When implementing Business Intelligence in our company, the first step is usually to focus on analysis of historical data. We want to know what happened in the past. On this basis we can determine our employees’ efficiency, inform the Management Board and shareholders whether we have achieved our assumed objectives, discover in which areas, departments, regions or product groups our company is doing well and where we are performing poorly.
But all this resembles watching a replay of a football match. There is more excitement among fans present at the stadium and tracking events ”live” than watching the replay on TV. By the same token, in data analytics it is more beneficial to anticipate what will happen in the future and organize your business to harness your strengths while avoiding the weakest areas. It’s better to predict the future than to over-analyze the past.
We can divide analytics into 4 types:
- Descriptive analysis. This answers the question “What happened?” It is usually based on relatively simple visualizations presenting aggregated data, replaced with the most important indicators. We analyze things like sales volume, margin value, number of acquired clients and many other factors.
- Diagnostic analysis. This answers the question “Why is it happening?” For this purpose, we use drill-down mechanisms. We can see in which region a given indicator performs the best and worst; which month recorded declines and which increases; which salespeople recorded good and bad results; which product group sold poorly despite an intensive promotional campaign.
- Predictive analysis. This answers the question “What will happen?” To predict the future, we need to find patterns in historical data and apply them to current and future situations. If we can predict that a given product will not sell well in a given region in a given month, we may not fill the warehouse with it. If we anticipate that a client generating satisfactory profitability may leave us, we can prepare a better offer to keep them.
- Prescriptive analysis. This answers the question “What will we do?” By default, to avoid losses or to take advantage of favorable circumstances. Here, analysis will help us do things like set prices at a given time for a given customer segment at a level that maintains margins on the one hand, while on the other hand doesn’t drive customers to the competition because of high prices. We can estimate what offer will have the highest chance of acceptance by our clients.
When implementing a Business Intelligence solution, it is important that, when switching between different types of analytics, we correctly define the problems and decisions to be made, and that we set goals to achieve. If we don’t correctly define our decision-making scheme, it is easy to fall into the trap of implementing a technically perfect set of reports which after a while will be useless.
Here is what you should think about first:
- What do I need the information for?
- How will I judge whether it means success or failure?
- What kind of actions will I take on the basis of the information?
- What mechanisms will help me plan these actions in detail?
A typical implementation of a Business Intelligence system most often addresses the first two types of analysis (descriptive and diagnostic). The other two (predictive and prescriptive) frequently require the development of predictive models and simulations, which can be done using machine learning.
I suggest that you start implementing Business Intelligence from the MS Power BI self-service platform . While it provides excensive functional possibilities, it does not present a high barrier to entry. If you want to go further with sales data-oriented analytics, we invite you to learn more about our Upsaily tool.