Retailer Peacocks has identified 20 cases of internal fraud within the first six weeks of implementing specialist data mining software (Computing,

7 July).

Already five cases of fraud identified by the software have resulted in staff being dismissed, says Chris Miles, Peacocks’ retail operations manager.

‘The cases that we have identified are relative to the six weeks that the software has been live and running,’ said Miles. ‘We could have got more exceptions from it otherwise.’

Peacocks has been using Fraud Alerter software from vendor Innovetra to highlight cases of fraudulent activity, including staff voiding genuine transactions and pocketing the cash from the till, illicit refunds and unauthorised staff discounts.

‘We spent the majority of July and August writing the algorithms and testing the software,’ said Miles. ‘But it generated far more exceptions from the sales data than we would have been able to deal with or wanted.’

Miles says before Peacocks installed the software, it was aware of exceptions in its data but was unable to review it in a meaningful way, or decipher it.

‘Now we have tightened up the software’s algorithms, we are getting less in the way of positive hits and more genuine exceptions,’ said Miles. ‘We can get it to look at every refund of more than £50, for example.’

Peacocks’ field investigators have changed the way they access sales data so it can be run through the data mining software.

‘When we first configured it, the field guys were dialling into the same server that our email ran on, clogging it up and making it run slow,’ said Miles.

Now the software is fully operational, Miles says it will not only identify fraudulent staff activity, but also examine where procedures are breaking down and address training needs.

The technology has proved fraud in three of the cases involving dismissal so far and flagged two others for investigation.

A further 15 cases are ongoing.