In the blog post "What is analytics and why do we need it?" I stated that the data from disparate data sources should ideally be collected into a common data warehouse. Let's take a midsize company as a simple example in order to make this more tangible. The company makes money by billing hours and would therefore at least track the following profit metrics: employee cost, utilisation and forecasted earnings. Even a small company would need the following systems:
- Time planning
These systems all belong to different domains which aren't always integrated in corporations. People working with HR are only responsible for HR, people working with sales are only responsible for sales and so forth. Only the company management is ultimately responsible for all domains.
The built-in reporting and analysis features in the individual systems can't satisfy the reporting needs as no single system has access to all the metrics.
|Employee cost||Utilisation||Forecasted earnings|
|CRM/Sales||No||No||Yes, on an aggregated level|
As management doesn't have access to all the required metrics, they might revert to running their company using only the Profit and Loss statements that are available to them. Business controllers can mitigate the issue by copying and pasting large amounts of data into spreadsheets in order to piece together the complete picture and to create management reports. This is of course time-consuming, error-prone, complex, difficult to oversee and difficult to harmonise across the entire business. This issue is of course amplified as organisations grow, acquire new systems, create new departments and add new key performance metrics.
In short: the data model used in a business not only gives analysts access to well defined and structured data. It also gives management the ability to influence key performance metrics and the way the business perceives itself and the surrounding world. This is a subtle way of making sure that everyone in the organisation works toward the same goals without requiring rigid oversight. For this to work the whole organisation needs to be represented in the data model.