From 7c0e43e3535c0b367e4782d374cf996808d0451f Mon Sep 17 00:00:00 2001 From: Myron Fung Date: Wed, 8 Jul 2026 09:16:03 -0700 Subject: [PATCH 01/10] add cohort size limit --- docs/users/cohorts.mdx | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/docs/users/cohorts.mdx b/docs/users/cohorts.mdx index 43e9f7d1..4ec084e7 100644 --- a/docs/users/cohorts.mdx +++ b/docs/users/cohorts.mdx @@ -185,6 +185,9 @@ You can delete any Cohorts from your project by navigating to **Lexicon > Cohort ## FAQ +### What is the cohort size limit? +Mixpanel supports up to 20 million users per cohort. + ### How can I create cohorts based on dropped-off or retained users? There are two ways to create such cohorts. @@ -195,7 +198,7 @@ The second way would be to use the Users page to create cohorts. For example, if If you would like to find those who have dropped off, you could switch “did” to “did not do”. Here is the cohort for reference: https://mixpanel.com/project/3187764/view/3699044/app/users#Umm8E4JZxLSe -### I created two cohorts, Cohort AA: for users watched videos the last 30 days, and Cohort B: for users who liked videos in the last 30 days. I want to find out the users who did not watch videos AND did not like videos in the last 30 days. How do I create such a cohort? +### I created two cohorts, Cohort A for users watched videos the last 30 days, and Cohort B for users who liked videos in the last 30 days. I want to find out the users who did not watch videos and did not like videos in the last 30 days. How do I create such a cohort? You can do so by using the “AND” operator. Here is a cohort for reference: https://mixpanel.com/project/3187764/view/3699044/app/users#Tc76sqFRzbao From c60b048ce89e4210d666cf2e19b9fa58ee877de2 Mon Sep 17 00:00:00 2001 From: Myron Fung Date: Wed, 8 Jul 2026 14:25:10 -0700 Subject: [PATCH 02/10] remove experiment top banner --- docs/experiments.mdx | 6 ------ 1 file changed, 6 deletions(-) diff --git a/docs/experiments.mdx b/docs/experiments.mdx index b4d67ba0..511da086 100644 --- a/docs/experiments.mdx +++ b/docs/experiments.mdx @@ -3,12 +3,6 @@ title: "Experiments: Measure the impact of a/b testing" sidebarTitle: "Experiments" --- - - The Experiment Report is a separately priced product add-on. It is currently only offered to those on the Enterprise Plan. See our [pricing page](https://mixpanel.com/pricing/) for more details. - - Customers who have not purchased the Experiment add-on will be able to create up to 3 experiments per project. Please note that creation is irreversible, so you would not be able to delete experiments. If you are interested in exploring the features fully, you can request a trial from your account team. - - ## Why Experiment? Experimentation helps you make data-driven product decisions by measuring the real impact of changes on user behavior. Mixpanel is an ideal place to run experiments because all your product analytics data is already here, providing you with comprehensive insights into how changes affect your users' journey. From 193a23482dd99fdf1b974457212d967e0fa6a7fa Mon Sep 17 00:00:00 2001 From: Myron Fung Date: Wed, 8 Jul 2026 14:39:14 -0700 Subject: [PATCH 03/10] move stick variant and delayed experience note to body. emphasize cohort limit as a note --- docs/featureflags.mdx | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-) diff --git a/docs/featureflags.mdx b/docs/featureflags.mdx index 520b19d4..7bfeb6d5 100644 --- a/docs/featureflags.mdx +++ b/docs/featureflags.mdx @@ -95,8 +95,15 @@ Control variant by is always set to non-sticky. This is to ensure this is the on Feature flags may consist of one or more rollout groups. Users are evaluated against each rollout group in order until one is found that the user qualifies for. -You can target 'All Users', or optionally target rollout of your feature flags to subsets of your user base rather than to all your users with [Cohorts](/docs/users/cohorts). -These are dynamic audiences of Mixpanel users based on user behavior or properties. Use these to target feature flags to specific subsets of your user base. For example, target only users who did 5 purchases in the last week. +You can target 'All Users', or optionally target rollout of your feature flags to subsets of your user base rather than to all your users with [Cohorts](/docs/users/cohorts). These are dynamic audiences of Mixpanel users based on user behavior or properties. Use these to target feature flags to specific subsets of your user base. For example, target only users who did 5 purchases in the last week. + +Cohorts used in feature flag targeting refresh on a periodic cadence (~every 2 hours). So once a user or group entity qualifies for a cohort, it can take up to 2 hours before they see a desired experience. + +If you want users or groups to continue seeing the same variant even if they disqualify from the cohort in the future, couple this with **Sticky Variants.