Best practices: preventing referral fraud without blocking real customers
Fraud control should protect your reward budget, not stop real customers. Leave Bloop’s automatic signals on, review each flagged order on its evidence, and set your offer so cheating does not pay.
Most abuse is self-referral: one person posing as both referrer and buyer to claim both sides of the deal. Blocking everything stops that, but it also frustrates your best referrers. The goal is to catch the few abusers without taxing the many honest customers.
Let detection flag, then decide on evidence
Section titled “Let detection flag, then decide on evidence”Bloop watches for the patterns abuse leaves behind and holds only the orders that deserve a second look. A flag is a prompt to check, not a verdict. A popular referrer on a shared office network can look like collusion.
Use this table to decide what to do with a flagged referrer:
| What you see | What to do |
|---|---|
| Exact email match, alias on another domain, or matching address | Reject. Ban if it repeats |
| Matching last name only | Look closer. Family referrals are usually real |
| Shared IP only, with unrelated real emails | Approve. It’s a household or an office |
| Shared IP and a self-referral signal | Reject |
| High volume with varied IPs, emails and names | Approve. This is a popular referrer |
| High volume with repeated IPs or alias emails | Reject and ban |
The Same IP as referrer and High referral volume signals use these default thresholds: 5 or more referred purchases from one IP in 30 days, and 7 or more referred purchases in 7 days. The self-referral signals are the strong ones.
Actions on flagged referrals are logged in Fraud history, so you can check what was approved or rejected.
Check the signals before you tune them
Section titled “Check the signals before you tune them”Open Settings → Referral → Fraud detection, or click Detection rules in the Fraud center. It has two groups:
- Fraud detection flags a referral for review. Every signal is on by default.
- Fraud Prevention blocks the referee’s discount outright when the referrer and referee share an IP address or a browser.
A referee whose email matches the referrer’s, including Gmail variations of the same address, is rejected straight away rather than held, as long as the self-referral email signal is on.
Remove the incentive to cheat at the source
Section titled “Remove the incentive to cheat at the source”Bloop already closes the most common loopholes. Existing customers get no referee reward, and by default a referral is revoked when its order is cancelled or 90% or more refunded. You change this in Settings → Referral → Rewards and orders.
Three offer settings reduce abuse further:
| Setting | What it stops | Trade-off |
|---|---|---|
| Minimum purchase | Makes self-referral cost the abuser real money | Small honest orders may not qualify |
| Reward review period | Rewards paid before the signals have surfaced | Referrers wait longer for their reward |
| Tighter audience | Abuse from a wide pool of referrers | Fewer people can refer |
The review period holds each reward for 6 hours by default. Lengthen it for high-value cash rewards, or if you have seen abuse. It is a store-wide setting in Settings → Referral → Rewards and orders. Suspicious orders are held separately either way.
Many merchants open the audience wide at launch to gather data, then tighten it only if abuse appears.
Scale your review effort to what one abuse would cost you. A high-value cash reward on an open audience needs closer review than a small store credit limited to past customers.
Set it up in Bloop
Section titled “Set it up in Bloop”- Review and resolve flagged orders in the Fraud center.
- Add minimum requirements in rewards, and set the review period in reward timing.
- Limit who can refer in target audience.