The Pipeline Nobody Dares Clean: Why Dead Deals Stay in the Forecast

Dead deals stay in a pipeline because removing one is a loss the remover feels immediately, while the benefit of removing it is a number that gets smaller and more honest for everybody else. A pipeline inflates quietly for the same reason a queue does: because every individual decision to leave something in place is defensible, and the accumulation is not.

The loss from removing a stale deal is immediate and personal. A number you put in a report goes down. A colleague who was counting on that value asks about it. You have to say out loud that a conversation you described as promising was not.

The loss from keeping it is deferred, distributed, and much larger. It is paid by whoever makes a decision on the basis of the inflated figure: the person who hires against a forecast, the person who sets a target, the person who concludes that this is a strong quarter and stops worrying. That person is not necessarily wrong. They are working from a number that has drifted, and they have no way to tell.

The asymmetry is the whole problem. One side of it is felt today by one person who is socially visible. The other side is felt months later by everybody, invisibly, and is almost never traced back to the deals that were never removed.

The constructed register below is a worked example. Every deal, value, and date is invented, and the arithmetic is shown underneath so it can be checked. It is here to make the size of the gap concrete, not to describe any real business.

A reminder to clean up the pipeline is not a rule. A rule is a set of decisions made in advance, and most of them are uncomfortable to write down. There are five, and skipping any one of them leaves the rule half-implemented.

The cost is not embarrassment. It is that decisions get made anyway, on a number everybody privately discounts, and those decisions are never revisited because the number itself was never taken seriously enough to challenge.

Hiring, borrowing, stock, and overtime are all decided against a pipeline figure. A team that knows the figure is inflated does not stop using it; it uses it and then compensates privately, by discounting it in their head by whatever amount they have learned to subtract. That subtraction is the real forecast. It lives in one person's head, it is not written down, it cannot be audited, and it differs between people, which means two people in the same business are working from two different pipelines.

The constructed example makes this concrete. Using 635,000, a business planning against 375,000 of real activity is over-planning by 260,000. Using 375,000, they are planning against a number they can trace to five specific deals with dated next actions. The first figure might produce a pleasant surprise. The second produces a plan somebody can execute and check.

A stale-deal rule run as a weekly instruction to the sales team is a reminder, and reminders decay. It is run as an output of a review that one named person is responsible for producing, it is a decision. The difference is that a decision leaves a record: this deal, marked stale on this date, by this person, under this rule, with the last recorded activity shown.

That record is what makes the rule survivable. It means a deal removed from the month's figure is not gone, it is documented, and if it turns out the person who marked it stale was wrong, there is a specific thing to correct rather than a vague suspicion that the pipeline is being kept tidy too aggressively.

Any rule that removes deals from a forecast will occasionally remove a deal that was going to close. This is not a defect to be engineered away; it is the price of a figure that means something. A rule that never misfires is a rule that is not doing any work, because the deals it fails to catch are exactly the deals that were never going to be caught by being careful.

Design for the misfire rather than pretending it will not happen. A holding area that a deal can be brought back from, with the re-entry recorded, converts an error into an inconvenience. A rule that permanently deletes live deals does not have a recovery path, and people who know that will route around it by not applying the rule.

NoxCRM keeps opportunities with owners, stages, dated activities, and stage history, which is enough to compute staleness and to show when a record last moved. It does not decide for you when a deal is dead, and there is no CRM-native forecasting model that would quietly write stale deals off for you. That decision is the one this article is about, and no tool is entitled to make it on a sales team's behalf.

There is also no marketing automation, no lead scoring engine, no email sequence builder, and no dialler in NoxCRM, so a held deal does not sit in a holding area quietly receiving automated touches from the platform. It sits still, visibly, until a person deals with it. Whether that is a feature or a limitation depends on how your business sells, and we would rather say so than imply automation that is not there.

In the dedicated NoxCRM deployment — scoped per business rather than sold as a separate product — the review rhythm, the holding area and the threshold are agreed during setup rather than picked from a settings page. The opportunities, owners and stages they operate on are part of every NoxOrigin plan. You can run this rule in a spreadsheet first, and several businesses should: the rule is a business decision, and a tool is only useful once the decision exists. If the review rhythm is the part you are still designing, /blog/crm-basics-small-business-whatsapp-spreadsheets covers the rest of the weekly loop, and /templates/lead-follow-up-sheet gives you a sheet to run the rule on before you automate anything.

We have not measured how much of a typical pipeline goes stale, and we are not going to give you a figure. Any such number would be invented, and an invented staleness rate would be indistinguishable from a real one once it had been copied into somebody else's deck. The 260,000 in the table above is arithmetic on invented deals, not a claim about how your pipeline behaves.

Your own figure is cheap to derive and worth more than ours: list the open deals, find the date of the last recorded activity on each, and count how many are older than the threshold you are considering. You will know within an hour whether the problem is real, and you will know it in a form you can act on.