** This will ensure they continue to see the new experience until the flag is turned off. + + + For optimal performance, remote evaluation supports cohorts of up to 20 million users per cohort. To target larger cohorts, contact your Mixpanel account team. + **Group Cohort Targeting** @@ -106,12 +113,6 @@ When the Variant Assignment Key is set to a group key (e.g., `company_id`, `acco - The cohort builder automatically corresponds to the the flag's assignment key. - Example: A flag with assignment key `company_id` lets you target a cohort like "Enterprise companies with >50 seats." - - - Cohorts used in feature flag targeting refresh on a periodic cadence (~every 2 hours). So once a user or group entity qualifies for a cohort, it can take up to 2 hours before they see a desired experience. - - If you want users or groups to continue seeing the same variant even if they disqualify from the cohort in the future, couple this with **Sticky Variants.** This will ensure they continue to see the new experience until the flag is turned off. - - For the optimal performance, cohorts used in feature flag targeting are officially limited to 20 million users. - - Note that cohort targeting for `device_id` assignment key flags builds cohorts against Users, not devices. Furthermore, since this targeting is against Users, it will only target authenticated Users, which could defeat the purpose of using an auth-insensitive `device_id` rollout. If you wish to do group cohort targeting against devices, make device a Data Group and then use group targeting against it. From 75e5731cbfdaff6550f3621c9ec17aeda1443ee3 Mon Sep 17 00:00:00 2001 From: Myron Fung Date: Wed, 8 Jul 2026 14:48:23 -0700 Subject: [PATCH 04/10] add number to FAQ for readability. and clearly explain what counts as a feature flag for billing --- docs/featureflags.mdx | 21 +++++++++++++-------- 1 file changed, 13 insertions(+), 8 deletions(-) diff --git a/docs/featureflags.mdx b/docs/featureflags.mdx index 7bfeb6d5..9c705f12 100644 --- a/docs/featureflags.mdx +++ b/docs/featureflags.mdx @@ -440,42 +440,47 @@ If the assignment key is _not_ present in the destination project, the operation Pricing Unit: Feature Flags are priced based on Feature Flag API Requests, and number of 'active' feature flags at any time -**What is a Feature Flag API Request?** + +**1. How are active feature flags are counted?** + +Mixpanel counts every active feature flag toward your plan capacity. It counts a feature flag shared across projects only once. Disabling or archiving a feature flag frees up your plan capacity + +**2. What is a Feature Flag API Request?** A feature flag API request is a call made by your application to a Feature Flag system’s API to determine **which all features** should be enabled or disabled, or which variant should be served to a user. All the active flags in the system are evaluated as part of this API request. -**How can I estimate Feature Flag API requests?** +**3. How can I estimate Feature Flag API requests?** In general, you can expect every user session to have 1 API request. When the session starts, all the flags for the user are fetched in the single API request. So monthly API requests is at minimum equivalent to the total user sessions that month. Few notes: - If you have multiple sdk inits each session, that would increase the count of API requests in a session - Considering above, general rule of thumb: would consider API requests to be 1.5 x User Sessions considering implementation challenges, or multiple sdk inits in a session -**How do I estimate API requests if implementing via server side sdk?** +**4. How do I estimate API requests if implementing via server side sdk?** - If you are using server side sdk, with only local evaluation: your API requests depend on a) how many server instances you have and b) how frequently you poll your servers - If you are using server side sdk, but with remote evaluation or leveraging cohorts: your API request estimation is similar to client-side sdk requests laid our above -**What happens if I go over my purchased Feature Flag API Request bucket?** +**5. What happens if I go over my purchased Feature Flag API Request bucket?** You can continue using Mixpanel Feature Flags, but you will be charged a higher rate for the overages. -**Is there any limit on number of feature flags per API request?** +**6. Is there any limit on number of feature flags per API request?