Decisions · Program mechanics and currency
How to set a point value
Fix the point's redemption value before you set any earn rate, and hold every catalogue item within a narrow band of that value. Decide the give-back you can afford as a share of gross margin, then derive the earn rate from it. A value set after the earn rate inherits whatever the earn rate implies.
Set the redemption value of a point before you set any earn rate, and derive the earn rate from the give-back your margin can fund. Programs that run the sequence in the other order publish a generous-sounding earn rate first, then discover at catalogue pricing what they have actually promised, and the correction they eventually make is called a devaluation.
## Members price the point off the redemption they want
The value that matters is set by the most desired redemption, never by the catalogue average. If the flight redemption returns 0.9 cents per point and the merchandise page returns 0.4, the members who want flights carry 0.9 in their heads and everyone else learns 0.4 the first time they check. There is no single value unless you build one.
## Hold every redemption inside a band
Pick a target value and keep each catalogue item within about twenty percent of it. An item that returns well above the band becomes the de facto value of the currency and an arbitrage magnet, drained by exactly the members who do the arithmetic. An item well below the band reads as a rip-off and contaminates trust in the rest of the catalogue. The band is the enforceable policy; a target alone is a wish.
## The earn rate is an output of two other numbers
Give-back as a share of spend equals earn rate multiplied by point value. Work it through: target a point at 1 cent and a return of 1 percent of spend, and the earn rate must be 1 point per dollar. At a 30 percent gross margin, that 1 percent give-back consumes 3.3 percent of gross margin. If finance will only fund reward cost at 0.6 percent of revenue, the choices are an earn rate of 0.6 points per dollar or a point value cut to 0.6 cents, and those two options are visible to members in completely different ways.
## Round numbers are part of the design
A 10 dollar reward at 1 cent per point costs 1,000 points. The same reward at 0.8 cents costs 1,250 points, which members can neither remember nor compute at a shelf. Legibility is worth real money. Pick values where the rewards people actually want land on round thresholds, because a currency nobody can price in their head defaults to being valued at zero.
## The spread between cost and value funds the generosity
A program only affords a high face value when the operator's cost per point sits below the member's value per point. Own-inventory redemptions create that spread: a hotel room with a marginal cost far below its cash price lets perceived value run high while fulfilment stays cheap. Cash and gift card redemptions close the spread entirely, since a point redeemed for cash is worth exactly what it costs to honour. If your catalogue is mostly cash-like, set the value low and honest, because nothing is available to fund a higher one.
No public benchmark of point values across programs exists in our data yet, so the target has to be built from your own margin arithmetic rather than from comparables. The failure mode is the one the opening rule prevents: an earn rate published before the value was modelled, then corrected two years later with a devaluation that members experience as a broken promise. The conditions that flip the recommendation are narrow. A pure cash-back currency has its value fixed by definition, and there the only real decision left is the earn rate itself.