Tracking First-Time Users

This article describes how you can do new user analysis in Mixpanel, by filtering on behaviors for new v/s returning users, by filtering on when a user did an event for the very first time in their lifetime.

First-time users arrive at your website or app for the first visit. It is important to distinguish first time versus returning users so you can answer questions, such as:

  • How do first-time users convert through sign up?
  • What is first-time versus returning user retention?
  • What traffic source do my first-time users come from?

The answers to such questions provide valuable product insight not readily available when analyzing all users simultaneously. Measuring first-time users gives you a granular view of what your users are doing when they have never used your website or app before.

You may find user experience insights that you cannot discover without analyzing these users.

For example, you might discover that most of your users do not sign up the first time they visit your site. This insight could mean your home page is too cluttered or your sign up process is not being emphasized enough, allowing you to make the necessary user interface tweaks to test the new layout for future first-time users.

1st time filters

We have recently launched the capability to filter down core analysis reports by "1st time", so as to only count the very first instance of a user doing a particular event in their entire lifetime. Earlier, users had to do a lot of heavy lifting to track special events or properties that marked whether that was the first time a particular user did that, but now this is available without any added tracking or SQL.

Questions you can now answer easily

  • First Time Retention: What % of users come back and watch another video after viewing  a video for the 1st time ever?
  • First Time Conversion: How quickly do users Purchase a paid plan after finishing their 1st ever course?
  • First Time Engagement: How frequently do users do complete a ride request for the 1st time ever each week?
  • First Time Flows: What did users do after viewing their 1st ever job listing?

FAQ on First Time

What reports can I use "1st time" in?

Insights, Funnels, Flows, Retention. Explore and Cohorts are coming soon.

Does "1st time" work with groups that I've setup?

Yes!

How far back are we searching to determine whether the event was the "1st time" it was done by a user?

The entire project history, so this is true 1st time doing an event unless events are deleted via GDPR or via our data retention policy that deletes events older than 5 years.

How does "1st time" work when there are filters applied to the event?

It applies after the inline filters and before the global filters.

Example 1: "1st time" with no other filtersmceclip1.png

This will return the number of 1st time purchases that happened in a given time period.

Example 2: "1st time" + inline filtermceclip0.png

This will return the number of 1st time subscription purchases that happened in a given time period, and not the first time purchases that just happened to be of type subscription; the event and the inline filter get evaluated before the 1st time check is done.

Example 3: "1st time" + inline filter + global filtermceclip3.png

This will return the number of 1st time subscription purchases filtered down to the ones that happened in United States in a given time period. The 1st time check is done on the Purchase event and purchase type equals subscription property filter but NOT on the Country equals United States, so if 1 user (Alice) did their 1st ever subscription purchase in United States and user 2 (Bob) did their 1st ever subscription purchase in Canada, but their 2nd purchase in United States, this query will include Alice but not Bob, because Bob's 1st ever subscription purchase was made in Canada.

Please note: Dashboard filters are also considered global filters.

Example 4: "1st time" + global breakdownmceclip4.png

This will return the number of 1st time subscription purchases segmented by the purchase category of those 1st time subscription purchases. So this query and response tells us that the most common purchase category for 1st time subscription purchases was "meTume Academy".

Example 5: "1st time" + multiple inline OR filtersmceclip5.png

OR filter queries behave just the same way as AND filters, in that all the inline filters get evaluated before the "1st time" check is done. So this above query would return all the 1st time subscription purchases OR all the 1st time meTube Family purchases in a given time period.

Example 6: "All events" + 1st time in Insights

mceclip1.png

This calculates the number of 1st ever events done by users(across all events done across their lifetime) in the last 30 days.

How is this different from "First Seen"?

First Seen is a property on each user that is the timestamp of the first event they performed. This behaves like a profile property and this is just a single timestamp per user.

"1st time" is the feature that lets you filter down your analysis to only count the very first instance of a user doing that event.

Will this update when events are deleted (GDPR), hidden (data views) or users are merged (ID management)?

Yes, it will respect GDPR, data views, Lexicon, ID management, with the only exception being Lexicon hidden events. If a user's first event is deleted, hidden, or merged, we will recompute based on the remaining events we have.

Can I use 1st time filter when defining a custom event?

No - 1st time filters are not supported when creating custom events, but customers can effectively do what they need by defining the custom event and then adding a 1st time filter in the query builder.

Can customers calculate doing an event Nth time instead of 1st time?

For conversion analysis, yes. You can define a funnel like this to get the "2nd ever" watch video for a user:

mceclip0.png

First Seen

You also have access to the timestamp of the very first event that a user did within Mixpanel via the new profile property "First Seen" (previously known as First Event Time).

Screen_Recording_2020-09-14_at_12.18.37_PM.gif

 

Use-cases

  • Calculate total purchases done by users that first showed up on our platform in the past week
  • Breakdown total video watches by the "month" of the users' join date, so in other words, what monthly cohorts are contributing to most engagement on the platform
  • Visualize the user flows for users that just joined the platform this week
  • Calculate conversion from search -> download for users that joined this month
  • Create a cohort of users that were "new" in the last 7 days.
    • Please note: Because "First Seen" is a profile property, building a cohort of users where "First Seen" is in the last 7 days" and visualizing the trend over time in Insights won't give you dynamic cohorts over time; it will show a flat line, which represents the number of users that were "new" in the last 7 days from present date.

FAQ on First Seen

Does this apply for users without a profile? What about for a user that has not done any events?

First Seen will be defined for any user that has done at least one event.

If a user has a profile, but no events, First Seen will not be defined for that user.

Will this work on Groups (eg. First Seen on a Billing Account)?

Yes, it will be the timestamp of the first event done by that Billing Account.

What are the limitations/caveats?

First Seen is computed based on the user's event history. This means that if events are deleted, either via GDPR or automatically via our data retention policy (by default: after 5 years of it being tracked), First Seen will update itself to the new earliest event timestamp for that user.

How does "First Seen" get calculated when you merge identities?

This is calculated at query time, so First Seen will be the timestamp for the first event across the two identities.

If I define a cohort of "new users", where First Seen was in the last 7 days, can I visualize the "new users cohort" trend over time in Insights?

Unfortunately, because "First Seen" is a profile property, building a cohort of users where "First Seen" is in the last 7 days" and visualizing the trend over time in Insights won't give you dynamic cohorts over time; it will show a flat line, which represents the number of users that were "new" in the last 7 days from present date.

Did this answer your question?

Comments

0 comments

Please sign in to leave a comment.