AntifrogACADEMYExplore Antifrog ↗
05 / THE LIBRARY

Stay with the question.

Longer reads for the problems already on your desk. Principles in theory; working methods here.

A stork and frog in the oak reading room
The practitioner collection / 24 reads

Read deeply.
Take something back.

Disputes, accounts, analysis and the practical work between them. Every article ends with a desk card and a question worth discussing.

Explore the continuing series

01
high-risk / practitioner / 9 min

Paying for players who never play.

CPA pays for an event. Revenue share pays for a relationship. The quality of an affiliate's traffic is only visible in what the cohort does after the commission has been paid.

Open article
02
accounts / practitioner / 9 min

Three layers of a takeover.

Session, device and behaviour each tell part of the story. A takeover is only visible where the three disagree with the customer's past.

Open article
03
payments / practitioner / 9 min

A thousand cards, one question.

How to read an enumeration burst as a defender, what it costs the merchant and the acquirer, and what to do in the first hour that you will not regret in the second.

Open article
04
high-risk / practitioner / 9 min

One welcome bonus, forty welcomes.

A bonus is a price the operator pays for a player. Multi-accounting is the same price paid forty times for one. The evidence is in the deposits, not the sign-ups.

Open article
05
operations / foundation / 7 min

Write the note for the person who reopens it.

A case-note template that separates what was seen from what was concluded, and survives a representment, an acquirer query or an audit.

Open article
06
disputes / advanced / 10 min

Two prior transactions and a matching footprint.

What Visa's Compelling Evidence 3.0 asks a merchant to hold, checked against the applicable pathway, and what matching records still do not prove.

Open article
07
high-risk / advanced / 10 min

The money has to leave somewhere.

A deposit tells you where money came from. A withdrawal to a different method tells you where it wants to go. The closed-loop rule, its exceptions and the source-of-funds question all live in the gap between the two.

Open article
08
disputes / practitioner / 9 min

A chargeback arrives in its own time.

Disputes land weeks after the sale they belong to. Measure by the month of sale, or the ratio you report will describe the wrong month.

Open article
09
disputes / practitioner / 9 min

When the cardholder is the customer.

A fraud dispute raised by the person who made the purchase is a different problem from theft. How to tell them apart without deciding too early.

Open article
10
analysis / advanced / 8 min

A shared attribute is a lead, a ring is a finding.

How to move from 'these accounts share something' to 'these accounts are one operation', and how to say which is which.

Open article
11
high-risk / advanced / 11 min

Reading a merchant before the first transaction.

Underwriting is a forecast with a review date. The PSP is deciding what one merchant will do to a portfolio ratio it does not control, and which terms make that forecast safe to be wrong about.

Open article
12
operations / practitioner / 8 min

The first fifteen minutes.

A runbook for the moment an authorisation burst appears: confirm, scope, act narrowly, write it down, and only then widen.

Open article
13
payments / foundation / 8 min

What a decline is trying to tell you.

A decline is a message from someone who saw something you did not. Learn who sent it, what they saw, and when a retry is a kindness and when it is a cost.

Open article
14
disputes / foundation / 8 min

Four families of dispute, four kinds of evidence.

Why the first digits of a reason code tell you what the issuer is asking, and what a reply would have to prove.

Open article
15
payments / practitioner / 10 min

The refund that was never a return.

Refund abuse hides inside a process that exists to keep customers happy. How to see it without punishing the customers who are simply unhappy.

Open article
16
analysis / foundation / 7 min

A rule is a claim. Write it like one.

A four-part template for fraud rules that a colleague, an acquirer or an auditor can read, test and switch off.

Open article
17
analysis / practitioner / 8 min

Compared with what, exactly.

Building a baseline by weekday and hour, reading a z-score or a ratio in plain words, and keeping the denominator in the room.

Open article
18
payments / practitioner / 10 min

The trial that keeps on giving.

Subscriptions turn one risk decision into a recurring one. How to read trial abuse, rebill disputes and the cohort that tells you which is which.

Open article
19
accounts / advanced / 10 min

A person who passes every check and does not exist.

A KYC pass answers narrow questions about documents and data. Whether one real person sits behind the account is a different question, and it is answered over time and in company.

Open article
20
payments / practitioner / 10 min

Who carries the risk when 3DS is skipped.

A working decision frame for SCA exemptions, liability shift and the customers in between.

Open article
21
analysis / practitioner / 8 min

The threshold is a price, not a score.

How to choose where a rule fires by pricing both kinds of error, with a worked matrix you can rebuild in a spreadsheet.

Open article
22
disputes / practitioner / 10 min

Two ratios, two denominators, one portfolio.

Visa's VAMP and Mastercard's ECP count different things over different months. A working guide to what moves each number.

Open article
23
payments / practitioner / 9 min

A velocity rule is a sentence with four parts.

Count what, per what, over how long, against what number. Most velocity rules that go wrong are missing one of the four, or were written for a business that no longer exists.

Open article
24
operations / practitioner / 9 min

What the acquirer asks, and why.

The five questions an acquirer puts to a PSP about a merchant near a monitoring threshold, and the records that answer them.

Open article

Original teaching material. Business names, scenarios and datasets are fictional unless an external source is explicitly identified. RSS feed