Decisions · Risk and fraud

Should you design loyalty for account takeover resistance

Yes, if your loyalty programme holds transferable points or points redeemable for cash equivalents. Build takeover resistance into balance changes, redemption velocity and recovery flows. For non-transferable, expiring, low-value points, ordinary authentication can be enough.

The decision

The decision is whether account takeover resistance should be a design requirement for the loyalty ledger itself, not a security patch around login. A loyalty account is a slow settlement account: a member earns points over months, then converts them in a single redemption. Attackers target the gap between credential theft and redemption. The mechanism is control at three points: balance mutation, redemption velocity and recovery of a taken account. If you design those points to be hostile to takeover, you reduce the expected loss from stolen credentials.

What it costs you

Designing for takeover resistance costs engineering time and operating friction before it saves liability. The real cost is not software alone; it is the added step a legitimate member takes during redemption and the support queue that follows. An illustrative calculation shows why the trade is often worth it. Take a programme with 1 million members, an average balance of 4,000 points and a redemption value of one cent per point. The stored member value is 40 million dollars. If takeover resistance reduces successful account takeovers from 0.2 percent of accounts per year to 0.05 percent, the expected annual loss falls from 80,000 dollars to 20,000 dollars. That is a saving of 60,000 dollars a year. If the design work costs three engineer months at 12,000 dollars per month, the cost is 36,000 dollars, leaving 24,000 dollars of annual benefit before extra support is counted. That margin pays for the friction. If your average balance is 400 points, the saving falls to 6,000 dollars and hardening may not clear the cost.

How to decide

To decide for your programme, run the stored value test. Multiply active member count by average point balance by one cent. If the result is below 250,000 dollars, account takeover resistance is not a design requirement; it is a compliance checkbox. If the result is above 2.5 million dollars, build the three controls into the ledger. Between those figures, apply the transferability test: can a member move points to another account or convert them to a cash equivalent within 24 hours? If yes, harden. If points are non-transferable, have a 90 day expiry and cannot become cash equivalents, basic authentication plus anomaly monitoring is enough. The test should be applied to the programme's own redemption rules, not to a generic threat model.

What would change our mind

We would reverse this recommendation if your loyalty points are non-transferable, expire quickly and cannot be converted into cash or near cash. In that configuration, a stolen account has little liquidation value and the engineering cost of ledger-level resistance exceeds the liability reduction. We would also change the recommendation if a regulator required issuers to reimburse members for all loyalty point theft regardless of programme controls. That would shift the loss from the operator's margin to the issuer's balance sheet, making account takeover resistance a regulatory cost rather than an operating decision. Without one of those conditions, we hold the position.

Related