix. Every stage needs one sentence answering what must be true before we move on. If you can't write it, the stage isn't real — it's a place to park deals.
A deal moves when a specific person does a specific thing on a specific date. If the next step doesn't exist as a task with an owner and a date, it doesn't exist. Quick audit: how many of your open deals have no future task on them right now? In most portals we see, the answer is between half and two thirds — and that is the list of deals that will stall next month.
In complex B2B deals you don't stall against your contact. You stall against someone you aren't talking to: the manager who signs off budget, legal, the security reviewer. Your contact keeps replying politely, everything looks fine, and the deal doesn't move because the blocker isn't in the conversation at all.
This is why mapping stakeholders is a management tool rather than documentation. If you know who has to say yes and when you last spoke to them, you can see the block. If you don't, all you see is a quiet deal.
The choice that decides whether this works: build the workflow on time in stage, not on inactivity. An inactivity alert misses the busy-but-stalled deals and generates noise about deals that simply don't need touching this week.
A structure that works:
What not to do: don't move deals between stages automatically, and don't auto-close them as lost. Automation that moves deals without a human deciding destroys the reporting you were trying to fix.
Before you fix the process, clear what has accumulated. A pipeline holding deals from last year won't forecast reliably no matter how good the new rules are.
Work through every deal past its threshold and make one of three decisions:
The rule we work to: a deal you can't name a next step and a date for is not a deal in the forecast. You can leave it open. Don't count it.
Almost always because the stage they're sitting in has no exit criterion. When stages are named after seller actions — "proposal sent", "call completed" — there is nothing the buyer must do for the deal to move, so it can sit there indefinitely without the system flagging anything as wrong.
Compare time in the current stage against the median time won deals spent in that same stage. Two to three times the median is the signal. Measure time in stage rather than inactivity — the most dangerous deals are the ones with plenty of activity and no progress.
Trigger on time in stage past a threshold set for that specific stage. The first action should set a field and open a task for the deal owner, not email the manager. Escalate to the manager only if the deal hasn't moved for another week. Automation shouldn't move or close deals by itself.
No. Auto-closing cleans the report without cleaning the problem, and it destroys the information about why the deal stopped. A manual pass with an explicit decision — revive, push, or close with a reason — is worth the hour.
IV-Lead builds sales processes in HubSpot for B2B companies in Israel and abroad, as a HubSpot Gold Solutions Partner. We define stages by buyer behaviour, write an exit criterion for each one, and build the alerting — in English and Hebrew.