To Catch a Thief

Our first goal is of course to prevent theft from happening in the first place (I talked about some of the ways we can accomplish this in Removing Avenues for Theft). However, there is only so much we can do, so we also have to ensure that the thief can easily be caught. In its simplest form, this typically requires that three things are in place:

  1. Adequate measurement of employees on any factors which, if skewed, might indicate that theft is occurring.
  2. Data aggregation, summarization, and correlation—potentially including data from sources outside of the POS.
  3. Robust tools that will allow managers and others to detect theft.

When reviewing a client's process and systems, I usually find that employees are already being measured on many of the most important points. Usually, however, the employee is kept in the dark about the kind of data that is being reported on the back end. Depending on the situation, I may recommend that the company make data collection more visible to the employee, because by doing so, we can often enhance performance and deter employee theft at the same time.

Imagine, for example, that on user sign-in, the system displays a status page showing various graphs. If this page included, for example, a graph of voided items for the previous day and month—and compared the individual's statistics with those of other employees—we might deter employees to use the "voided item" trick to steal money intended for the register. And such a page would become even more powerful if it indicated that the manager has access to even more types of data to monitor potential employee theft.

Where most companies tend to break down is in the aggregation and presentation of data to managers. Typically, detecting employee theft is a very cumbersome process, with managers having to take the time and initiative to review transactions to look for anomalies. Also, there is very little “pushing” of potential problems to the managers. Rather than the system flagging odd trends and atypical transactions for the managers, all the burden usually rests with the managers to root out thieving employees.

Rather than make owners/managers continually seek out potential theft, designers should look for ways to reduce their burden by informing them when user behavior goes outside of the norm. Required data and algorithms for accomplishing this will depend on the situation.