Give a rep enough discount authority to close the ordinary deal without waiting on anyone, and route everything past that to a person who owns the margin. The specific number comes from your own data, not from a benchmark, because it depends on your list pricing, your segment mix, and how much room the pricing model already carries.
Pull your own distribution first
Export every closed-won deal from the last four quarters with its discount off list. Plot the distribution. You will usually find a dense middle band where most wins cluster and a long tail of exceptions.
Set the rep band to cover the dense middle. That way routine business quotes instantly and only genuine exceptions reach an approver. Set it below the middle and you have built a queue for your own standard pricing, which is the fastest way to make approvals the bottleneck. Set it well above the middle and the band grants authority nobody needed, so the average selling price drifts down to whatever the ceiling allows.
Segment the analysis before you set the number. Enterprise and SMB rarely share a distribution, and a single company-wide band will be wrong for both.
Every point past the band buys something
Authority past the standing band should be conditional, not merely deeper. A longer term, an annual prepayment, a narrower scope, a reference commitment, or a case study. When a rep asks for an extra ten points and the answer is yes with no trade, the discount stops being a negotiating instrument and becomes the price.
Write the trades into the approval rules so the exchange is automatic rather than argued deal by deal.
Watch the market, not only the policy
Discount bands go stale when conditions shift. ORM saw price pressure and buyer indecision across its customer base through the first half of 2026, and pressure of that kind shows up as deeper discount requests before it appears anywhere in bookings.
Re-pull the distribution every two quarters. If requests are drifting deeper while win rates hold flat, reps are buying deals that would have closed anyway. If requests are drifting deeper and win rates are falling with them, the problem is competitive and a wider band will not fix it.
The forecasting consequence
Discount authority determines the gap between pipeline value and closed value. ORM has seen that gap run wide in customer data, with a pipeline averaging $80,000 per deal while closed-won deals averaged $40,000. A forecast built on the pipeline number is wrong before any deal slips.
If you want forecast accuracy that holds up, weight pipeline by the price your team actually realizes at each band rather than by the number entered at deal creation. See sales forecasting best practices for how that adjustment fits the wider model.
Frequently Asked Questions
How much discount authority should a rep have?
Enough to close the ordinary deal without an approval, and no more. Pull the discount distribution on your last four quarters of closed-won deals and set the band to cover the range where most wins already land. That number is specific to your pricing and segment, so borrowed benchmarks are worse than your own data.
Should reps have any discount authority at all?
Yes, for speed. Zero authority sends every deal into an approval queue, which slows quoting and trains reps to pad the ask so the approved number lands where they wanted it. A narrow standing band removes the queue for routine business.
Should discount authority vary by rep?
By segment and deal size rather than by tenure. An enterprise deal and an SMB renewal need different bands because the dollar consequence differs by orders of magnitude. Rewarding tenure with deeper authority mostly rewards whoever is best at asking.
How do you stop discount creep at quarter-end?
Require a trade for every point past the standing band, log every exception with the approver and the reason, and re-approve rather than inherit when a deal slips into the next quarter. Creep happens because a one-time exception becomes the reference price for the next deal.
Put these metrics to work
ORM builds custom revenue forecast models that turn concepts like how much discount should a sales rep be allowed to give? into prescriptive action for your team.
Schedule a Demo