What a Pipeline Stage Actually Is

Open your pipeline and read the stage names.

Interested. Warm. Nurturing. Proposal. Negotiating.

Now ask what has to be true for a deal to be in Warm rather than Interested. Not what it feels like — what has to have happened.

If you can't answer that in one sentence for every stage on the board, you don't have a pipeline. You have a list of names sorted by vibe, and everything downstream of it — the forecast, the follow-up, the sense of whether this month is going to be fine — is running on that.

Adjectives can't be wrong

The problem with a stage called Warm isn't that it's vague. It's that it's unfalsifiable.

A deal can sit in Warm forever. There's no condition it violates, no point at which it obviously doesn't belong there anymore, and no way for anyone else to disagree with the placement. Which means the stage carries no information — it's a record of how you felt about a conversation, and feelings about conversations decay while the label doesn't.

A stage you can't be wrong about is a stage that can't tell you anything.

Watch what this does to your board over six months. Deals accumulate in the middle stages, because nothing pushes them out. The total value climbs. Your sense of the pipeline gets better while your actual pipeline gets older, and the gap between those two things is what produces the quarter where the number doesn't arrive.

The four things a real stage has

A stage is a state, and states have properties. Four of them, and each one does a specific job.

An entry condition tied to an observable event

Something happened, in the world, that a second person could verify. A call took place. A proposal was sent. A specific question was answered.

Proposal is a legitimate stage name because a proposal was either sent or it wasn't. Interested is not, because interest is a state of somebody's mind and you have no instrument for it.

This is the property that makes everything else possible. Once entry is an event, the stage has a timestamp — and a timestamp is what makes age, velocity and conversion computable.

An owner

One name. The person responsible for this deal moving out of this stage.

Founder-led pipelines almost never have this, because the answer is obviously the founder for all of them, so writing it down feels redundant. It isn't. The moment there are two people, or the moment you want to know why nothing moved last week, ownership is the field that answers it.

A timer

How long a deal should reasonably stay here. Not a rule — an expectation, against which the exceptions become visible.

This is the property that turns a stalled deal into a fact rather than a feeling. A deal that's been in Proposal for nineteen days when your median is four isn't a deal you should follow up on. It's a deal that has already told you something, and the board should be saying so without anybody reviewing it.

An exit — including the one nobody defines

Every stage needs at least two exits: forward, and out.

The forward exit is the entry condition of the next stage, which means your stages are actually a chain of events rather than a set of labels. The out exit is the one that's almost universally missing, and it's the one that keeps the board honest.

The test that tells you which kind you have

Ask two people in your business to place the same five deals on the board, independently, and compare.

If they agree, your stages are events. If they don't, they're adjectives — and the disagreement isn't a training problem, it's the board telling you it has no definitions.

You can run a single-person version of the same test. Take a deal, look at its stage, and ask what would have to happen for you to be wrong about that placement. If nothing would, the stage isn't doing any work.

Rebuilding it: fewer stages, harder edges

The instinct when fixing this is to add stages, and it's the wrong direction. More granularity in a system with no definitions just produces more places for deals to sit.

Start from the events that actually happen in your sales process, in order. For most founder-led service businesses there are four or five, and no more.

Someone made contact. A first conversation happened. They qualified — meaning specific things you can name are true. A proposal went out. They agreed and signed.

That's it. Each of those is an event. Each has a natural owner. Each has a plausible timer you can derive from your own history in about twenty minutes. And the edges between them are hard, because the edge is did the thing happen.

Then add the exits. Lost, with a reason. Dormant, with a definition — no next step and no contact for a defined period. Both of those need to be states a deal actually moves into rather than a status somebody remembers to set.

Make interest easy to express and impossible to lose. A board with hard edges is what makes the second half of that sentence true.

The field that matters more than the stage

If you only change one thing, change this instead: put a scheduled next step on every open deal.

Not a task list — a scheduled event, with a date, that exists in the system. And then make the absence of one visible.

A deal with no scheduled next step is a deal that's currently decaying, regardless of what stage it's in. That single field converts a passive board into something that surfaces its own failures, and it does more for a founder-led pipeline than most stage restructures.

It's also the thing that makes follow-up stop depending on your memory. When your system handles the remembering, you get to focus on the connecting — and remembering is precisely the faculty that fails during the months when you land good work, which are the months your pipeline most needs to keep moving.

Why founders resist hard edges

Two objections come up, and one of them is right.

The wrong one is that hard stages make the pipeline look worse. They do, immediately and dramatically — the first time you apply real entry conditions, a third of your board moves backwards or disappears. That's not the system making things worse. That's the first accurate reading you've had, and the discomfort of it is exactly the information you were missing.

The one that's right is that a rigid process can flatten a relationship-led sale. Founder-led selling genuinely is a relationship journey rather than a funnel, and a business that starts treating people as records moving through states has lost the thing that made it work.

But that objection is about behaviour, not about definitions. Hard stages don't change how you talk to anyone. They change whether your business can tell, without you, that a conversation from three weeks ago has gone quiet. The relationship stays entirely human; only the remembering gets mechanized.

Where the stages come from

If you're rebuilding, the fastest source of accurate stages isn't a template. It's your last ten closed deals, walked backwards.

For each one, list the events in order — what actually happened, with dates, from first contact to signature. Then lay the ten lists side by side.

Two things fall out immediately. The events that appear in all ten are your stages. And the ones that appear in three or four are either optional steps that shouldn't be stages at all, or a second sales motion you've been running without noticing, which is a genuinely useful thing to discover.

You'll also get your timers for free. The median gap between each pair of consecutive events is the expectation for that stage, derived from your own history rather than from a benchmark that describes somebody else's business.

Do the same exercise on five deals you lost, and you'll get the third useful output: where they stopped. Losses cluster, and the cluster is almost never where founders assume — it's usually one stage earlier, at a step nobody thought of as a step.

An afternoon, ten deals, and you have stages, timers and a conversion baseline that are all specific to how your business actually sells.

What you get back

Three things become computable the moment stages are events, and none of them are available before.

Conversion by stage. Where deals actually die, as a number rather than an impression. Almost every founder who computes this for the first time is surprised — the stage they thought was the problem usually isn't.

Cycle length. How long a deal genuinely takes, which is what makes a forecast possible and what tells you whether this month's quiet pipeline is a problem or an artifact of timing.

Stage age against the median. The single most useful operational number in a small pipeline, because it identifies the deals that need attention today without anybody reviewing the board.

Those three are the difference between a pipeline you look at and a pipeline that tells you things. And they're all downstream of one decision: that a stage means an event happened, not that somebody feels good about it.

That's what the Pipeline pillar is actually about. We've turned the word into something mechanical that has stages — but a real pipeline isn't mechanical, it's the bridge between interest and trust. The stages exist so the relationship doesn't depend on anyone's memory of where it got to.