The sales-cycle number changes when the stages change

A sales-cycle number is a definition before it is a benchmark. It becomes comparable only after the start event, stop event, outcome categories, and treatment of skipped or returned stages are fixed. Without those choices, two teams can calculate correctly and still compare different paths.

The problem hides in ordinary language. The statement that the cycle is six months can mean first form submission to contract, qualified opportunity to contract, first human action to contract, or first conversation to a decision of any kind. The number is not wrong until it is used as if all four paths were the same.

Why does the sales cycle metric distort when pipeline stages change?

Marvasti and colleagues model four B2B buying states: no-funnel, early-funnel, middle-funnel, and late-funnel (Marvasti et al., 2021). Their model permits a state to remain, advance, or return to no-funnel (Marvasti et al., 2021). That is important for measurement. A path that returns to an earlier state has not necessarily failed, and a path that stays in one state has not necessarily moved toward a contract.

If a cycle metric counts only the final contract path, it hides the time spent waiting, returning, or leaving the funnel. If it counts every return as a new cycle, it may turn one buying process into several observations. The correct choice depends on the decision the metric is meant to support.

The stop event changes the population

Virtanen, Parvinen, and Rollins distinguish won, lost, and canceled leads and model the three outcomes separately (Virtanen et al., 2015). That distinction changes the denominator. A team that excludes canceled leads reports a duration for decisions that reached a particular endpoint. A team that includes them reports a different object. Neither is automatically better.

Their one-company global IT setting reports an average lead time of two quarters, with 15% of leads lasting more than 12 months (Virtanen et al., 2015). Those figures describe that company’s data and definitions. They do not establish the expected cycle for another B2B market. The long tail is the more useful lesson: an average does not tell you whether a metric mixes short resolved paths with a small number of still-open or unusually long paths.

Measurement field	Option to declare	Why it changes the number	Check before comparing
Start event	Inquiry, qualified opportunity, first human action, or another named event	Earlier starts include more waiting and exploration.	Is the event recorded consistently?
State path	Straight progression, return allowed, or restart rule	A return can be progress, delay, or a new observation.	Does the system preserve state changes?
Stop event	Won, lost, canceled, contract signed, or another endpoint	Endpoints select different populations.	Are open paths excluded or censored explicitly?
Time unit	Calendar days, business days, quarters, or elapsed periods	Unit and boundary rules change duration.	Are pauses, weekends, and waiting periods treated alike?
Outcome	Won, lost, canceled, open, or separate outcome	Mixed outcomes can hide different mechanisms.	Can the reader reproduce the denominator?

Forecasts have a similar boundary problem

Goodman separates market potential from a sales forecast conditional on marketing decisions (Goodman, 1972). The distinction carries into cycle reporting. A potential cycle describes what might happen under a set of conditions. A forecast cycle describes what the current process, coverage, and decisions make plausible. Neither number can be read without its conditions.

This is why a cycle metric should carry its denominator and censoring rule. If open opportunities are removed, the metric describes completed paths. If they are included as right-censored records, the metric carries information about time not yet resolved. If they are silently moved between stages, the number loses its path.

Use the number for one decision

Start with the decision. If the goal is response time, start at the first human action. If the goal is forecast timing, start at the opportunity definition the forecast system owns. If the goal is process learning, preserve returns and canceled outcomes instead of forcing them into won or lost.

Then write the definition beside the result. A cycle number without that sentence should be treated as a label, not as evidence. The best comparison is not the one with the most precise decimal. It is the one where a reader can see which paths were counted and which were not.

The start event is a selection rule

Choosing a start event does more than set a timestamp. It selects which part of the buying process becomes visible. An inquiry-to-contract measure includes waiting, qualification, information search, and seller response. A qualified-opportunity measure excludes everything before qualification. A first-human-action measure may capture response work while excluding the period in which the buyer was still deciding whether to engage.

None of those choices is automatically correct. The choice is correct only for the decision the measure is meant to support. Marvasti and colleagues’ state model makes the point in another way: the journey can stay in a state, advance, or return (Marvasti et al., 2021). A start rule that ignores those states can turn waiting into missing data or treat a return as a new beginning without saying so.

