One of the most common practice of store employees to defraud retailer is to bring down (Mark Down) the price of merchandise at a level which is much lower than the regular sales price of the product. Mark downs also take place regularly due to various promotional schemes which retailers offer to its customers. Products nearing 'best before date' also needs to be marked down so as to clear the stock before expiry date. In the interest of business, such privileges are also delegated to individual stores. Dishonest employees use this privilege to either derive personal monetary advantage or use it to pass unauthorised benefits to their friends and families. Since the billing is done on POS, the sales would get recorded in data ware house.
To protect the interest of business, it is most desirable that analysis of this 'marked down' data be carried out regularly at back end. That would indicate if marked downs are being done for operational reasons or for personal reasons. Once such analysis is completed then only investigator should visit store for final findings. This flows from the fact that it is highly undesirable that loss prevention / security officer visits stores to detect malpractices. His presence at store is more of interruption to business rather than to expedite the business.
It is obvious from the above, that we need to have as much detailed and focused analysis of data, as far as possible, at the back end so that minimum time of investigator is spent at store.
To help investigators interpret the data, a sample and simple method of recognizing a pattern indicative of fraud, has been discussed in slides below.
To protect the interest of business, it is most desirable that analysis of this 'marked down' data be carried out regularly at back end. That would indicate if marked downs are being done for operational reasons or for personal reasons. Once such analysis is completed then only investigator should visit store for final findings. This flows from the fact that it is highly undesirable that loss prevention / security officer visits stores to detect malpractices. His presence at store is more of interruption to business rather than to expedite the business.
It is obvious from the above, that we need to have as much detailed and focused analysis of data, as far as possible, at the back end so that minimum time of investigator is spent at store.
To help investigators interpret the data, a sample and simple method of recognizing a pattern indicative of fraud, has been discussed in slides below.
Slide 1 This fist picture explains the methodology to narrow down data from millions of entries to few hundreds and then analyze those few entries further.
Slide 2 After having narrowed down the data to few hundred entries from million entries earlier, now the focus is to narrow down further to specific stores.
Slide 3 If the analyst has reached the point as given in this picture, be sure that he / she would be able to home on to the most suspicious entries and the chances are that even sitting at back end, he would be able to clearly discern as to what actions at POS (whether malicious or otherwise) would have led this entry to crop up in data warehouse. 90% chances are that the analyst shall be able to home on that single suspicious data out of whole data – in other words, he / she would be able to find a needle in the hay stack.
There could be many more patterns available to be flagged depending upon extent of automation in the company, type of reports being pulled out of data warehouse and also on experience of the investigators / loss prevention officers to interpret the same. Since it is not possible to discuss all reports and patterns in the given time and space available – a sample analysis is attached.
Slide 3 If the analyst has reached the point as given in this picture, be sure that he / she would be able to home on to the most suspicious entries and the chances are that even sitting at back end, he would be able to clearly discern as to what actions at POS (whether malicious or otherwise) would have led this entry to crop up in data warehouse. 90% chances are that the analyst shall be able to home on that single suspicious data out of whole data – in other words, he / she would be able to find a needle in the hay stack.
There could be many more patterns available to be flagged depending upon extent of automation in the company, type of reports being pulled out of data warehouse and also on experience of the investigators / loss prevention officers to interpret the same. Since it is not possible to discuss all reports and patterns in the given time and space available – a sample analysis is attached.
No comments:
Post a Comment