Deferred discounting: never give away today what buys you tomorrow

There are two places you can put a discount. You can take it off the order in front of you, or you can issue it forward against an order that has not happened yet. Almost every pizza shop does the first. The second is materially better on nearly every dimension that matters, and it is the single change we push hardest with new locations.

What deferred discounting is

A deferred discount is value issued at the moment of a completed order, redeemable only on a subsequent order, within a defined window. The customer pays full price today and leaves holding something. Nothing comes off today’s ticket.

The classic shape is "$10 off your next order, valid for 21 days." The important structural features are that the value is conditional on returning, time-boxed, and costed only on redemption.

An immediate discount buys an order you were probably getting anyway. A deferred discount buys the order after it.

Why it outperforms an immediate discount

Five reasons, in rough order of financial significance.

1. Today’s margin is intact. A 20% immediate discount on a $30 order costs $6 of pure margin on an order the customer had already decided to place. The deferred version costs nothing today.
2. You are buying an incremental occasion, not subsidising an existing one. This is the whole argument. Discounting the order someone was already placing is not marketing, it is a price cut. Discounting the next order changes behaviour.
3. Cost is incurred only on success. Unredeemed offers cost nothing. Redemption on a well-built deferred programme typically runs 15–35%, which means your effective cost per campaign is a fraction of the face value, and every dollar of it is attached to a return visit.
4. It self-selects. The people who redeem are the people who came back. You are, by construction, spending money exclusively on retained customers rather than on everyone who walked in.
5. It creates a reason to return that is not price. Holding something unused is an open loop. The customer returns to close it, and the return trip becomes a habit-building event rather than a bargain-hunting event.

The window is the most important parameter

Get the expiry window wrong and the whole mechanism fails. Too short and it reads as a trick; too long and it stops driving urgency and starts sitting in a drawer.

Set the window from the customer’s own order rhythm, not from a house rule. The right window is roughly 1.5 to 2 times their median inter-order interval: long enough that returning is natural, short enough that it pulls the next visit forward rather than simply subsidising it.

Where it earns the most: the first order

The single best place to deploy a deferred discount is on a customer’s first ever order. First-order customers have the steepest churn curve in the entire file, the majority never return, and the second order roughly doubles the probability of a third.

Issuing forward value at first order costs nothing on an order you have already won, and directly attacks the highest-leverage drop-off in the whole customer lifecycle. If you do only one thing from this chapter, do this one.

Structuring the value

Three structures, each with a different job.

• Flat value ("$8 off your next order"): simple, easy to communicate, and the customer knows exactly what they hold. Set a sensible minimum spend so it cannot be used to buy a single side at a loss.
• Item-based ("free garlic bread with your next order"): lower food cost than the perceived value, higher perceived generosity per dollar, and it drives an attach rather than a discount. Usually the best margin per unit of goodwill.
• Escalating ladder ("$5 next order, $8 the one after, $12 the one after that"): the strongest frequency builder and the most complex to run. Reserve it for a defined reactivation or habit-building sequence rather than as a standing offer.

The mistakes that break it

Deferred discounting fails in predictable ways.

• Issuing to everyone, every time. Then it is not an offer, it is your price. Regulars in particular should not receive it. They were returning anyway, and you have simply cut your best customers’ prices.
• No expiry. Removes urgency, creates an unbounded liability, and makes forecasting impossible.
• Stacking with everything else. Define what it can and cannot combine with before launch, not after the first bad ticket.
• Not tracking issuance and redemption separately. These are two different events with two different dates and two different financial meanings. Reporting that only sees redemption cannot tell you the offer’s true cost or its true lift.
• No holdout. Without a control group you cannot distinguish customers who returned because of the offer from customers who were returning anyway, which is exactly the distinction the whole strategy rests on.

Accounting for it properly

A deferred discount is a contingent liability at issuance and a discount at redemption. Track both. The two numbers you want on the weekly scorecard are outstanding face value (issued, unexpired, unredeemed) and realised cost (redeemed face value in the period).

The gap between them is not profit. It is unrealised liability plus breakage. Assume a redemption rate based on your own history rather than an industry figure, and revisit it quarterly. Redemption rates move when the window moves, when the value moves, and when the channel moves.

Questions