“Lies, damn lies and statistics”
As Mark Twain is famous for saying, statistics can be twisted to prove any point.
For many managers, BI applications are a critical decision making tool. The statistics, trends and analytics presented are used to determine future actions and to track historic performance.
There are two significant ways in which poor quality data can make this kind of decision making a lottery.
Most visibly, a manager may be faced with two sets of reports, about the same business area but from different systems. These may provide contradictory analysis – one suggests investing more in a region, the other suggests pulling out. Which does she believe? She has to make a strategic decision that could cost the business millions and cost her her job!
Possibly of more concern, the reports may show a clear leaning to a particular position – but this may be based on incomplete or inaccurate underlying data. One of my clients showed a report indicating a 87% compliance to a key supply chain metric. On investigation we found that the underlying data was only populated for 17% of the records – in fact they had no idea whether or not they were compliant. Are they making good decisions?
Many organisations are turning to data profiling applications to test whether the underlying data supports the reported findings. An increased understanding of data quality issues means two things:
1.) You can make better informed decisions using your confidence in the underlying data to weight reported metrics.
2.) Steps can be taken to address data quality problems that are critically affecting your business.
Improving your key decision making must be a good idea!