Analytics Manager

AM enables Sphonic to deliver computed data in rapid time to clients. This in turn takes out the need for complex computation and processing of data by decision engines built on legacy platforms and code.

Clients’ fraud engines are then used to develop rules based on computed data.

Example 1 - Velocity

As an example many fraud engines enable clients to run velocity rules such as the number of card transactions performed by a client in a defined period. Many of these systems allow rules to be run on a value or set of values over a limited set of time periods
As customers can do so much more in a consumer not present environment such linear rules are not necessarily relevant requiring more context based on Mid or event or expanded timeframes and including comparison analytics based on a customer history or history of all customers in a Merchants portfolio
AM enables clients to compute data such as a client may have performed 5 transactions on their card today but based on that Merchant category (MCC0 or Merchant (MID) itself the client usually performs on average 5.2 transactions in a 24 hr period and customers in general perform 5.6 transactions on average for that merchant category (or MID) in the same timeframe.
This enables decisions to be made truly “knowing your customer” and how they compare to other customers too returning values illustrated below:

  • Count of transactions in 24 hrs by Mid = 5
  • Average rolling count of transactions in 24 hrs by this consumer at this Mid = 5.2
  • Average rolling count of transactions in 24 hrs by all consumers at this Mid = 5.6
  • Variance current to average for current customer
  • Variance current to average compared against all customers

As a result a client just needs to run rules based on the value returned above or just the variances without the fraud engine having to run through and process all of the computations and changing data models

Example 2 - Associations

AM can create X/Y pairs comparing a static data element (X) to another data element (Y) returning data such as:

  • How many cards have been seen against a unique device
  • How many devices have been seen against a unique card
  • On Average against a Merchants portfolio how many cards are seen against a device
  • On Average against a Merchants portfolio how many devices are seen against a card
  • Variance current to average for both scenarios

X/Y Pairs can be created against all data elements and also segmented further by Timeframes or Merchant Categories or Events e.g. How many identities are seen on average against a unique device for the event “account registrations"
This complex processing brings “context” to the decision making process by delivering computed values to the end clients decisioning process
Built with bleeding edge technology, AM has been designed to deliver truly valuable data with very low latency ensuring decisions can be made on relevant data.