Key takeaways
- Mixpanel counts events, not storage: read them in Organization Settings > Plan Details & Billing and the Data Usage tab, which needs the Owner or Billing Admin role.
- The Raw Event Export API returns every event with its properties as JSONL, one date range at a time, at 60 requests an hour per project.
- Data Pipelines is a paid add-on that sends events to a warehouse on a schedule, but its backfill reaches back at most 6 months.
- You don't need to export anything to apply: have your event count or size, years of history and seat count ready, and the data vendor anonymizes the records before any AI lab sees them.
01
What is Mixpanel data, and why do AI labs and data vendors buy it?
The short answer
Mixpanel data is a company's record of what people do inside its app or website: every sign-up, tap, page view and purchase, stored as a timestamped event. AI labs (companies that build and train AI models) and data vendors (companies that anonymize business records and package them for those labs) buy a copy because a long event stream shows, step by step, how real users adopt features, where they give up and what they do after each product release.
What a Mixpanel project records
Mixpanel is a product analytics tool, meaning software that records what users do inside a product so the company can see which features people use. The product's code sends Mixpanel an event each time a user does something, such as "Signed Up" or "Added to Cart".
Each event carries an ID for the person (the distinct ID), the time it happened and properties, which are extra details such as the device platform, the customer's plan or the page that sent them. Events are stored in a project, a separate container for one product, and a company can have several.
Why an event stream teaches an AI model something
An AI model learns by studying a very large number of examples and finding the patterns in them, and those examples are called training data. A single event says little, but millions of them in order show which paths end in a purchase, where people abandon a sign-up and how usage changes after a new feature ships.
AI labs now want to train AI agents, software that carries out a job step by step, and an agent learns a job from records of that job being done. A data vendor can turn anonymized event sequences into worked examples of real user behavior. The full mechanism is in AI training data: why AI companies want your company's data.
How Mixpanel fits into a dataset
Mixpanel is usually one of several sources in a company's dataset, which may also include web analytics such as Google Analytics, customer records and support tickets. Companies are paid $100K to $2M or more per dataset, according to micro1, the data vendor we typically refer companies to. The price covers the whole dataset, not each source on its own, and what your data is worth explains what moves it.
02
Where does Mixpanel show how much data you have?
The short answer
Mixpanel counts events, not storage. Open the gear icon, then Organization Settings > Plan Details & Billing for this billing period's event count, and the Data Usage tab for past periods. You need the Owner or Billing Admin role in the organization, and the only gigabyte figure comes from a raw export you run yourself.
How to read your event count
The steps below total the events Mixpanel has billed over the months it still shows. Each step names the menu you need.
- Sign in to Mixpanel as an organization Owner or Billing Admin. Mixpanel's pricing and plans page names these two roles as the ones that can see billing.
- Click the gear icon to open Organization Settings, then choose the Plan Details & Billing tab.
- Read the event usage for the current billing period on the right side of the page.
- Open the Data Usage tab and choose View Historical Usage to see earlier billing periods.
- Add up the monthly event counts for every month still listed. The sum is your event total.
What the free allowance means for the number
Free plans cover up to 1 million events a month, and a Growth plan bought after February 2025 does not charge for its first million events. A small company can therefore have a low monthly count and still hold a long history, so add the months together instead of reading one month alone.
Why the total can differ from the real history
Mixpanel counts an event in the month it reaches Mixpanel's servers, not the month it happened. A company that imported two years of old events in one afternoon will see them all in that one month.
Usage history also stops where retention stops, since Mixpanel deletes events once they pass the retention period (the export section gives the figures). Events are also not storage: Mixpanel never reports gigabytes, so a size in GB exists only after a raw export.
03
What should you know about your Mixpanel data before you apply?
The short answer
Have three facts ready: the size of your data or, without an export, your event count, how many years you have used Mixpanel and how many people have access to the organization. They go into the data manifest, the written summary of which apps hold your records, how many years each covers and roughly how much. You don't need to run an export to find them.
The size or event count
If someone has already run a raw export, add up the sizes of the JSONL files it produced (JSONL is a file format with one event per line) and give the total in GB. Without an export, give the event total from Data Usage and say plainly that the number counts events, not bytes.
