Descriptive analytics primarily looks at the past. With the help of historical data, it examines and visually depicts what happened. Literally translated, descriptive analytics means the description of data. On the one hand, this data (without predictive analytics) can describe the course of the past or the actual state. The actual state is in turn the result of all events in the past. Descriptive analytics accordingly includes a view into the past (retrospective) as well as into the present. Predictive and prescriptive analytics, on the other hand, look to the future.
Use of descriptive analytics
We use descriptive analytics with a focus on the following two goals. On the one hand, descriptive analytics can be used to gain insights for later predictive analytics. On the other hand, the methods of descriptive analytics offer excellent opportunities to build BI dashboards for controlling.
Descriptive Analytics in Data Mining
Descriptive analytics is an elementary component of data mining. Following the CRISP-DM, it is used in the phase of data understanding. Exploratory data analysis is a sub-area of descriptive analytics. When we are in the data understanding phase, the aim is to understand the data without the help of any models. An initial understanding is generated of how the data is structured, interrelated, correlated and has behaved in the past. These insights can later be used in predictive analytics (i.e. modelling) or in BI dashboards.
There are numerous methods, procedures and visualisations for this.
Statistical method include among others:
- Summary statistics such as means, standard deviations, quantiles, minimum and maximum
- Cross tabulations
Visualizations include among others:
- Density plots
- Correlation plots
Descriptive Analytics in BI-Dashboards
Ideally, descriptive analytics are integrated into BI dashboards that enable interactive handling of descriptive analytics and their visual representation. Here, too, descriptive analytics enables the representation of the past (retrospective) or the present (actual state). The dashboards we use with our clients are characterised by interactivity. This means that different dimensions, such as time or categorical variables (such as regions) can be changed interactively. This results in graphics and tables that are updated in real time.
BI dashboards in sales
Dashboards can be used in two ways in sales: To support the sales rep and for sales controlling.
Support of sales rep
A combination of descriptive and predictive analytics is optimally used in these dashboards. Descriptive analytics shows the retrospective and the actual state, while predictive analytics writes predictions for the future.
In general, there are almost no limits to the possibilities for these dashboards. Descriptive analytics can, for example, visualise which products the customer owns, when they were bought, how often they were bought and what the average value of the purchases was. Furthermore, the customer can be compared with similar customers or other customers in the region (benchmarking). Interactions, touchpoints (sales and marketing) can also be visualised. The objectives and informative value of the respective visualisations and tables are to be defined individually in each case.
Predictive analytics helps salespeople prioritise customers. Models can provide a score for each customer per product, enabling salespeople to target the most promising customers first. Predictive analytics enables the following predictions, for example:
- Closing scores for different products
- Customer Lifetime Value (CLV)
- Assigned segments (clustering)
- Churn score
While the individual representative works at the customer level from a data point of view, sales controlling looks at the data in aggregated form. Aggregations can be done by geographical regions, product lines or other arbitrary allocations. Here, too, a distinction can be made between descriptive analytics and predictive analytics. In both areas, meaningful dashboards can be used for sales controlling.
The figures and metrics to be considered are often not very different from those used by individual salespeople. However, this data is now aggregated to allow comparability between, for example, geographic regions or product lines.
Below are some examples of key figures and metrics already used in dashboards by our clients:
- Number of units sold in region X in period X
- Number of customers in regions
- Monetary contribution of all customers in region X
- Sales potential in region X
- Total customer value (CLV) in region X
- Expected revenue in region X
The presentation can be realised via interactive tables, graphs and geographical maps like the one below.