How to Set Up Sales Pipeline Stages That Match How You Actually Sell

Stages should be defined by what the buyer has done, not by what the seller thinks.

Compare:

The left column measures optimism. The right column measures reality. A forecast built on the left is a mood; a forecast built on the right is a projection.

This single change is worth more than any other pipeline advice.

Adapt to your motion — this is a starting point, not a template to adopt unchanged.

1. New / Unqualified. A lead exists. Nobody has confirmed it's real yet. Exit criteria: qualification attempted and outcome recorded.

2. Qualified. Confirmed there's a genuine need, rough budget plausibility, and access to someone who can decide. Exit criteria: meets your qualification bar. (See How to Qualify a Lead.)

3. Discovery. You understand their situation — the problem, the current state, what happens if they do nothing. Exit criteria: discovery call completed, notes logged, decision process mapped.

4. Demo / Evaluation. They've seen the product against their actual use case. Exit criteria: demo delivered to at least one decision-maker.

5. Proposal. Commercial terms are with them. Exit criteria: proposal sent and receipt confirmed.

6. Negotiation. Terms are being worked. Legal and procurement are engaged. Exit criteria: open points documented with owners.

7. Closed Won / Closed Lost. Terminal. Closed Lost always carries a reason code.

Five to seven for most B2B motions.

Fewer than five and stages become too coarse to forecast against — everything sits in “in progress.” More than eight and reps stop maintaining them, which is worse than having too few. Every stage you add is a small tax on the person doing the actual selling.

The test: can a rep move a deal without thinking hard about which stage it belongs in? If it takes deliberation, you have too many, or your criteria are ambiguous.

A stage without written exit criteria is a label. Write down, for each stage, exactly what must be true before a deal leaves it.

Rules for good criteria:

• Observable. “Champion identified” is vague. “Champion named in CRM with role and confirmed influence” is checkable.
• Buyer-side wherever possible. Something they did beats something you did.
• Binary. Yes or no, no judgement call.
• Documented in the CRM, not in a wiki nobody opens.

Where a CRM supports it, enforce criteria as required fields — a deal can't advance to Proposal without a value and a close date. Some teams find that heavy-handed; the alternative is a pipeline full of deals with no numbers attached, which is what makes forecasts meaningless.

Each stage carries a probability, used for weighted forecasting.

Don't invent them. The default weights that ship with most CRMs (10%, 25%, 50%, 75%) are placeholders, not data.

Derive them from your own history. Look at your last 12 months of closed deals: of everything that reached Proposal, what percentage actually closed? That's your Proposal probability. If it's 40%, use 40% — not the 75% the CRM suggested.

If you don't have enough history yet, start with the defaults, mark them explicitly as estimates, and replace them as soon as you have thirty or so closed deals to learn from. (More: How to Build a Sales Forecast.)

1. Stages that describe seller activity. “Followed up” isn't a stage. It's a task. Stages should track the buyer's progress toward a decision.

2. A “Nurture” stage inside the active pipeline. Deals with no momentum sit in Nurture inflating your pipeline number indefinitely. Nurture is a different list, not a pipeline stage.

3. No stage-skipping rule. If deals routinely jump from Qualified to Negotiation, your middle stages don't reflect reality — either fix them or remove them.

4. Closed Lost without a reason. The single most valuable data you collect, and the most commonly skipped. Use a fixed dropdown (price, timing, competitor, no decision, lost to status quo) rather than free text, or it's unanalysable.

5. Setting it once and never revisiting. Your sales motion changes. Review stages every couple of quarters against what actually happens.

Every deal should have a maximum time-in-stage. When it's exceeded, the deal is flagged.

Set thresholds from your own cycle data — if a healthy deal spends two weeks in Proposal, flag at four. The point isn't to punish anyone; it's that stalled deals are the primary source of forecast error, and they're invisible unless something surfaces them.

Most teams discover, when they first run this, that a meaningful chunk of their “pipeline” hasn't moved in months. That's the number leadership actually needs.

1. 1.Map how you really sell. Interview your reps. Trace three recent won deals and three lost ones, step by step.
2. 2.Draft 5–7 stages, each defined by a buyer action.
3. 3.Write exit criteria — observable, binary.
4. 4.Derive probabilities from historical close rates.
5. 5.Set time-in-stage thresholds per stage.
6. 6.Build a Closed Lost reason list — fixed options.
7. 7.Configure it in the CRM, with required fields where it matters.
8. 8.Train the team on the criteria, not just the names.
9. 9.Review after one quarter and adjust.

A pipeline isn't a reporting structure imposed on selling — it's a shared definition of where things actually stand. When stages are defined by buyer behaviour with written exit criteria, three things follow: the forecast becomes defensible, stalled deals surface early, and pipeline reviews stop being negotiations about optimism.