How to Run a Pipeline Scrub

What is a pipeline scrub?

A pipeline scrub is a scheduled pass that tests every open opportunity against age and activity rules and removes the ones that fail. It ends with records changed, not with a meeting summary.

Pipeline accumulates dead weight by default. Nobody is rewarded for closing a deal as lost, reps prefer to leave hope in the system, and stage-based reporting makes a stale opportunity look identical to a live one. The result is a pipeline number that flatters everyone and predicts nothing.

The share varies by company, but more than 10 percent of a typical pipeline is stale, untouched for twelve months. That volume is sitting inside every coverage ratio you report.

How is a scrub different from a pipeline review?

A review works the deals you intend to win. A scrub decides which deals should not be counted at all.

The two need separate sessions because they compete for the same hour and the review always wins. Managers will spend sixty minutes on the deals they can close this quarter and zero on the ones that should have been closed as lost eighteen months ago.

What test do you apply to each deal?

Test against meaningful activity, not against logged activity. Meaningful movement on an opportunity is a change in stage, close date, or amount.

That distinction is what makes the scrub work. Logged calls and emails are easy to produce without a deal advancing, and any activity-based hygiene rule gets satisfied within a week of being announced. Stage, date, and amount are harder to fake because changing them has consequences in the forecast.

Apply a tiered age test:

The twelve-month line is a reasonable default because it sits well beyond the point where most opportunities resolve. When opportunities are grouped by pattern and given a predicted time-to-close curve, those curves run from one week to eighty weeks, with most of the closing expectation landing before week twelve. Very few groups carry meaningful expectation past week fifty-two. A deal that has shown no stage, date, or amount movement in a year is outside the window where its cohort closes.

What do you do with deals that fail the test?

Close them as lost with a reason code, and keep the record.

Deleting stale opportunities destroys the most useful training data you have. A closed-lost record with a reason tells you which sources, segments, and deal shapes never convert, and that feeds directly into scoring and sales forecasting. A deleted record tells you nothing forever.

Use a short, mandatory reason list so the data stays usable:

- No decision, buyer stalled - Lost to competitor, named - Lost to internal build or status quo - Budget removed or reallocated - Unqualified, should not have entered pipeline - Champion left the account

The last two categories are the most valuable. A high share of unqualified entries points at a qualification problem upstream. A high share of champion departures points at single-threaded selling.

Give managers a sign-off step rather than a veto. Every deal the rules mark for closure goes to the owning manager with a 48-hour window to document why it should stay, and silence closes it. Managers who can block closures without evidence will preserve the same stale pipeline the scrub exists to remove, usually because coverage against quota is what they are measured on. Requiring a written reason costs a minute per deal and cuts the save rate to the deals that genuinely deserve it.

How does the scrub change your reported numbers?

Expect coverage to fall, and expect that to be an improvement.

A scrub that removes 10 percent or more of open pipeline will visibly reduce your coverage ratio. Leadership sometimes treats that as a step backward, which is exactly the confusion the scrub exists to fix. Coverage built on stale opportunities was never protection.

Two things get better immediately. Forecast inputs improve, because the model is no longer weighting deals that will never close. And the composition of the quarter becomes readable, because the pipeline you can see is now pipeline that is actually in play. Across ORM's customers, coverage runs from 1.4x to 5x with most companies near 3.5x, and a scrubbed 3x is worth more than an unscrubbed 4x. This is the practical reason the 3x pipeline coverage rule fails as a standalone measure of health.

How do you keep the pipeline clean between scrubs?

Automate the flagging so the quarterly pass has less to find.

Set three rules to run continuously:

- Any deal with no stage, close date, or amount change in 30 days appears on the manager's weekly exception list. - Any deal whose close date moves twice drops out of the committed forecast until a manager reinstates it. A close date change is the strongest deal slippage signal you have, and a deal that slips across a quarter boundary is less likely to close even when it stays in commit. - Any deal with a close date in the past auto-flags on the day it expires rather than at the end of the month.

Then hold the standard where it belongs. The scrub is a process, and consistency in how deals are entered and updated matters more than perfect data quality. A pipeline that is consistently optimistic can be modeled accurately. A pipeline where hygiene depends on which manager owns the segment cannot.