Safety in numbers

Listening to the radio on my way into work this morning I was entertained by a story relating to an error in a mathematical program that was used to estimate the weight of dinosaurs.

It seems that the error was overstating their weight by rather large amounts – for instance there was a 16 ton weight gain for the Brachiosaurus. Interestingly it appears that the mistake has been perpetuated for the past 25 years.

Although I am not a palaeontologist I am not entirely surprised by the revelation. Carrying on the theme of my recent posts regarding the use, or overuse of Excel, in business (and sometime substitute for a business system) it reminded me that a great many spreadsheets contain errors that often lead people to draw the wrong conclusions, which they then use to make important decisions.

Although our main focus is on implementing and supporting ERP such as Microsoft Dynamics AX and Infor ERP LN, as well as other complimentary products, we also develop custom applications, if required, for our customers. One recent example was a secure web site for a European company to allow them to collect their health and safety statistics on line. Information about incidents such as slips, trips and falls and safety inspections had to be recorded by each of the 60 plus sites.

The web based solution was designed to replace the previous reporting regime that was based wholly around the use of Excel. Each month end the chief safety officer had to badger the sites for their reports and then spend several days collating the data into regional and global analysis. It was time consuming and tedious. He had spent some time trying to speed things up by writing complex macros – he was becoming quite an Excel expert.

We finished the development of the new website and as a part of the testing he entered some historical data. Unfortunately, the results on the web site for some critical statistics were coming out ‘wrong’. This clearly had to be corrected as the statistics were reported to the global organisation and, quite importantly, many people had an element of their bonuses based on achieving certain health and safety targets.

We checked our program. It appeared to be correct. We then checked the spreadsheet. It was wrong. It transpired that the previous two years of statistics had been wrong. Had people lost money because of this error? I’ll let you guess.

The lesson is that if you are going to rely on Excel for collecting, collating and calculating critical data then at the very least get it checked.