- Augmented Analytics
- Big Data Analytics
- Business Analytics
- Data Governance & Data Fabric Architecture
- Data-Driven Strategy & Transformation
- Performance Management & Improvement
- About us
- Insights Room
- Webroom & Events
If organisations are to learn from the numbers and gain business benefit as a result then it makes sense that analysis should be performed on data that is consistent and trusted.
Oct 02, 2019
To this end I am regularly horrified that one of the most cited reasons for users preferring to use Excel as a data analysis and reporting tool is that numbers can be manipulated before being presented.
This is no way to become a data-driven organisation. If anomalies exist then they should be corrected at source if at all possible and adjusting a figure because it does not agree with a chosen narrative undermines its business value. For the same reasons I am not an advocate of “correcting” data within a data warehouse.
There are sometimes legitimate business reasons for applying adjustments (updating estimates, for example) but where this is the case it should be done in a repeatable, transparent, consistent and auditable fashion. Even then, the processes leading to the requirement to adjust data outside of the system responsible for it should be carefully examined and care should be taken not to get out of synch with that system. If this is not done then any hope of achieving a single version of the truth, hence maximising the ability of the business to learn from the numbers, is lost.
Another area in which numbers are inadvertently influenced is inconsistency in how data is used, particularly in the definition of calculations and key reference data. We regularly come across departments using different calculations for something as key as revenue or profit and a common source of inconsistency of figures is down to the particular date used by each party (e.g. order date, manufacture date, date shipped to warehouse, delivered date, invoice date etc.). Agreeing and applying a common set of business rules to govern the use of data across the organisation is extremely important if its value is to be maximised; a topic that I am sure we will return to in the future.
We also need to be careful to avoid introducing bias into our analysis. This topic will be covered in more detail in a later blog in the context of advanced analytics but there are many ways in which bias can be either consciously or unconsciously introduced to even basic reporting scenarios. At the simplest level, it is very easy to choose narrow data sets, ignore inconvenient data or to draw conclusions from partial analysis if these approaches support a preferred outcome. It is human nature to seek reinforcement of one’s opinions rather than to have them challenged. We need only look at the egregious use of statistics in the political arena to see this in action on a daily basis but the business world is by no means immune from this practice.
This is the difference between a data-driven organisation and one that has chosen to weaponise data. The former has a genuine desire to learn from the numbers and to continually improve the business whereas the latter simply wants a vehicle to support its pre-determined point of view.
No prizes for guessing which approach leads to long-term, sustainable business success.