How to Calculate Sales Cycle Length by Stage

How do you calculate sales cycle length by stage?

For each opportunity, subtract the date it entered a stage from the date it entered the following stage, then take the median of that duration across all deals in the period.

``` Days in stage = Date entered next stage - Date entered this stage Stage cycle length = Median days in stage across the deal set ```

Repeat once per stage and the sum of the medians for won deals reconstructs the full cycle. That reconstruction is the point. A team quoting a 76-day average cycle knows one number. A team that knows discovery takes 12 days and proposal takes 21 knows where the time goes and which stage to work on.

Where does the stage timing data come from?

The opportunity history table, which logs every stage change with a timestamp. Current-state fields hold only the latest value, so they cannot answer how long a deal sat anywhere. The history object is the only source that can.

Two constraints apply. Field history tracking has to have been enabled on the stage field, and the usable window starts the day it was turned on. Backfilling is not possible, so a team enabling it this quarter has one quarter of data. Second, history rows record every change including backward moves, so a deal that went from proposal back to discovery and forward again produces multiple entries for the same stage. Decide whether to sum those entries or count only the final pass, and apply the choice consistently.

Should you measure time in stage or time to stage?

Both, because they answer different questions. Time in stage isolates one step and points at the specific activity that consumes the days. Time to stage is cumulative from creation and tells you how far into the cycle a deal is when it reaches a given point, which is what a forecast needs.

The cumulative view also produces usable thresholds. If deals reach negotiation at day 51 on average, an opportunity in negotiation at day 140 is behaving unlike anything that has closed before, regardless of what the rep says about it.

What does a stage duration table look like?

Won and lost deals diverge at every stage, and the gap widens as deals move later in the funnel. Winning deals complete the cycle in 76 days. Losing deals take 168 days to reach the same endpoint and produce nothing. Blending the two sets returns a cycle length somewhere in between, which describes no real deal and overstates how long a healthy opportunity should take.

Negotiation and contract show the widest ratios, both near three to one. A deal sitting in negotiation past 46 days is following the losing curve, and that threshold is more useful than any stage probability field. Pair it with win rate by stage to see both how many deals convert from a stage and how long the survivors take.

Why do stage curves matter more than a single average?

Because close timing is a distribution, and a single average discards the shape of it. ORM groups opportunities with a machine learning model and predicts a close curve for each group. Those curves run from 1 to 80 weeks, with most of the expectation landing before week 12 and very few groups extending past 52 weeks.

That structure means two deals created on the same day can carry completely different reasonable timelines based on the group they belong to. A stage average flattens that into one number and then treats every deviation from it as noise. Stage-level medians, split by outcome, recover enough of the shape to be operational without needing a model.

What does a lengthening stage tell you?

Which part of the buying process changed, and whether the cause is inside or outside your control.

Late-stage expansion points at the buyer's side. Procurement scrutiny, legal review, and budget approval all live in negotiation and contract, and time growing there usually means deals are being examined harder rather than sold worse. Buyer uncertainty produces exactly this pattern, stretching the span from qualified to closed while volume and deal counts hold steady.

Early-stage expansion points at your side. Discovery and solution fit stretching out means opportunities are being created before they are real, and reps are spending weeks qualifying deals that should have been disqualified in the first call.

The two situations call for opposite responses, and only the stage split distinguishes them. A blended cycle length rising from 76 to 94 days says something got slower and stops there.

What breaks stage timing data?

Skipped stages, bulk updates, and deals that never leave.

Skipped stages are the most common. A rep moving an opportunity from discovery straight to proposal produces a zero-day solution fit record, and enough of those pull the stage median toward zero. Exclude stages a deal never entered rather than recording them as instant.

Bulk updates create the opposite distortion. When an admin reclassifies 400 opportunities in an afternoon, every one of them gets a stage change timestamped that day, and the following month of stage duration data is fiction.

Deals that never exit a stage are excluded by definition, since the calculation needs two timestamps. That exclusion systematically removes the slowest opportunities and shortens every median. Track the count of deals sitting past twice the median for each stage as a companion metric, and watch deal slippage alongside it, since a rep pushing a close date is the clearest signal that a stage duration is about to run long. The cycle number also feeds directly into the velocity calculation covered in sales velocity, so an understated cycle inflates that metric too.