Why deals go cold: building a sales pipeline that doesn't leak
Most deals aren't lost to a competitor. They're lost to silence — a follow-up nobody owned, a next step nobody wrote down, and a stage that meant something different to everyone looking at it.
If you go back through the deals you didn't win last year, a decent share of them will have no dramatic ending. No "we went with someone else." No price objection. Just a conversation that was going well, then a gap, then nothing.
That's not a sales skill problem. It's a systems problem, and it shows up the same way in almost every business we've built software for.
Three reasons deals go quiet
- No owner. A deal that belongs to "the team" belongs to nobody. The moment two people could plausibly be following up, neither reliably does.
- No next action. "Waiting to hear back" is not a next action, because it has no date and requires no one to do anything. It's how a deal quietly becomes a memory.
- No date. An action without a date is a wish. It surfaces only when someone happens to scroll past it.
Every deal, at every moment, should have an answer to: who owns it, what happens next, and when. If any of the three is blank, the deal is already drifting.
Pipeline stages that actually mean something
Most pipelines are decorative. They describe a feeling ("warm", "interested", "negotiating") rather than a fact, so two people looking at the same deal put it in different columns and forecasts become fiction.
The fix is to define each stage by an exit criterion — a verifiable thing that must be true for the deal to move on. For example:
- Qualified — we know their problem, their rough budget, and who decides.
- Proposed — a written proposal has been sent, with a number in it.
- Verbal — the decision-maker has said yes; paperwork is outstanding.
- Won — signed, or a deposit has landed.
Notice that each of those is checkable by someone who wasn't in the room. That's the whole test. If a stage can't be verified by a third party, it isn't a stage — it's an opinion, and it will make your pipeline value meaningless.
A quick pipeline audit
- Can you state the exit criterion for each stage in one sentence?
- Does every open deal have an owner, a next action, and a date?
- What's the oldest deal still sitting in an active stage — and is it actually alive?
The follow-up cadence
Persistence beats charm, and structure beats persistence. A simple, defensible rhythm after a proposal goes out:
- Day 2 — confirm they received it and ask if anything needs explaining. Low pressure, high response rate.
- Day 7 — a short nudge tied to something useful: an answer to a question they raised, a relevant example.
- Day 14 — ask a direct question. "Is this still something you want to do this quarter?" A clear no is worth more than an open deal that isn't real.
- Day 30 — move it to a long-cycle nurture and stop counting it in the forecast.
The last step is the one people skip, and it's the most valuable. A pipeline stuffed with deals that will never close doesn't just flatter your forecast — it hides the ones that deserve attention this week.
Automate the timing, not the relationship
The right split is almost always the same. Automate anything that is a trigger, and keep anything that is a judgement.
Worth automating:
- Creating the follow-up task when a proposal is sent
- Moving a deal to a review state when it's been untouched for N days
- Assigning an owner the moment a lead is captured, so nothing lands ownerless
- Notifying someone when a deal crosses a value threshold
Worth keeping human:
- The actual wording of a follow-up to a specific person
- Whether to discount, and by how much
- When to walk away
Automated nudges that read like automated nudges do real damage. Let the system remember the date; you write the sentence.
The tool matters less than the discipline — up to a point
Plenty of businesses run a decent pipeline on a whiteboard or a spreadsheet, and for a solo operator that's often the right call. The wheels come off at predictable moments: when a second person needs the same view, when someone leaves and takes the context with them, or when you want to know which channel actually produced the deals you won.
At that point the requirement isn't "a CRM" in the abstract. It's something fast enough that people actually update it — because a pipeline nobody maintains is worse than no pipeline at all. It gives you confident, wrong numbers.
Start with the three questions. Owner, next action, date — on every open deal, today. Most of the leak closes right there, before any tooling changes.