Creating a Random Cohort with 5% of Users in Mixpanel
is it possible to create a random cohort which contains 5% of our users in mixpanel?
is it possible to create a random cohort which contains 5% of our users in mixpanel?
Hi Lara M., Please stand by as I search the Mixpanel knowledge base for an AI-powered answer. I normally take about 15 seconds or so. In case AI cannot answer your question, feel free to reach out to Mixpanel support at mixpanel.com/get-support. If you have follow-up questions for this thread, please tag me again with your question: @kapa.ai <follow-up question> đ
Feel free to ask a follow-up question once I've answered by mentioning me with @kapa.ai
Yes, you can create a random cohort of a specific percentage of your users in Mixpanel using Feature Flags or Query Time Sampling, depending on your objective.
1. Using Feature Flags (For Targeting)
If your goal is to target 5% of your users for a specific experience or experiment, you can use Mixpanel's Feature Flags. You can create a rollout group and set a Rollout Percentage. ** How it works: You define a "Variant allocation" (e.g., 5% to Variant A, 95% to Control). ** Consistency: Use "Sticky Variants" to ensure that once a user is assigned to that 5% random group, they remain in it even if rollout percentages or cohort memberships change [Feature Flags: Rollout with precision control; Rollout Percentage].
2. Using Query Time Sampling (For Analysis)
If you want to analyze a random subset of your users to speed up report loading, you can use Query Time Sampling. ** How it works: When enabled (via the lightning bolt icon in reports), Mixpanel selects a uniformly random sample of 10% of the total population for the analysis. ** Limitation: The knowledge sources specify that the sample size for this feature is fixed at 10% of the total population, so you cannot manually set it to exactly 5% using this specific tool [Query Time Sampling].
3. Manual Workaround (For Static Cohorts)
If you need a specific static cohort of exactly 5% of your users:
1. Export your user list as a CSV [Explore your user base in Mixpanel].
2. Randomly select 5% of the IDs in your own external tool (like Excel or Python).
3. Upload that specific list back into Mixpanel as a CSV and assign them a unique property (e.g., Random_Group == "Group_A") [Creating a Cohort via a CSV].
4. Create a cohort in Mixpanel based on that new property [Creating a Cohort via a CSV].
kapa.ai is still learning and improving, please let me know how I did by reacting below.
âšī¸Â Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.
Let me check the knowledge base for you again. One moment! đ
Based on the provided knowledge sources, it is not explicitly stated that you can directly "sync" a Feature Flag variant as a Cohort to another platform. However, you can achieve this by combining Feature Flags with Cohorts.
Here is how you can sync that 5% of users to another platform:
1. Identify the 5% via Feature Flag
When you set a 5% rollout for a Feature Flag, Mixpanel assigns those users to a specific variant. To sync these users, you first need them to exist in a Cohort.
2. Create a Cohort based on the Flag Variant
Mixpanel allows you to create cohorts based on user properties or behaviors. While the documentation doesn't explicitly show a "Flag Variant" property in the Cohort builder, it mentions that users who qualify for a Feature Flag can be targeted with Cohorts [Feature Flags: Rollout with precision control].
If you track the assignment as an event (e.g., "Experiment Started" or "Feature Flag Assigned"), you can create a Cohort of users who performed that specific event with the property of your 5% variant.
3. Sync the Cohort to Another Platform
Once you have defined that 5% group as a Cohort, you can sync it to various third-party platforms using Cohort Sync [Cohort Sync Overview].
Supported platforms for syncing include: ** Engagement/Marketing: Braze, Airship, MoEngage, Iterable, and Appcues [Cohort-sync Integrations]. ** Advertising: Facebook Ads, Google Ads [Cohort-sync Integrations]. ** Experimentation: VWO, Kameleoon, and Taplytics [VWO Integration; Kameleoon Integration]. ** Data Routers: Segment [Segment Integration]. ** Custom Destinations: You can also sync to a Custom Webhook if your destination platform is not natively supported [Syncing cohorts to a custom webhook].
*Important Note: Cohorts used for targeting and syncing typically refresh on a periodic cadence (often every 2 hours), so there may be a delay between a user being assigned to the 5% group and appearing in the downstream platform [Feature Flags: Rollout with precision control].
kapa.ai is still learning and improving, please let me know how I did by reacting below.
âšī¸Â Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.