The same discipline applies to the stop event. Contract signature, a won decision, a lost decision, and cancellation are not interchangeable endpoints. Virtanen and colleagues model won, lost, and canceled leads separately (Virtanen et al., 2015). If a report combines them, it should say why. If open paths are removed, it should say that they were removed. Silent selection is what makes a precise number misleading.

Durations need their distribution

An average is a useful summary only when the population and the spread around it are visible. The one-company global IT observation of two quarters and 15% lasting more than 12 months illustrates the issue (Virtanen et al., 2015). The long tail may represent a small group of unusually complex decisions, open paths that were still being counted, or a mixture of outcomes. The figure cannot tell which without the event definition and the underlying distribution.

A measurement record should therefore preserve more than the mean. Report the number of completed paths, the number still open, the outcome categories, and a spread measure such as a median or declared percentile when the data permit it. That is a method recommendation, not a claim about a universal sales-cycle distribution. Its purpose is to prevent a single average from hiding the selection rule that produced it.

The distribution also helps separate process questions. A longer upper tail may point to a return path, a waiting period, a complex buyer group, or a definition that begins too early. A shorter average may simply reflect that slow paths were excluded. The number becomes more useful when the possible mechanisms remain visible beside it.

A stage change is a measurement change

Suppose a team adds a qualification stage between inquiry and opportunity. The next cohort may show a longer inquiry-to-contract cycle because work that was previously implicit is now recorded. The qualified-opportunity-to-contract measure may stay stable while the end-to-end measure rises. That does not prove the process got worse. It shows that the event sequence changed.

Marvasti’s model is helpful because it treats states as part of the journey rather than as decorative labels (Marvasti et al., 2021). A new stage can represent a new state, a new observation of an old state, or a policy that filters the population. The reporting system should say which. Otherwise a process change is later read as a market trend.

Historical comparisons need the same warning. If the start, stop, stage, or restart rule changes, place a boundary in the time series. Keep the old and new definitions rather than forcing them into one trend line. A break in comparability is a result to document, not a defect to smooth away.

How should RevOps build a standardized sales cycle dictionary?

Before calculating, write a small dictionary that another person could use without asking for an oral explanation:

1. Population. Which leads or opportunities enter the measure, and which are excluded?
2. Start. What observable event begins the clock?
3. Path. Can a record remain, advance, return, or restart, and how is each action counted?
4. Stop. What ends the clock, and are open records censored or removed?
5. Outcome. Which endpoints are reported separately?
6. Version. Which stage and field definitions were active during the period?

The dictionary is small enough to sit beside a dashboard. It is also strong enough to explain why two numbers differ. In parallel, establishing unambiguous stage boundaries ensures that win-loss analysis needs an outcome before commercial teams attempt to categorize reasons for deal attrition. Goodman makes the broader measurement point by separating potential from a forecast conditional on decisions (Goodman, 1972). A cycle measure needs the same humility. It is a conditional observation of a declared path, not a property that the market carries around for us.

Boundary

The held research supports stateful journeys, separate outcomes, and a distinction between potential and conditional forecast. It does not support a universal B2B cycle benchmark or a causal claim that a longer average causes weaker performance. Fix the path first. Only then decide whether two numbers belong in the same comparison.

The boundary problem connects to pipeline coverage and conversion distribution and forecast value added as a process audit, where the definition behind a number is kept visible.

References

1. Goodman, C. S. (1972). Measuring industrial markets: Uses and limitations of available data for market measurement. Industrial Marketing Management, 3, 279-293. https://doi.org/10.1016/0019-8501(72)90020-X
2. Marvasti, N. B., Huhtala, J. P., Yousefi, Z. R., Vaniala, I., Upreti, B., Malo, P., Kaski, S., & Tikkanen, H. (2021). Is this company a lead customer? Estimating stages of B2B buying journey. Industrial Marketing Management, 97, 126-133. https://doi.org/10.1016/j.indmarman.2021.06.003
3. Virtanen, T., Parvinen, P., & Rollins, M. (2015). Complexity of sales situation and sales lead performance: An empirical study in business-to-business company. Industrial Marketing Management, 45, 49-58. https://doi.org/10.1016/j.indmarman.2015.02.024