Fraud and chargeback prevention

Finding the fraud an online business already pays for, read from its own records, and tightening the rules without losing the customers it wants.

Write a brief Book a meeting

Kinds of loss

Most engagements cover a few of these, chosen once you have said where the losses are turning up.

Chargeback patterns: Reading your disputes to separate genuine fraud from buyers who could have been refunded before a chargeback was ever filed.
Friendly fraud: Customers who received the goods and then claimed they did not, checked against the delivery evidence behind each dispute.
Promotion abuse: First-order discounts and referral bonuses claimed again and again by the same people under different accounts.
Return abuse: Serial returners and wardrobing, read from return rates against the accounts and payment methods behind them.
Affiliate and click fraud: Affiliates paid for sales that were never incremental, and ad spend drained by traffic that was never going to buy.
Rule changes: Adjusting the checks that stop fraud so they catch more of it while turning away fewer of the customers you want to keep.
Measuring the cost: Putting a figure on each kind of abuse, so the fixes can be ordered by what each one costs.

The fraud a business already pays for

A business that sells online is usually paying for fraud it has not counted. The money leaves in small amounts, such as a chargeback on an order that was delivered or a first-order discount claimed twice by the same person. None of it is large enough on its own to force a decision, so it sits in the accounts as a cost of doing business and nobody adds it up.

The data to find it is already in the company's own systems. Delivery records read against account histories carry the pattern once they are joined, and so do the reasons buyers give for their disputes. A customer who files the same dispute reason on a third order shows up, as does an account that keeps reappearing behind a swapped letter in an email address. The work here starts from those records and reports what each kind of abuse is costing, with how firm each figure is.

The usual fix is to tighten the rules until the fraud stops, and that turns away paying customers along with the fraudsters. A rule that blocks every order from a new address also blocks a real customer who just moved. The useful question is which rule changes remove the most loss while rejecting the fewest good buyers, and that can be read from how your existing rules have performed against orders whose outcome you already know.

How to start

The brief

Say where the losses are turning up and what you cannot afford to block. Mention what data you hold on orders and disputes, even if it is messy.

Confidentiality

A confidentiality agreement can be signed before you send anything about your customers or payments.

Who does the work

Amos Weiskopf has run customer acquisition and conversion work at a Fortune 500 company, and he builds the data checks he recommends.

Questions buyers ask

What is fraud and chargeback prevention?

It is reading an online business's own data to find the fraud it is already paying for, then changing the rules that catch it without blocking paying customers.

How much does fraud prevention consulting cost?

That depends mostly on how much data has to be read and the state it is in. The price is agreed after the brief.

Will tighter rules cost me real customers?

That is the risk the work is built to avoid. The rule changes are read from orders whose outcome you already know, so they can be chosen to remove loss while rejecting as few good buyers as the data allows.

Do you work only from our own data?

The reading starts from the records you hold about orders and disputes. Public or lawfully obtained material can be used alongside them where it bears on a finding, and nothing unlawful is done to gather it.

Does our customer data stay under our control?

Yes. The analysis runs where you choose, and customer records are not passed to anyone else.

Can you tell friendly fraud from real fraud?

Often, though not always cleanly. Delivery records read against account histories separate buyers who received the goods and claimed otherwise from genuine fraud, and the report says where the data cannot tell the two apart.

Engagements by written brief.

Work begins from a documented problem statement, not a sales call.

Write a brief Or book a meeting