Weight Matters: Tell Me What’s Important

One of the challenges that Facet nested table users face is getting quick access to important dimension value results obtained from a query. For example, a campaign manager at a DSP may want to monitor campaign performance on a daily basis by sorting on the percentage change in impressions.  To be efficient, the campaign manager may be interested in finding any changes for the important campaigns. The high ad spend or impressions generated by this campaign may make it an important campaign to monitor. Based on the patterns of the metrics used across all our customer dashboards, we have heuristically determined that  a correlation exists between certain performance metrics and the importance of what our customers are interested in. The new weighted sort functionality in Facet Nested Tables utilizes this heuristic to predict the importance of a result. If this option is selected, it limits the results to the top 100 values of first split dimension by importance.

Let us see how this works. Consider the following campaigns that are sorted by percentage change for the click-through rate (CTR) on a week-to-week comparison as shown in Figure 1. It is difficult to understand which of these campaigns are important from the point of other metrics.

Figure1

Figure 1: Campaign names sorted by percentage change for CTR on a week-to-week comparison.

The weighted sort feature is turned on by clicking on the option in the top right hand corner and the results are shown in Figure 2.

Figure2

 

Figure 2: Campaigns weight sorted by percentage change for CTR on a week-to-week comparison.

Figure 3 shows the campaigns that are no longer on the list and others that have moved up to take their place. The reason why these results were removed was because they were not in the top 100 list that was determined by the weighted sort algorithm.

Figure3

 

Figure 3: Campaigns that were removed by the weighted sort algorithm.

Let us try and understand and see if we can explain why these results were removed. We choose Campaign 60-1 as the baseline and compare it with all the results above it that were removed by the weighted sort algorithm.  Figure 4 shows that Campaign 60-1 had a much higher impression volume and clicks compared to those above in the list. From the point of view of the weighted sort algorithm, Campaign 60-1 is more important than others on the list and hence is bubbled up.

Figure 4

 

 Figure 4: Comparison of Campaign 60-1 with other campaigns that were removed by the weighted sort algorithm

Now, It’s Your Turn

Give the new weighted sort feature a try and let us know what you think.

Filed in Corporate