The Need for Information Governance
By frank.buytendijk on Jan 05, 2010
I had a wonderful discussion with Wouter van Aerle of Cap Gemini, about the need for information governance. He pointed out that this is taken care of very well, when it is about financial information, but often surprisingly poorly managed for non-financial information. This is concerning, because required external reporting increasingly consists of non-financial information.
Non-financial external reporting is often still based on Excel spreadsheets, or a simple Access database that is hacked together. The data may look nice in a dashboard or official report, but may be directly keyed in by the person responsible for the report. No auditability, no reliability, no repeatability.
One of the things that you could do is demand that all external reporting goes through the financial consolidation system. It has all the controls that are needed for a reliable reporting process. This is increasingly the case for sustainability reporting, including CO2, health and safety and other performance indicators from e.g. the GRI framework. Even if there is no legal obligation for these reports to be auditable, it still makes sense, as critical stakeholders pay a lot of attention to them. But how does the data get into these reports? For financial data this is taken care of very well, the general ledger is also a highly controlled environment.
Is a data warehouse enough of a governance framework to ensure the data quality for externally reported performance indicators? The numbers can still be typed in, and uploaded from a simple spreadsheet.
The key difference between financial and non-financial information is the wide variety of data sources of the non-financial information, and recording that data is not always a very well-defined process.
Maybe this is the core of the matter. Data is still insufficiently seen as an asset. Any employee that handles money or materials in the same way data is usually handled, would probably get fired immediately. The best definition of BI that I ever heard was "BI is using information as a factor of production" (like labor, capital, materials and facilities). From a governance point of view, organizations therefore do not have a lot of true BI.
I recently attended a conference where a county in of the United States presented their results. One of the performance indicators was about the punctuality of public transportation. This used to be measued by sending out surveys to team leaders. Team leaders are not very motivated to fill in surveys like that accurately, and particularly not if the results are let's say not optimal. At one moment, all the buses were equipped with GPS chips, and this was also the opportunity to improve the measurement process. The data could simply be extracted from a system.
Measurement processes need to be as invisible as possible, and not hold up operational processes. I expect a lot of sensor technology that will become pervasive over the coming years. It allows us to create elegant solutions for measuring a wide range of results in operational processes.
But what to do in the meantime? A solution that have used in the past is to apply data quality color coding to numbers, or define a performance indicator that specifically measures data quality (a quality indicator). The indicator is set to "high" if the data comes from a well governed source system, set on "low" when it originates from a spreadsheet and perhaps two or three qualifications in between. It doesn't solve the problem, but it creates a feedback loop that shows how the data must be interpreted. Then through a process of prioritization improvement activities can be put in motion. (it is important that these quality indicators appear in the same report as the performance indicator, and not in a separate report. A separate report will not have the impact, it will be seen as "technical").
I am not aware of any formal data governance framework that can be used. You?