** Yes, there is a limit depending on the plan you purchase. You can choose from 3 plans: to have upto 50, 200 or 1000 'active flags' per API request. - Once you reach this limit, no more flags will will be fetched as part of the API request until you disable some others or upgrade your plan - Active flags are flags which are marked 'Enabled'. Only these flags are counted under the 'active flag' limit - If you are on the 50 active flag limit, the first 50 flags based on the 'start' date are fetched -**Do I get double charged if I create the same feature flag across projects?** +**7. Do I get double charged if I create the same feature flag across projects?** No. We charge based on active 'feature flag keys'. If the same flag-key is used across projects, it will be charged only once. -**How can I monitor my account’s Feature Flag API requests consumption?** +**8. How can I monitor my account’s Feature Flag API requests consumption?** You can see your feature flag usage by going to Organization settings > Plan Details & Billing. -**If I purchase the feature flag add-on, do I still need to purchase the experimentation add-on as well?** +**9. If I purchase the feature flag add-on, do I still need to purchase the experimentation add-on as well?** Yes. Both these are separate add-ons to ensure we're compatible with your tech stack. So, if you are also looking to analyze experiments, please check our [experimentation offering.] (https://docs.mixpanel.com/docs/experiments) From cebff3d0be38a2174f5c34ba4b6fe03a600b96ae Mon Sep 17 00:00:00 2001 From: Myron Fung Date: Wed, 8 Jul 2026 14:52:34 -0700 Subject: [PATCH 05/10] re-add banner for experiment on pricing --- docs/experiments.mdx | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/docs/experiments.mdx b/docs/experiments.mdx index 511da086..e964233d 100644 --- a/docs/experiments.mdx +++ b/docs/experiments.mdx @@ -3,6 +3,12 @@ title: "Experiments: Measure the impact of a/b testing" sidebarTitle: "Experiments" --- + +The Experiment Report is a separately priced product add-on. It is currently only offered to those on the Enterprise Plan. See our pricing page for more details. + +Customers who have not purchased the Experiment add-on will be able to create up to 3 experiments per project. Please note that creation is irreversible, so you would not be able to delete experiments. If you are interested in exploring the features fully, you can request a trial from your account team. + + ## Why Experiment? Experimentation helps you make data-driven product decisions by measuring the real impact of changes on user behavior. Mixpanel is an ideal place to run experiments because all your product analytics data is already here, providing you with comprehensive insights into how changes affect your users' journey. From 14bb555c4e8359a9231194e122badce677163603 Mon Sep 17 00:00:00 2001 From: Myron Fung Date: Wed, 8 Jul 2026 14:54:07 -0700 Subject: [PATCH 06/10] readd relative URL --- docs/experiments.mdx | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/experiments.mdx b/docs/experiments.mdx index e964233d..5f1912ed 100644 --- a/docs/experiments.mdx +++ b/docs/experiments.mdx @@ -4,7 +4,7 @@ sidebarTitle: "Experiments" --- -The Experiment Report is a separately priced product add-on. It is currently only offered to those on the Enterprise Plan. See our pricing page for more details. +The Experiment Report is a separately priced product add-on. It is currently only offered to those on the Enterprise Plan. See our [pricing page](https://mixpanel.com/pricing/) for more details. Customers who have not purchased the Experiment add-on will be able to create up to 3 experiments per project. Please note that creation is irreversible, so you would not be able to delete experiments. If you are interested in exploring the features fully, you can request a trial from your account team. From e2c5e980698b3379e37830b9fa16d9a0f3da6e00 Mon Sep 17 00:00:00 2001 From: Myron Fung Date: Wed, 8 Jul 2026 14:56:03 -0700 Subject: [PATCH 07/10] add faq about FF being included with XP --- docs/featureflags.mdx | 11 ++++------- 1 file changed, 4 insertions(+), 7 deletions(-) diff --git a/docs/featureflags.mdx b/docs/featureflags.mdx index 9c705f12..eb6c323c 100644 --- a/docs/featureflags.mdx +++ b/docs/featureflags.mdx @@ -3,10 +3,6 @@ title: "Feature Flags: Rollout with precision control" sidebarTitle: "Feature Flags" --- - - Feature Flags is a separately priced product add-on. It is currently only offered to those on the Enterprise Plan. See our [pricing page](https://mixpanel.com/pricing/) for more details. - - ## Overview **Feature Flags** let you control who sees a feature and when. Use them to: @@ -441,7 +437,7 @@ If the assignment key is _not_ present in the destination project, the operation Pricing Unit: Feature Flags are priced based on Feature Flag API Requests, and number of 'active' feature flags at any time -**1. How are active feature flags are counted?