A retention window is the period after a sale during which commission can still be reversed if the customer cancels, lapses or stops paying. Payment processors and subscription businesses commonly set it between 90 and 180 days; insurance carriers often run twelve months. Once the window closes, the money is the rep's permanently.
Ninety days or a hundred and eighty is the argument, and in most companies it gets settled by whoever is most senior in the room. There is a better way to pick the number. Plot when your customers actually churn. If two thirds of the losses land inside four months and the curve goes flat afterwards, a 120-day window recovers nearly everything a 365-day window would, and it buys back eight months of reps not knowing whether their earnings are real.
That second effect is the one companies underprice. A long window does not merely delay certainty, it changes how reps treat their own bank balance. Somebody who cannot say which of their last six payments is safe will either under-spend and resent the plan or spend anyway and get hurt, and neither outcome produces the behaviour the commission was meant to buy. A rep should be able to look at a statement on payday and name the date each line becomes untouchable.
Bands work better than a single edge. Reversing 100% in the first month, 50% through month three and nothing after tracks the company's real loss far more closely than a flat rule, and it removes the cliff that reps argue about, where a cancellation on day 179 costs everything and the same cancellation two days later costs nothing.
Sometimes the window is not yours to choose. An insurance carrier reclaims on its own schedule, a lender's investor agreement sets an early-payoff period that varies by product, and a processor's boarding bonus has terms written by the processor. Where the upstream window is fixed, the only real decision left is whether the rep's window matches it or sits inside it, and a rep's window shorter than the carrier's means the employer is knowingly eating the difference.
An ISO pays an $800 boarding bonus on every merchant an agent signs. Forty merchants board in a quarter. Three of them close inside 90 days and two more close between day 91 and day 180.
Whether 5% justifies three more months of uncertainty for every agent is a judgement, and it is the judgement the window is really making. Run this against your own attrition curve before copying anyone else's number.
A plan will state 180 days and leave the clock ambiguous. Signature date, boarding date, first transaction, first funding and go-live can sit weeks apart on the same deal, and two systems reading two different fields will disagree about whether a cancellation was inside the window. The rep's spreadsheet uses the date they remember closing it. Reconciling that argument after the fact is far harder than writing one extra clause naming the field the clock reads.
The long form lives in the guides: Commission clawbacks.
Commish pays retention window the way your plan describes it, shows the arithmetic on every line, and traces each payment back to the deal that earned it.