I was recently reminded, as we are doing our own financials, that dirty data is the bane of any Business Intelligence or Activity Based Costing implementation team. It seems that there is a wealth of data available but if it is not clean or validated it adds significantly to the overall implementation process. It has been my experience when drafting implementation plans for large ABC projects that we allow quite a significant percentage of time (could be up to 33%) to find, source and cleanse data and develop automated processes so this can be easily replicated each update period.

I could rattle off numerous examples of bad data issues we have struck, everything from millions of dollars of explosive ordnance allocated to a child care facility (yes its true) or a one person office that was supposed to have four periscopes hidden under his desk somewhere. In both cases it was a data entry error and data entry errors are very common, particularly if there are no specific instructions on what to do in a particular situation, or if it requires manually typing in a string of alpha-numeric codes.

So how can you fix it? I’m a big proponent of fixing data issues at the source, there is no point going to the huge effort of cleansing and rectifying the data in your BI or ABC model if this is not fed back to the source data owners, because the same issues will occur the next update cycle. There are also basic improvements that can be made to data entry screens, like removing the requirement to enter strings of identifying characters and simply allow them to choose from a drop down menu.

Also it has been our experience that simply implementing an ABC model is excellent feedback on data quality issues. Because the model results are pushed out beyond the finance department to those with budget responsibility they start to see all costs that are allocated to their area, not just their budget. If the Engineering Faculty is suddenly lumbered with a large facility depreciation cost, and that building actually belongs to the Business Faculty then they are quick to point this out, particularly when these reports are transparent to senior management.

Data quality has always been and continues to be an issue for all types of BI, BPM, CPM, ABC any type of Management Information system, some keys to improving this process:

• Make the systems transparent – do not build a magic black box
• Push the results of the system out of the finance department down to lower levels of management, this provides important feedback
• Make changes at the SOURCE not in the BI, ABC system.