I rewatched Mean Girls last week to switch my brain off and instead spent 90 minutes thinking about promotion suppression. Yes, this is a personal problem, yes, stay with me anyway.
One scene hit me: when Karen is forecasting the weather and landing on a 30% chance that it’s already raining, which, said with love for this industry and for myself as a member of it, is the rigour behind most promo planning. There’s a 30% chance the discount was already unnecessary and there’s a 100% chance nobody checked.
On the other hand, Regina’s operation was two lists ran with far more care than most brands give their pricing.
The first was the Plastics: 3 people, not a lookalike audience, and definitely not “women 25 to 40 who have opened an email.” Chosen on specific criteria, constantly reviewed, and cancellable on any day. That’s targeting. The second was the Burn Book. A detailed record of exactly who was excluded and why. That’s suppression or your do-not-fire list.
If you want the full take of the second one, negative eligibility, stacking, budget options, the margin-aware bit, I wrote it up properly over here. This piece is about the part that comes after which nobody talks about.
The best incentive is usually the one that doesn’t fire
So what happens when a suppression rule works? Well…nothing.
No redemption event or attributed conversion. Somebody was about to receive a discount they didn’t need and then they didn’t, and the entire evidence is an empty space where a coupon would have been.
That’s why promo suppression lists often don’t get built, by the way. Targeting comes with a slide and suppression comes with a shrug.
So how do you justify a decision with no evident paper trail?
Log the block
Most promotion engines throw away the most valuable data they produce. When suppression fires, something perfectly trackable happens: this customer with this cart, this rule and this discount amount that didn’t leave your business. If your incentive engine writes that down as a suppression event with the rule name and the value attached, you can add it up. If it doesn’t, the money you saved is the same money that was never in play.
Go check whether your setup logs suppression events at all. Most don’t. It’s the highest-value hour of work in this whole exercise and it’s boring enough, so I get the procrastination reflex on this one. But (!) if you flex these muscles, what you get is a running total of blocked discount value, split by reason. That’s the first step in writing your very own Burn Book.
Then hold some out
Blocked discount value on its own is a lie you’re telling yourself. Or as Gen Z says it, a delulu.
Some of those suppressed customers would only have bought with the discount and by refusing them you lost the order. You can’t know how many by staring at the total or suppressing everybody.
Take the population your rule applies to and split it. Most of them get the new rule and the discount gets blocked. A small slice, say 10%, keeps getting the discount exactly as before. That slice is your control group or your "what could have been" pile.
Then compare. And don’t compare on redemptions or on conversion rate!
The golden rule: compare contribution margin per session.
A good suppression rule will often make conversion look slightly worse while making the business meaningfully better, and if conversion rate is the number on the wall, someone can kill your rule with a graph that’s technically true and completely misleading.
If the control group (folks who got the discount) converts a bit higher while your suppressed group (folks who didn’t get the discount) carries more margin on every order, the promo suppression is working. If conversion in the suppressed group falls off a cliff, that discount was doing real work, put it back and figure out why customers only buy on sale.
Wait longer than feels comfortable
Someone you suppressed today might buy next week at full price, which makes the rule look better than your first report said. Or they might never come back, which makes it look worse, and that shows up in repeat purchase rate at 30 and 60 days rather than in this week’s numbers.
So measure the cohort, not the session: repeat rate, return rate, revenue per customer over a window that matches how people actually buy from you. If you sell mattresses, a 2-week read tells you nothing.
Give the absence a name
The reason promo suppression goes unappreciated is that it has no hot artefact, so make one.
Start with one line in the monthly review: blocked discount value, net of holdout-estimated conversion loss, split by rule. Two numbers and a breakdown. It turns the least legible work in your promo strategy into something a finance person recognises on sight, and finance people are excellent allies to have!
It also gives you something better than a raw number. A per-rule read on which of your discounts were ever incremental in the first place, which is the question sitting underneath all promotional spend.
The bus
Obviously I have to deal with the elephant (bus) in the room.
Regina’s system worked right up until the Burn Book got photocopied and scattered down the hallway and then the school was on fire and she almost got her neck twisted.
The lesson? The control layer was a physical object owned by one person, undocumented, running on her judgement and her vibe. Nobody could audit it or defend it in a meeting. When it burned, it burned all at once.
Plenty of promotions run exactly like this. The rules live in one person’s head, or in a spreadsheet, or in the tribal knowledge of whoever knows not to stack the loyalty thing with the shipping thing. That person takes August off and August is when things come crashing down.
Rules in a person are a Burn Book. Rules in the promo engine, logged and held out and measured, are architecture.
Be more like Regina about who gets in. Just keep receipts, and don’t be the only copy.
Anyway, if you’ve got a suppression rule that saved your quarter, or a ghost promotion that took a week to find, that’s the sort of thing we trade in marginmaxx. WhatsApp, invite-only group.