Overview of Cohort Tables in Analysis Workspace overview-of-cohort-tables-in-analysis-workspace

The Rolling Calculation setting within Cohort Tables can be used to analyze cohorts period-over-period, to understand how the same users are retained (or churn) over time.

In this video, I’m going to share with you an overview of the settings and features included in the cohort table in Analysis Workspace. So let’s begin. Here, you can see the builder state for the cohort table. There’s a lot here, and I’ll go through each feature individually to outline what tools are available and what they can do. So one of the first things you’ll notice is the section to define segments for both the inclusion and return criteria. You can drag a segment over from the left rail, or you can use the dropdown menu to select a segment. So let’s say I wanted to analyze my US users as part of my inclusion, but I only want to track Chrome users on return. I can add different segments to each of these sections, and I can add up to 10 different segments for both inclusion and 10 for return as well. As far as the actual criteria itself, you need to define a metric in order to actually build a cohort table. Segments are optional, as you can see. You can add up to three different metrics. So for example, I’ll select three different metrics here. We also have an operator, so you can choose to either group them all together so that the visitors must meet all three of these criteria, or you can set this to OR so they meet one of these three criteria. Additionally, we have operators and numeric values. So if you want to filter out or make users meet a certain threshold in terms of number of visits or orders or whatever it might be, you can come in and add and make changes to the operators here. So you can really filter and narrow down through the user groups that you want to include as part of your segment. You can do the same thing over here on return. You can add up to three different metrics. For this demonstration, we’ll keep things simple and compare visits and orders. But those are the inclusion and return criteria settings that enable you to add multiple segments and multiple metrics as needed to suit your criteria definition. Secondly, you can see down here we have two types of cohort tables. Retention is our default cohort table. It’s the tried and true standard one that you see. After it builds, you can see here is the standard cohort table. We also have a cohort type called churn. If you choose churn and select build, churn will do the inverse of our retention and is marked by the red indicating the fallout instead of the retention of users. That is how often are they churning and not returning to your GiveInsider app. Churn is a great, easy way to see the type of behavior of users who are not coming back and possibly the opportunity of those who you know you want to do a little bit more to engage and focus on that user set. Next, we have a setting called rolling calculation. Currently, the cohort is based on users who meet the inclusion criteria and do the return criteria at any point in the subsequent date range that you have. So users in month four don’t have to meet the return criteria in months three, two, and one. But if I change this to rolling calculation, now you can see I have a completely different type of cohort table. Users must meet the criteria in the previous period. So these five here in month two are of this 50 in month one, which are a subset of the inclusion group here of the 85,000. And that’s the same across the board. In order to be included in month two, you have to have met the criteria in month one and so on. And as you can see, the persistence of my users down through months three, four and five who are doing continual month after month behavior has dropped significantly down to one. And so rolling calculation is really good for period over period retention and analysis to know how your users are coming back on a repeat basis. The next setting is here under advanced where we have two features. The first is a latency table. Latency tables provide a good view of pre post analysis. So with a latency table, you can see that the cohort is a little bit different. Our inclusion has now shifted out to the middle and everything to the right of it is a standard cohort showing you repeat and return users after the inclusion event. But I can also now with the latency table, see pre inclusion activity on the same table really quickly and easily. So this is great for analyzing specific events like product launches or campaigns to see what the behavior was like before the event and then the change in behavior after. And so latency tables again are really good for pre post analysis. The last feature we’ll cover is a custom dimension cohort. And in many cases, we’ve had users request and say, hey, I want to do something not time based here on the left. I want to compare something else, a different dimension. And so the custom dimension cohort allows for that. So if I select the browser dimension, I want to compare browsers side by side to know which one is driving the most online orders for my company. So once that builds, I can now see instead of time here on the left, I have a list of my top 14 dimension items that have returned for that specific dimension. In this case, the different browsers. So I can see which browsers are driving a lot of inclusion and return. So Internet Explorer 6.0 is doing great here. But I can see how that stacks up against Safari, Firefox, Chrome and others to quickly evaluate which browsers are performing best at driving online orders and which ones might not be doing so well. So maybe my web experience on some of these other browsers isn’t as good, or maybe most of my visitors are coming from Internet Explorer. So I want to be doing specific things to improve their experience. But the dimension column is a great way to do some really cool non time based dimension analysis. So you could also compare campaigns, products, pages, any other dimension you can think of to do quick side by side, period over period analysis. You can also use the filter option here on the specific dimension if you want to analyze three or four specific dimensions and not the entire list. You can add those here, or if you have one that’s not on your top 14, you can find that as well. So those are some of the great features we have for the cohort table. I hope that these are useful to help you uncover new insights for your business. And thank you for your time.

For more information, please see the documentation.