Decisions · Earn and redemption
Should you round loyalty earnings up or down
Round up. For any purchase that earns a fractional point, award the next whole point. The extra liability is usually a fraction of a cent per transaction, while rounding down forfeits value the member believes they earned. Round down only where the rounded-up point exceeds the transaction margin.
The decision
The choice is not about arithmetic. It decides who keeps the fractional point that appears when a purchase does not divide evenly by the earn rate. A floor rule keeps that remainder with the programme; a ceiling rule gives it to the member. The mechanism is a transfer of value on every transaction, and the implementation effort is the same either way.
What it costs you
Rounding up increases your points liability and reduces recognised margin by the value of the extra fractions. Rounding down avoids that liability but creates member friction and support contacts. The arithmetic is small per transaction but compounds across volume.
An illustrative calculation makes the trade visible. Suppose your programme awards one point for each dollar spent. A member buys goods worth 9.50 dollars. The exact earning is 9.5 points. Rounding down awards 9 points; rounding up awards 10 points, an extra 1 point on that transaction. If each point carries a redemption liability of 0.01 dollars, that transaction costs the programme 0.01 dollars more under a round-up rule. Now scale that to 250,000 transactions in a month, every one of them earning a fractional point, with an average fractional remainder of half a point. Rounding down discards 125,000 points of earned value. Rounding up, by contrast, credits one full point for each transaction, so compared with rounding down it issues 250,000 extra points. At 0.01 dollars per point, the monthly cost is 2,500 dollars, or 30,000 dollars per year.
The technical effort is identical for either rule. Floor and ceiling are single operations inside the points engine, so this is not an engineering trade. The hidden cost of rounding down sits in member support and churn, not in code. When a buyer notices that a 9.50 dollar purchase earned 9 points under a published one point per dollar rate, they see broken arithmetic, not a rounding policy. That perception drives contacts and attrition that the points ledger never records.
How to decide
Run this test on your own transaction data. For a representative month, take every transaction that produces a non-integer number of points. Sum the discarded fractions under a round-down rule. Multiply that total by the cash value you assign to one point for liability purposes. Divide by the number of active members. If the cost per member per month is under one dollar, round up; the trust gain outweighs the liability. If it is over ten dollars, your earn rate is too generous and the rounding question is masking a larger unit-economics problem.
What would change our mind
We would switch to rounding down if the average transaction margin is lower than the value of the rounded-up point. In that situation, rounding up can turn a small, low-margin transaction into a loss after the loyalty cost is accrued. For example, if a transaction has a gross margin of 0.03 dollars and a half point rounds up to a full point worth 0.05 dollars, the programme loses 0.02 dollars on that sale before any other costs. Where that condition holds across a material share of transactions, rounding down is the only sustainable choice, and the remedy is to reduce the earn rate or raise prices, not to borrow margin from future redemptions.