Power BI: Key Influencers Top Segments
21 February 2019
Welcome back to the Power BI blog. This time, we’re going to be talking about the other part of the Key Influencers visual: the Top Segments.
There are two big headings at the top of the visual – Key Influencers and Top Segments. Top Segments is intended to help identify areas where a subset of the data is particularly associated with the key criteria that we’re looking at.
Last week, we looked at sales of over $3,000. This week, we can look at transactions under $200. In this instance, Power BI has identified six different segments that have a higher-than-usual incidence of transactions under $200.
To investigate, we can start by clicking on the first segment, up at 96%.
Firstly, we are told that 96% of the transactions in this segment have a sales value of under $200. This is compared to an overall average of just under 75% of transactions, which is a significant difference. On the left hand side, we are given the characteristics that define this segment:
- Customers who made their first purchase after 28 July 2003
- Customers who made their first purchase on or before 28 August 2003.
So, it appears that people who signed up during this period, from 29 July 2003 to 28 August 2003, purchase significantly lower value items than others. This might be due to a sale or other promotion that occurred during that time, or might be reflective of a particular sales strategy or product release. Either way, this is an insight that has been identified automatically.
We can click on the X in the top right hand corner to close this down, in order to go back to the first screen, or we can click on the circled percentages at the top of the visualisation. Let’s choose segment 5.
Segment 5 is a little different. Here, there are three criteria that define this segment:
- The commute distance of the customer is less than 1 mile
- Customers who made their first purchase after 5 November 2003
- Customers who made their first purchase on or before 31 March 2004.
This is quite a different dataset, stretching over 5 months’ worth of new customers, caveated by the commute distance of 1 mile. Again, there may be reasonable explanations – someone who has a short commute may already have a bike, and may only be looking for spare parts. Alternatively, they may have no need for an expensive bike, and only purchase accessories such as bottles and clothing.
Segments give us a great way to start storytelling with our datasets, and may go to help explain curious subsets of our data that would otherwise be brushed away as an anomaly. Next week, we’ll come back with more thoughts on how to use this great visual. See you then!