Hello, we've trying to figure out the best way to ...
# questions
s
Hello, we've trying to figure out the best way to track the data for a complex user event that we want to be able to do a lot of deep analysis on. We've tried a few different ways of tracking this data, but we're still struggling to find the best way. We're still new to Mixpanel, so there may be something obvious we're missing. Any advice you have on the following situation would be appreciated. We have an event where users are publishing data to our platform. The user can segment their data into one or more ranges and then publish multiple ranges at one time. Each range has unique characteristics that we want to be able to analyze together. For example, we want to be able to see for a publish how many rows and columns were in each range. It's not sufficient to have two separate lists representing row counts and column counts. We need to be able to correlate column count to row count to understand the overall characteristics of each range. There are also properties of the overall publish event that we want to be able to analyze and correlate with range-specific characteristics, so we can understand things like the publish took x amount of time, had x number of ranges, and those ranges breakdown into ranges with x rows and x columns. See the image for an example of the general properties and range specific properties. We tried having two separate events generated when the user publishes data. One event for the overall publish and one event for each range in the publish, but we found that we couldn't correlate and breakdown the overall publish data to the individual range data, as described above, because there isn't a way to link events together in a breakdown like you can in a SQL join. We have also tried having the data in a single event by generating one instance of the event for each range, but then we can't do accurate averages and percentiles on general publish properties. We couldn't find a way to remove the duplicates created by having multiple instances of the event when there is more than one range in a publish. For example, if there are 3 ranges, 3 publish events with a publish time get created. The publish time would count 3 times towards the average, even though it should only count once, which skews the overall average. Thanks in advance for any help you can provide.