** +**1. How are feature flags are counted for billing?** Mixpanel counts every active feature flag toward your plan capacity. It counts a feature flag shared across projects only once. Disabling or archiving a feature flag frees up your plan capacity @@ -480,8 +476,9 @@ No. We charge based on active 'feature flag keys'. If the same flag-key is used You can see your feature flag usage by going to Organization settings > Plan Details & Billing. -**9. If I purchase the feature flag add-on, do I still need to purchase the experimentation add-on as well?** +**9. If I purchase the Experiment add-on, do I need to purchase the Feature Flag add-on as well?** + +Feature flags are included with the Experimentation add-on and are not billed or metered separately. -Yes. Both these are separate add-ons to ensure we're compatible with your tech stack. So, if you are also looking to analyze experiments, please check our [experimentation offering.] (https://docs.mixpanel.com/docs/experiments) From e14bcc3e4ee5389f1d2660abad1af9d478930971 Mon Sep 17 00:00:00 2001 From: myronkaifung <97630035+myronkaifung@users.noreply.github.com> Date: Fri, 10 Jul 2026 13:40:19 -0700 Subject: [PATCH 08/10] Update featureflags.mdx --- docs/featureflags.mdx | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/docs/featureflags.mdx b/docs/featureflags.mdx index eb6c323c..9074fc5c 100644 --- a/docs/featureflags.mdx +++ b/docs/featureflags.mdx @@ -437,9 +437,9 @@ If the assignment key is _not_ present in the destination project, the operation Pricing Unit: Feature Flags are priced based on Feature Flag API Requests, and number of 'active' feature flags at any time -**1. How are feature flags are counted for billing?** +**1. How are feature flags counted for billing?** -Mixpanel counts every active feature flag toward your plan capacity. It counts a feature flag shared across projects only once. Disabling or archiving a feature flag frees up your plan capacity +Mixpanel counts every active feature flag toward your plan capacity. Specifically, it is defined as a feature flag having the same flag key across projects within the organization. If the same feature flag key is used in different projects in your organization, it will only be counted once. Disabling or archiving a feature flag frees up your plan capacity **2. What is a Feature Flag API Request?** From 8ef0a5c3cdb337428e5a41b2faec62f2ea341ebc Mon Sep 17 00:00:00 2001 From: Myron Fung Date: Thu, 16 Jul 2026 13:58:21 -0700 Subject: [PATCH 09/10] remove reference to 20M size limit (not true) --- docs/users/cohorts.mdx | 3 --- 1 file changed, 3 deletions(-) diff --git a/docs/users/cohorts.mdx b/docs/users/cohorts.mdx index 4ec084e7..4c01a564 100644 --- a/docs/users/cohorts.mdx +++ b/docs/users/cohorts.mdx @@ -185,9 +185,6 @@ You can delete any Cohorts from your project by navigating to **Lexicon > Cohort ## FAQ -### What is the cohort size limit? -Mixpanel supports up to 20 million users per cohort. - ### How can I create cohorts based on dropped-off or retained users? There are two ways to create such cohorts. From 58b1f043751e4ec7b7db9a9aeaa78679deaa7c02 Mon Sep 17 00:00:00 2001 From: Myron Fung Date: Thu, 16 Jul 2026 14:03:36 -0700 Subject: [PATCH 10/10] add internal note about modifying language --- docs/featureflags.mdx | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/docs/featureflags.mdx b/docs/featureflags.mdx index 05918e18..747a89fa 100644 --- a/docs/featureflags.mdx +++ b/docs/featureflags.mdx @@ -97,6 +97,12 @@ Cohorts used in feature flag targeting refresh on a periodic cadence (~every 2 h If you want users or groups to continue seeing the same variant even if they disqualify from the cohort in the future, couple this with **Sticky Variants.** This will ensure they continue to see the new experience until the flag is turned off. +{/* ============================================================ */} +{/* ⚠️ DO NOT REMOVE OR MODIFY the 20M user cohort limit below. */} +{/* We are contractually/legally required to maintain it. */} +{/* Before any change, confirm with the PM for Feature Flags */} +{/* AND Legal (contract implication review). — Myron, 2026-07-16 */} +{/* ============================================================ */} For optimal performance, remote evaluation supports cohorts of up to 20 million users per cohort. To target larger cohorts, contact your Mixpanel account team.