Behavioural analysis against each customer’s own history

DetectX® Behavioural Analysis learns how each customer normally pays. It compares their transactions with their own history and with customers who behave like them, and raises an alert when the deviation is significant. The thresholds are set by the system and adjust as the data changes.

Request a demoTalk to us
  1. 1Observe
  2. 2Detect
  3. 3Review
  4. 4Platform tour
DetectX - One Platform. Total Compliance (FINAL CAPTIONS)

One Platform. Total Compliance.

The DetectX® Approach

Unusual, measured against what is usual.

Behavioural analysis is the comparison of a customer’s current payment behaviour with what is normal for them and for customers like them, so that unusual activity stands out even when it breaks no fixed rule. In DetectX®, customers are grouped into clusters from their transaction history, statistics are collected over months, and an alert is raised when the deviation is significant.

Unusual for whom? Behavioural Analysis builds on the risk profiles and transaction history you already hold. It compares each customer with customers like them and with their own past, and raises an alert when the difference is significant.

RequiredClient Risk ScoringRisk profiles, and the data for the initial clustering model.
RequiredTransaction MonitoringThe transaction history, and the daily alert run.
OptionalYour Own CategoriesCustomer and transaction categories and risk assessments.
Behavioural AnalysisCompared TwiceEach customer’s activity is compared with their cluster and with their own history.
Customer clustersStatistics over monthsDynamic thresholds
How customers are clustered
Significant deviationAn Alert Is RaisedMonthly and daily, worked by your analyst in AlertViewer.
Also producedCustomer ThresholdsUsed in your monitoring and filtering rules.Thresholds applied to each payment

Your categories sharpen the clusters but are not required.

Key Benefits & Impact

Unusual only means something once you know what is usual.

A fixed rule knows nothing about the customer it is checking. DetectX® Behavioural Analysis learns what normal looks like for each customer and for customers like them, and alerts when behaviour departs from both.

Compared Twice

Against customers like them, and against themselves.

Each customer’s activity is measured against the cluster they belong to and against their own past months. A deviation that shows on both comparisons is one worth opening.

How customers are clustered
Grouped by Behaviour

A change of level is a reason to look.

The system places each customer on an activity level from 0 to 3, from how often and how much they transact. A customer who jumps to a new level with rising risk stands out, without anyone writing a rule for it.

How profiles are built
A Reason for Every Alert

Your analyst sees why, not just a score.

Alerts are raised when behaviour changes level with rising risk, when it departs from expected patterns, or when a customer belongs to a high-risk cluster. Each alert carries the reason that raised it.

How alerts are worked and reviewed
Thresholds That Move

Limits that keep working as your base changes.

Thresholds are determined by the system instead of fixed in advance, and adjust as your customer base or the wider economy changes. Nobody has to re-tune a limit that the whole base has quietly outgrown.

Spotting patterns over time
Shaped by Your Own Data

Clusters that fit your bank, not a generic model.

The clusters are built from your own customers’ anonymised history. Your customer and transaction categories and risk assessments can be added to make them sharper.

How client data is prepared
Built on What You Already Run

Added to the monitoring you already run.

The clustering model starts in Client Risk Scoring, and daily alerts come through Transaction Monitoring. The customer thresholds it produces can also check each payment through Transaction Filtering.

Thresholds applied to each payment

Why DetectX®

Most rules are written once. DetectX® learns what normal is.

Request a demoTalk to our team
  • Measured against the customerOne limit for everyone, whatever their patternEach customer compared with their own past months and with customers like themHow customers are clustered
  • A change that stays under the limitA fivefold jump below the limit goes unseenA new activity level with rising risk raises an alertKey benefits
  • Slow drift caughtA customer drifting from their own pattern stays quietA significant deviation from their own history raises an alertPattern Recognition
  • A reason for every alertA limit breached, nothing moreEach alert carries its reason: change of level, broken pattern or high-risk clusterHow alerts are worked
  • Thresholds nobody has to re-tuneOrdinary customers alert once the base grows past the limitThresholds set by the system adjust as your base or the economy changesThresholds that move
  • One platform, not another systemA separate model with its own dataBuilds on Client Risk Scoring and raises its daily alerts through Transaction MonitoringClient Risk Scoring

In production

Variables

Five measures, one decision on whether to alert.

5 Variables in Every Decision

Activity level, transaction risk, Risk, Risk 2 and cluster risk, calculated during clustering and read together.

