Every point your program issues is a promise to give something away later: a free coffee, a discount, a flight. That makes a points balance a pot of money with your name on it, owed to your members. Loyalty fraud is what happens when someone takes value out of that pot who was never promised it.
This guide is written for the owner of the program, not for a security team. It covers the four ways points leave, why the controls belong in the build, and what to check before your program opens.
Points are money you owe
A bank account gets strong locks because everyone agrees it holds money. A loyalty account holds money too, in a different shape, but it often gets a weak lock: an email address, a reused password, and no warning when the balance moves. Member stories repeat the same pattern, paraphrased here. An account is opened with a password the member used elsewhere. The balance is drained in a single session. The member finds out weeks later, when they try to spend points that are no longer there.
Two facts make loyalty balances attractive to steal. Points can usually be turned into something resalable, such as gift cards, goods, or bookings. And members check loyalty accounts far less often than bank accounts, so theft can sit unnoticed for months. If points are money you owe, the system that holds them needs the care you would give to money.
The four ways points leave
Fraud against a loyalty program comes through four routes. Each route exploits a different weakness, so each needs a different control.
Someone signs in as the member
This is the route members fear, and the most common one. Lists of email addresses and passwords leak from breached websites every year. Software then tries those pairs against loyalty programs at scale, because a password that worked on a shopping site often works on a loyalty account too. Quiet accounts with large balances are the prize: the owner rarely signs in, so nothing looks wrong until the balance is spent, fast, on gift cards or bookings that can be resold. By the time the real member notices, the value has left and the trail is cold.
Fake members and joining bonuses
A joining bonus is money given to a stranger on the promise that they are a new customer. If joining takes thirty seconds and asks for nothing hard to fake, one person can open many accounts, collect the bonus from each, and spend the combined value. The same trick feeds referral rewards: the new member and the referring member are the same person with two email addresses. Nothing here requires hacking. The weakness is a joining flow that cannot tell one person from the same person twice.
Staff access used the wrong way
Every program needs staff tools: a way to fix a missed scan, credit points after a till failure, or correct a mistake. Those tools can also add points that no purchase earned, move a balance, or cancel an expiry. Used honestly, they keep members happy. Used badly, they mint money.
The weakness is usually permission, not character. If every login can adjust any balance by any amount, with no second person involved and no lasting record, the program has no way to tell a correction from a theft. Our guide to the types of loyalty apps makes the same point from the other side. The simple stamp card with a shared counter code is the easiest program to run and the easiest to abuse, because a copied code lets anyone add stamps.
Bending the rules from the outside
The last route breaks no locks at all. It uses the program exactly as written, in ways the writer did not intend. Points earned on a purchase that is then returned, but the points stay. A referral reward claimed between two accounts one person controls. A promotion meant for one use per member, claimed once per email address by someone with twenty.
These losses look small per event and add up quietly. They also corrupt the numbers the program runs on: balances owed that should not exist, and behaviour records that describe fraud as if it were loyalty. The defence is written rules with the awkward cases settled in advance. Our loyalty program design guide covers how those rules get decided.
Fraud controls belong in the build
Most fraud writing is aimed at programs already running on someone else's platform, and it ends at buying a detection tool. A detection tool watches transactions and flags odd ones. That has value, but it sits downstream of the real decisions: how a member proves who they are, how many accounts one person can hold, what staff can change, and what the system remembers.
In a custom program, those decisions are made once, at design and build time, and they are cheap there. Retrofitting them is not. A stronger sign-in step added after launch touches every member's account. Tighter staff tools mean retraining every team. And a record of past points changes cannot be rebuilt after the fact: what was never kept is gone. This is why our loyalty app development guide puts rules and ownership before screens. Fraud control is part of that same order of work.
The controls, in plain terms
Six controls cover the four routes. None of them requires a security department to understand or to run.
- One member, one record. The system should spot likely duplicates at joining: the same phone number, payment card, or device behind several accounts. Bonuses and referral rewards are paid once per person, not once per email address.
- A stronger check at risky moments. Everyday earning stays effortless. A large redemption, a change of email or phone number, a points transfer, or a sign-in from an unusual place asks for a code sent to the member before it goes through.
- Alerts the member actually sees. A redemption above a set size, a change to the email or phone number on the account, and a password reset each send a message at once. The member becomes the fastest fraud detector the program has, because nobody knows their balance like they do.
- Limits and short holds on unusual spending. A balance built over two years does not normally leave in one minute. A brief hold and a confirmation step on redemptions far above the member's usual pattern stops the fast drain that account theft depends on, while ordinary spending is untouched.
- Staff tools with edges. Each staff login can do only what that person's job needs. Adjustments above a set size need a second person's approval, and every adjustment asks for a reason from a short list, so corrections have a shape that theft does not.
- A record that cannot be quietly edited. Each points change, who made it, when, and why, goes into a history that staff tools can read but not rewrite. When a member disputes a balance, the program answers from the record instead of from memory.
Read the list again as a founder test. Each control can be explained to a co-founder in one sentence, and each maps to a route in the section above. If a proposed build cannot say how it handles all six, the fraud conversation has not happened yet.
Check the locks before you open the doors
Controls on paper are not controls in the system. Before launch, run the attacks yourself, or watch your builder run them:
- Join twice with different email addresses and try to collect two joining bonuses.
- Reset a password, then immediately try to spend the whole balance.
- Ask a junior staff login to add a large points adjustment without approval.
- Buy, earn, return the item, and check whether the points left with the refund.
- Refer the program to a second account you control and see if both sides get paid.
Each test takes minutes, and each failure found here costs a settings change. The same failure found after launch costs refunds of value, support time, and members who stop trusting the balance on their card. Treat these five tests as the security page of the wider launch work in our loyalty program launch checklist.
What prevention costs
Prevention inside a custom build is mostly decisions, priced in the scoping work rather than sold as a separate product: how sign-in works, what staff can change, what needs approval, what gets recorded. The expensive version is the one bought after a loss, when sign-in, staff tools, and the points record all have to be reopened while the program is live.
The loss side is easier to picture than to price. Stolen points are value handed out for nothing. Restoring a member's balance pays the same value twice. Support staff spend hours on disputes the system cannot settle because it kept no record. Our loyalty program cost guide treats fraud prevention as a real cost line of running a program, alongside connections and staff training, which is where it belongs in a budget.
Bottom line
Loyalty fraud is not a hacker problem with a software answer. It is a money problem with a design answer. Points leave through four routes: a member's account, fake joins, staff tools, and rules bent from outside. One record per member, stronger checks at risky moments, alerts, limits, edged staff tools, and a record that cannot be edited quietly close all four routes before the first member joins.
Talk to the studio
LoyaltyPass is a studio that builds custom loyalty software for businesses that have outgrown off-the-shelf tools. Fraud controls are part of the design conversation from the first call, not an extra sold later. Projects start at $9,500. Book a scoping call, bring your program rules, and say where you worry the points could leak. The proposal will show how the build closes those routes.


