Pull your CRM data and the pattern is almost always the same: deals with a discount close at a higher rate than deals without one. In most companies that number lands somewhere between 10 and 25 percentage points. It looks like proof that discounting works.
It is not. And acting on it is one of the more expensive mistakes a sales organisation can make, because the cost is invisible — it shows up as margin you never see rather than deals you lose.
Why the number is misleading
Discounts are not handed out at random. A rep decides when to offer one, and that decision is informed by everything they know about the deal. Consider who actually receives a discount in a typical pipeline.
- —Deals that reached late stage — the ones already close to signing
- —Deals where the buyer is engaged enough to negotiate in the first place
- —Larger accounts, which have both more leverage and more intent
- —Deals a strong rep is running, because strong reps get further and negotiate harder
Every item on that list independently predicts closing. So when you compare discounted deals against undiscounted ones, you are not comparing discount versus no discount. You are comparing late-stage deals run by good reps with engaged buyers against everything else in the pipeline.
The discount did not cause those deals to close. Being the kind of deal that closes caused the discount.
What the honest question looks like
The question you actually need answered is a counterfactual: for a deal at this stage, this size, with this rep and this buyer, what happens to the probability of closing if we offer 10% off, versus if we do not?
That question cannot be answered by filtering a dashboard, because the comparison group you need does not exist in the data as a clean segment. It has to be constructed — you compare each discounted deal against the undiscounted deals that look like it in every respect that matters, and take the difference.
What you need to hold constant
The whole analysis turns on getting this list right. Anything that influences both the decision to discount and the likelihood of closing has to be accounted for.
- —Deal stage at the point the discount was offered — the single most important one
- —Deal size and account segment
- —Rep, or at minimum rep experience level
- —Lead source — inbound and outbound deals behave very differently
- —How long the deal has been open, and how engaged the buyer has been
- —Quarter or month, since end-of-quarter pressure moves both discounting and closing
If you only have time to control for one thing, control for stage. Discounts offered at proposal stage and discounts offered at first call are two completely different actions with the same name in your CRM.
The answer is usually a curve, not a yes or no
When teams run this properly, the typical finding is not that discounting works or does not work. It is that discounting works up to a point and then stops.
A common shape: below roughly 10% the discount adds a meaningful number of points to close rate. Between 10 and 15% the effect flattens. Above 15% the curve is flat — the extra margin buys nothing, because at that point you are discounting deals that were going to close anyway, or deals that were never going to close regardless.
That flat section is the finding worth having. It tells you where to set an approval threshold, and roughly what it is worth.
And the effect is not the same for everyone
Aggregate answers hide the useful detail. Discounting frequently behaves very differently across parts of the business — strong in competitive mid-market deals where the buyer is actively comparing vendors, close to useless in enterprise deals where procurement timelines matter more than price, and negative in inbound deals where the buyer had already decided.
A single company-wide discount policy averages over all of that. Knowing which segments respond turns the analysis into something a sales leader can act on: not "discount less", but "discount here, stop discounting there".
Where to start
You need one row per deal, with the outcome (won or lost), whether a discount was given and how much, and the factors listed above as they stood when the discount decision was made. Most CRM exports already contain this. The most common gap is stage-at-time-of-discount, which often has to be reconstructed from stage history.
That last detail matters more than it sounds. If you use final stage rather than stage at the time of the discount, every deal that closed will look late-stage, and the analysis will quietly tell you discounting is magnificent.