Profiling & Scoring

Your Next Step

Three ways in, depending on where you are.

Each one is a conversation with the people who build DetectX®, with no obligation.

Start Here

Request a Demo

Watch Behavioural Analysis work through real alerts with a product specialist.

  • Alerts explained, field by field
  • Thresholds tuned in front of you
  • Your questions answered on the call
30 min · Video callRequest

Talk to Our Team

Bring your data, volumes and core banking system. We map where DetectX® fits.

  • How DetectX® connects to your systems
  • Which data you would use
  • A realistic plan for going live
45 min · Video callBook a call

Ask for a Reference

Speak with a compliance team already running DetectX®.

  • A peer with a set-up close to yours
  • Their experience, in their words
  • Arranged with their agreement
By arrangement · CallAsk us

How Behavioural Analysis Works

Try it yourself: four steps, from a customer’s cluster to a decision on its alert.

Unusual only means something once you know what is usual. DetectX® learns how each customer and customers like them normally pay, and raises an alert when behaviour departs significantly from both. Every step is yours to drive.

Illustrative exampleInvented Names & Data

Explore the DetectX® Platform

Start with Behavioural Analysis. Add what you need next.

Every DetectX® solution runs on the same underlying platform and shares its capabilities, AlertViewer and audit trail. Pick a solution to see what it adds to Behavioural Analysis.

All solutions
Business IntelligenceWorkflowsRegulatory Reporting

Behavioural Analysis FAQs

What compliance teams ask before choosing DetectX®.

Something we have not covered?Talk to our team

Behavioural analysis compares a customer’s current payment behaviour with what is normal for them and for customers like them, so that unusual activity stands out even when it breaks no fixed rule. In DetectX®, customers are grouped into clusters from their transaction history. Each customer is compared with their cluster and their own past, and an alert is raised when the deviation is significant.

PlatformHow customers are clustered

There are three reasons:

  • A change in the customer’s activity level combined with rising behavioural risk, or with an individual high-risk transaction.
  • A significant deviation from the patterns expected for the customer and their cluster.
  • Membership of a high-risk cluster.

The decision rests on five variables read together: activity level, transaction risk, two measures of behavioural risk, and cluster risk.

SolutionHow alerts are worked and reviewed

A business rule checks an absolute threshold that someone has set. Behavioural Analysis sets its thresholds itself, for each customer, and adjusts them as your customer base or the wider economy changes. The two work together: the customer thresholds can be used as variables in your Transaction Monitoring and Transaction Filtering rules.

PlatformUsing thresholds in your rules

Client Risk Scoring builds the initial clustering model, and Transaction Monitoring generates the daily alerts. It can run without client risk profiles, but works best alongside them. Your own customer and transaction categories and risk assessments can be added to sharpen the clusters.

SolutionThe risk profiles it works alongside

At least two years of historical data is recommended. The system needs several months of statistics before it starts raising alerts, and initialising it with historical data shortens that start. The data collected for modelling is anonymised, and the alerts are reviewed and the parameters tuned during implementation.

PlatformHow client data is prepared

Alerts are generated monthly, on the previous calendar month, and daily, on a set time window. Real-time alerts on each payment are possible through integration with Transaction Filtering, with additional implementation effort.

SolutionThresholds applied to each payment

Go Deeper

All resources
InsightHow Behavioural Analysis Enhances Transaction Monitoring

Why rule-based monitoring misses changes in customer behaviour that stay under a fixed limit. It also covers how comparing customers with their own history and with similar customers helps separate genuine anomalies from routine activity.

Read the articleInsightHow Behavioural Analytics Detects Emerging Fraud Patterns Effectively

How behavioural signals, from changes in payment patterns to the way a session is used, reveal fraud that a single transaction check cannot. It also covers where they fit alongside rules.

Read the articleCase studyPLUSCARD

How the fraud team at Pluscard GmbH uses DetectX® to detect fraud and work its alerts, in the words of its Team Leader Chargeback Fraud.

Read the case study

Behavioural Analysis · DetectX®

See Behavioural Analysis in a demo

Book a session on your own data. Bring your questions about the clustering inputs, the history a start needs and how the thresholds feed your monitoring rules.

Request a demoTalk to us
Own history and cluster
Each customer compared with their own patterns and with similar customers.
Dynamic thresholds
Determined by the system and adjusted as the data changes.
Monthly and daily
Daily alerts through Transaction Monitoring; real time optional, through Transaction Filtering.