Pipeline stages exist to make a forecast possible. A stage that means “the rep feels good about this” produces a forecast that means nothing, which is why most pipeline reviews consist of a manager asking why last month’s commit did not close.
The fix is defining each stage by something observable that the buyer did.
Buyer actions, not seller activities
This is the whole principle. “Demo completed” is a seller activity — it tells you what you did. “Buyer confirmed the demo addressed their requirements and introduced two colleagues” is a buyer action, and it tells you something about whether the deal is real.
Every exit criterion should be a fact someone else could verify from the record. Not a judgement, not a feeling, not a stage the rep moved it to because the quarter is ending.
A workable stage set
Five to seven stages suits most B2B sales. Fewer and the forecast is coarse; more and reps stop maintaining them accurately.
1. Qualified. Exit criteria: the buyer has described a specific problem in their own words; they fit the ideal customer profile; there is a plausible timeframe; you have identified who else is involved.
2. Discovery complete. Exit: you understand the current process, what it costs them, their decision criteria and the decision process. The economic buyer is identified by name.
3. Solution validated. Exit: the buyer has seen the product address their specific case and has confirmed it does. Technical questions resolved. Any required evaluation completed.
4. Business case agreed. Exit: pricing has been discussed with someone who can approve it, the scope is agreed, and the buyer has confirmed the value justifies the cost. A champion is actively advocating internally.
5. Procurement and legal. Exit: contract under review, security review underway or complete, the approval path is known with named approvers.
6. Closed won or lost.
Adapt the middle stages to your own process, and keep the principle: each exit criterion is a verifiable buyer action.
Probability, and its misuse
Attaching a percentage to each stage is standard and routinely abused. Two rules make it useful.
Derive the percentages from history, not from intuition. Calculate what share of deals that reached each stage actually closed over the last year. The real numbers are frequently much lower than the assigned ones, particularly in the middle stages.
Do not let reps adjust them per deal. The stage determines the probability. A rep who believes a deal is more likely than its stage suggests should advance the stage by meeting the criteria, not override the number.
Stage ageing, the most useful report nobody runs
Track how long deals sit in each stage, and compare each open deal against the historical median for that stage.
A deal in “solution validated” for three times the median is not progressing, whatever the rep says. This single report identifies most of the deals that will not close long before the forecast does, and it requires no judgement from anyone.
Set a maximum age per stage. Deals exceeding it get reviewed explicitly and either advanced, re-dated or moved out of the active pipeline.
Keeping the pipeline honest
Four practices, each addressing a specific form of inflation.
Require the exit criteria to be recorded, not asserted. A field saying who the economic buyer is, by name. Empty means the stage is not met.
Allow deals to move backwards. If a champion leaves and you lose access, the deal is not in late stage any more. Pipelines where deals only move forward are pipelines that lie.
Enforce close dates. A date that has been pushed three times is not a date. Require a reason and a new one tied to a buyer-side event.
Review losses as rigorously as wins. Which stage did they die in, and why. This is where the pattern that is costing you deals becomes visible.
What the pipeline tells you
Beyond the forecast, the stage data answers diagnostic questions.
Conversion rate between stages shows where deals die. A sharp drop between discovery and solution validation suggests a product or positioning gap; a drop at procurement suggests you are not engaging the right people early enough.
Pipeline coverage — total pipeline value divided by the target — tells you whether there is enough in the system. Three to four times is a common benchmark, though the right figure depends on your actual win rate rather than a rule of thumb.
Average cycle length by stage shows where time goes, which is usually not where people assume.
Common mistakes
Too many stages. Ten stages produce precision nobody maintains.
Stages named after seller activities. “Demo scheduled”, “proposal sent”. These measure effort, not progress.
No definition of qualified. If anything can enter the pipeline, the pipeline means nothing. See lead qualification frameworks.
Changing the stages frequently. Every change breaks historical comparison. Get them right and leave them alone for at least a year.
Running a pipeline review that is worth the hour
Most pipeline reviews consist of a manager asking about each deal and a rep describing optimism. Three changes make them useful.
Review by exception, not by list. Do not walk through every deal. Filter for deals that are ageing beyond the stage median, deals whose close date has moved, and deals above a value threshold. That is where the time should go.
Ask about the buyer, not the rep. “What did they do last?” rather than “what’s your next step?” Buyer actions are facts; seller next steps are intentions.
Make it safe to be pessimistic. If a rep is punished for moving a deal backwards or pushing a date, they will not do it, and the forecast becomes fiction. The behaviour to reward is accuracy, not optimism.
Forecasting on top of the stages
Stage-weighted pipeline is one input to a forecast, not the forecast itself. Three views together produce a more reliable number.
Weighted pipeline — each deal’s value multiplied by its historically derived stage probability. Mechanical, unbiased, and poor at the top end where a single large deal distorts it.
Rep commit — what each rep believes will close. Biased optimistic, and valuable because it contains information the CRM does not.
Historical run rate — what the team closed in comparable recent periods. Dull and frequently the most accurate of the three.
When all three agree, the forecast is reasonably safe. When they diverge sharply, the gap is the thing to investigate rather than an average to take.
Changing stages without losing history
Redefining stages is occasionally necessary and it breaks every historical comparison, which is why it should be done rarely and carefully.
Map old stages to new ones explicitly before the change, so historical deals can still be analysed. Keep the old stage recorded on closed opportunities rather than overwriting it. And make the change at the start of a quarter, not mid-period, so the reporting boundary is clean.
Then leave it alone for at least a year. Conversion rates between stages only become meaningful once enough deals have passed through the same definitions, and a team that adjusts its stages every few months never gets that data.
Where pipeline comes from
Stage definitions govern what happens to opportunities once they exist. The separate question of generating enough of them determines whether the pipeline has anything to measure.
For teams relying on outbound, the arithmetic is unforgiving and worth doing honestly before setting targets: cold email benchmarks set out what reply and meeting rates to expect, and therefore how much activity a given number of qualified opportunities requires.
Frequently asked questions
How many pipeline stages should there be?
Five to seven for most B2B sales. The test is whether each stage represents a genuine change in the probability of closing.
What is good pipeline coverage?
Three to four times the target is commonly cited. Derive your own from your actual win rate: if you win a quarter of qualified deals, you need four times coverage.
Should deals be allowed to skip stages?
Rarely, and when it happens the criteria for the skipped stage should still be recorded as met. Allowing skips without evidence defeats the purpose.
Who owns pipeline hygiene?
Reps maintain it, managers enforce it in weekly reviews. Without consistent enforcement it degrades within a quarter.
