One of the basic ideas that we introduce to you is that the data in your systems is of different value to your organization. A smart Data Management strategy should identify this value and deploy tactics and solutions that help you to handle the data inline with its value. Typical questions we discuss on this matter are:
- How do you measure the value of data to our organization?
- Which criteria can be used to define the value? Who should define these criteria?
- How can you measure these criteria? What is the unit of measure?
Following a very simple analogy many vendors use temperature grades hot, warm and cold to classify data by importance. Whilst I like this model for its simplicity I experience that it lacks differentiation. Hence I want to introduce a slightly refined version and rate the temperature in 3 dimensions: type of data, granularity and age.
Type of data indicates that certain data is “hotter” than others. Sales pipeline will be more valuable to your business than user logs. This dimension is necessary to identify temporary or technical data, historical applications or to differentiate hot from warm data.
Granularity is a more important dimension. It helps split transactional data (high granularity) from data that serves analytical purposes (low granularity).
Using data age as a dimension is certainly the most intuitive. Current research suggests, that the access rate to data older than 90 days drops by 80%. Our experience teaches us, that data from the current 2 quarters can be called hot in the sense of business critical. Of course certain applications may refer to the current financial year as hot. While warm data is typically characterized by its use for comparison (e.g. previous quarter, 4 rolling quarters, previous fiscal year, last Olympic games) data being older than 2 years of age data is generally referred to as cold.
If applied to the real world you will need all 3 dimensions to categorize your data in temperature grades. Where a single invoice may be useless quickly, the total financial revenues of a quarter may be accessed for a long time to come. On the other hand, data from technical meters may only be valuable for a couple of days.
Apply this to your data
These concepts and ideas are an invitation to start thinking about smart data management. While this post focuses on strategic and tactical aspects I encourage you to take action. Remember the Japanese saying: You miss 100% of the shots you don’t take.