A Data Science Central Community
There are different levels of reporting, depending on how much you demand yourself as a writer of the report. Firstly, a good report is another expression of a regression model. It tells you what variables are causing the change in business performance. Secondly, a good business report tells you new trend in terms of recent movement. For example, it tells you there are some new opportunities somewhere out there which your competitors may not aware.
To deliver a good analysis of causation, one needs to think about the hypothesis in the.mind, and find numbers to support. The support has to tell whether it is caused by the market factor in general, seasonal trend that repeats the same in every year, individual actions or simply a noise that should be ignored. Does it sound like a regression model? In fact it should be supported by quantitative model if it can be. Of course, when communicating in a written report, there should not be any smell of any models.
To make sure you spot the right trend, you need to monitor all dimensions to its finest level. E.g to capture the growth opportunities of any market, there should be monitoring of individual countries, regardless of its size such Vanuatu.
There are so many dimensions to monitor, which one is a better one? The one that could drives customer behavior. If you need have two choices: one to report performance by sales team's country another for customers' location, I would prefer to monitor the fundamental- by customers' location, except that your product has no real economic value- its sales is purely by how cunning you are- then use sales force will explain most of the purchase behavior of the customers.
Not sure if this is clear enough to explain what makes a good report. Next then I can share about the experience of building up the predictive models.