When we're looking at behavior, one important way to measure success is actor activity. We tend to care about how active and engaged our users are, when our devices are most heavily loaded, which discussion topics are most active, and how that changes week to week. Interana makes it simple and painless to examine the activity of observed actors by defining an Activity measure for per-actor metrics. You've already seen it in action on the landing dashboard, and here we're going to take a deeper look.
Going back to the beginning
Let's jump back to the landing dashboard and find the activity charts. We'll use them as a starting point for exploring actor activity. Now that you have your own dashboard, you'll need to look under All dashboards to find the landing dashboard:
or you could just click the Interana logo like we did before. Head back there now, and scroll down until you come to the bar chart titled "Days Active out of 7 for Existing vs. New Users (last 28 days)" that looks something like this:
The chart shows the share of users who were active a certain number of days out of 7 over the last 28 days. It also compares the activity between newly registered and older users. Looking at the activity, you can see that newer users skew toward being active fewer days out of 7 than older users. Let's use this chart to take a deeper look at activity measures. Click the compass to bring the chart into the Explorer.
How can I use it? In this example we're looking at recently acquired users. Knowing how newer and older (retained) users differ in their activity helps you figure out how to adapt your system for the behavior you want. For example:
- Web services companies can see how their most valuable users started out and try to encourage similar behavior in their new users.
- Enterprise software companies can track the on-boarding process for new users and drive them toward taking advantage of their most powerful and sticky features.
- Electronics companies can monitor for DOA or early failures in their connected devices.
Activity measures and per-actor metrics
Once the chart shows up in the Explorer, take a minute to examine how the query was defined. The View we're using is a Bar chart. Like last time, it's comparing users in and not in the new users cohort. The measure being reported is a count of unique users. The activity measure shows up in the compare group. Notice that we're grouping by a metric named "User_DaysActiveLast7".
This is the activity metric we want to examine in more detail. Click the little tool tip icon to see how the metric is defined:
You can see the publisher-supplied description on the top line, and the automatic description generated by Interana below that. We want to see how that metric was defined. We could use the navigation panel to jump into Metrics (the icon that looks like a calculator) and find it in the list, but thanks to the tool tip there's an easier way. Just click that blue Edit link and you'll go straight to the metric:
What you'll see is a box that looks like this:
Here's a bit of background definition: This is a per-actor metric, which means we're calculating the metric for each actor in our observation window that matches our filter. The actor in the For Each box is
user. The Measure is Activity, with the active period being defined in terms of Days. There are no filters, so all events are examined. The critical part here is the Time override section. By default, per-actor metrics get the time range to examine from the query definition that uses the metrics. Here, we're overriding the default and telling the metric to calculate the measure as a series of values defined by one week time windows. For example, when we use this metric in a query over 4 weeks, we will get a series of 4 values for each user. You can learn much more about metrics and measures on our documentation site.
Ok, so let's say that we wanted to modify the metric to work over the full 4 week period. You can't edit this particular metric because it was published by somebody else, but you can conveniently copy, modify, and save it for yourself. Just click the friendly blue button in the bottom right corner!
You'll be presented with a box asking you to name the new metric. Let's call it "User_DaysActiveLast28" and then click the blue OK button. In the background, Interana went to the Metrics screen. Your new metric should now show up in the list of Per-Actor Metrics:
Notice what's different about the new metric? It shows up with Owner listed as "Mine", with a trash can icon next to it so you can delete it. If you're ever working with a very long list of metrics, you can sort the list by Owner or Last Modified to easily see your metrics. Or use your browser's text search function to find the name.
Ok, you now have an editable copy of the original metric. But it's still using the original definition. Edit it by clicking the pencil icon next to the name:
The name and dataset can't be changed now that the metric is defined, but you can edit everything else. Change the Time window toward the bottom of the box from 1 to 4 Weeks. Update the description to match the 4 week time window. Then click the big blue Save button in the lower right corner. Your metric definition gets updated, and you can now use it in your queries.
What makes it special: Interana makes activity metrics simple to define. We provide a visual, intuitive and interactive experience to encourage iterative exploration. We always calculate behavior on the freshest data whenever the query is run. Interana is efficient and powerful enough to deliver results across billions of events in seconds, without pre-aggregation or tons of expensive RAM.
Ok, let's go back to the Explorer by clicking the compass icon on the navigation panel. The query using the Days Active out of 7 should still be in the Explorer, but if not, you should be able to pull it back up from the query history. Or just start again from the dashboard. Once there, delete the original metric in the compare groups box by clicking the X:
and select your new metric by typing part of its name (e.g., 28) and selecting it from the matching list:
Click the GO button and check out the resulting chart:
Cool! It's definitely showing more days. But something still isn't 100% right. There are fewer than 28 days displayed. And there's an "All others" column listed in the legend but not displayed. That's due to the chart default settings that are easy to adjust! Just go to the CHART CONTROLS and change the number of bars from 10 to 28:
Click APPLY and take a look at the resulting chart:
That's more like it. We have all 28 days showing on the chart. We can see that the pattern still holds: older users are generally active for more days than newer users. If you hover over the leftmost bars, you can see the exact percentages:
Save your new chart by pinning it to your dashboard, and let's go on to learning about Metrics in general.
Please keep in mind that you're using a shared demo system meant for learning by everybody. The dashboards and objects you create will stick around for a while, but we will periodically clean up the system and remove stale accounts.