Group analytics

Last updated:

Note: This is a paid feature and is not available on the open-source or free cloud plan. Learn more here.

Group analytics is a powerful feature in PostHog that allows you to perform analytics on entities other than single users. This feature is slightly more advanced, but opens up entirely new avenues for analytics which are essential for most B2B use cases.

Group analytics can sometimes feel confusing to new users, and this often times comes from skipping over the basic concepts too quickly. We'll start with an overview of how Group analytics works, and then jump into creating new groups and viewing them within PostHog.

1. What is a group?

In PostHog, a 'Group' is an entity that we can send events and hold properties that is beyond a single user.

To get a more concrete idea of what this looks like in practice, here are a couple examples of common entities that we can model using groups:

  1. Organizations/Teams: It's very common in B2B apps to have multiple users under the same 'Team' account, in which case groups allow us to perform analytics on individual teams exactly as if they were users.
  2. Threads/Chats/Posts: If you're creating a social app, you might want to track events that are connected with specific conversations such as replies

Looking at these examples, it might be tempting to think that a group is just a group of users; however, this is not exactly the case. We'll see this later, but the basic idea is that only events are connected to groups directly, and any user who sent one of those events will also be linked indirectly. In this way, there isn't a way to directly 'assign' a user to a group - instead, this user would send an event with a special property linking the event to a group.

Groups vs. group types

Another important distinction is between a Group and a Group type. At a high level, a Group type is the abstract type of whatever our group represents (e.g. Teams, Group chats, Posts, etc.), whereas a Group is a specific instance of one of these types (e.g. a specific team or a specific group chat).

2. Creating a group type

Now that we're familiar with how Group analytics works, let's start by creating our first group type. In this example, let's imagine that we have a B2B app and want to track when a user join a new team.'company', '84359520')
posthog.capture('User joined company')

To associate an event with a specific group, we provide two pieces of information the group type (company) and the ID of the specific group (84359520). This will create a new company group type in PostHog and associate the User joined company event with the ID of the group we specified. This ID behaves similar to the distinct_id for users we saw before, and best practice is to use the unique ID of the object within your database.

We can also use groups to associate events with other entities. Let's imagine our B2B app has a chat feature, and we want to track when a user sends a message in a group chat.

// This works almost exactly like `identify` but for groups instead
// All future events from this session will be associated with this group'chat', '84359520')
posthog.capture('Message sent')

3. Setting group properties

In addition to associating events with Groups, it's also common to want to set properties on them. Continuing our example, let's try adding the name of the group chat to our group.'company', '84359520', {
name: 'Hedgehog corporation limited',

In posthog-js, we can use the same group method we saw above, while in the backend libraries we can use the groupIdentify method.

4. Using groups in PostHog

Our app is now fully set up to track events for each of our companies and view the groups that a user is related to.

Furthermore, we can use groups now in the following places:

  1. Creating insights: Groups can be used in insights exactly the same as people - we can analyze the total number of active groups and filter events based on group properties.
  2. Related groups: See all the different groups a specific user is related to.
  3. Feature flags: Roll out features to entire teams without worrying about different users seeing different things.
  4. Experiments: Run A/B tests on entire groups instead of individual users.


Was this page useful?

Next article

Next steps

Now that PostHog is set up, you might want to import historical events . Or just skip ahead and learn what you can do with the product below:

Read next article