The years of history
Years in use run from the first month you tracked events. Mixpanel's default retention period is two years for projects created since September 1, 2025, so older events may already be gone unless your contract keeps them longer, as Mixpanel's retention changelog explains.
The number of seats
Seats are the people who can sign in to your Mixpanel organization. Count everyone with access, including analysts, product managers and developers.
What you don't need to do first
You don't need to export anything to apply, and you don't need to clean or scrub an export. The data vendor anonymizes the records before anyone else sees them, and nothing moves until a lawyer-reviewed license is signed. The application, the intro call and each later step are laid out in how to sell your company's data.
04
How do you export your full Mixpanel history?
The short answer
Mixpanel offers two ways to take out every event: the Raw Event Export API, which a technical person runs project by project in date slices, and the Data Pipelines add-on, which sends events to a data warehouse on a schedule. Both return only events inside the retention period, and the API is the one that reaches the oldest of them.
What you need before you start
The API needs a service account (a login made for software, not a person) that has access to the project, or the project's secret key. Someone who can run a short script, usually a developer or data analyst, does the work. Mixpanel's export guide says the API works on Free, Growth and Enterprise plans, although a few projects on older Free plans lack it and need Mixpanel's support.
Raw Event Export API: the steps
- Create a service account with access to the project, or find the project secret. With a service account, every request also names the project's ID.
- Choose the server for the project's region. Most projects use data.mixpanel.com, projects stored in the EU use data-eu.mixpanel.com and projects stored in India use data-in.mixpanel.com.
- Split the history into date ranges, such as one month each, and send one request per range with a from_date and a to_date in the form year-month-day. Both dates are inclusive.
- Save each response as its own JSONL file. Each line of a file is one event with all of its properties.
- Repeat for every project. Then add up the sizes of all the files to get the figure in GB.
The command below shows one request for one month. A developer runs it in a terminal, and it asks Mixpanel for every event in January 2025 and saves the reply to a file; the capitalized words are placeholders for your own values.
curl --get "https://data.mixpanel.com/api/2.0/export" \
-u "SERVICE_ACCOUNT_USERNAME:SERVICE_ACCOUNT_SECRET" \
--data-urlencode "project_id=PROJECT_ID" \
--data-urlencode "from_date=2025-01-01" \
--data-urlencode "to_date=2025-01-31" \
-o events_2025-01.jsonlThe Raw Event Export reference allows 60 requests per hour, 3 per second and 100 running at once, and it answers with an error code 429 when you go over. Waiting a minute between requests keeps a script inside the hourly limit, and the export is not deduplicated, so the same event can appear twice.
Data Pipelines: the other route
Data Pipelines is a paid add-on for Growth and Enterprise plans, with a 30-day free trial. It sends events on a schedule to Snowflake, BigQuery, Databricks, Redshift Spectrum, Amazon S3, Google Cloud Storage or Azure Blob Storage, all of which are places that store large amounts of data.
A pipeline's backfill, which copies events from before the pipeline existed, can start at most 6 months in the past, and a trial pipeline backfills only one day. Large backfills can take weeks. Mixpanel's Data Pipelines page has the details, so use the API for anything older.
How long Mixpanel keeps events
Mixpanel deletes events two years after the event date. Projects created before September 1, 2025 keep events for five years, until the project changes plan or moves to the Free plan, and then the two-year period applies.
Older history is only reachable if the contract keeps it longer, and Mixpanel's account team handles that request. Export early, since an event that ages out cannot be recovered.
What a Mixpanel export leaves out
The raw event export contains events and nothing else. User profiles, the stored details about each person, come from a separate call to the Engage API, which allows 60 requests per hour and 5 at once.
CSV downloads from reports are a different tool, limited to 10,000 rows each, with a 50,000-row ceiling for certain Insights views on Enterprise, so they cannot hold a full history. Timestamps in the raw export are in UTC for projects created in 2023 or later and in the project's own time zone for older ones.
How long an export takes
The time depends on how many events the project holds. With only 60 requests an hour, a large project exported in daily or monthly slices can take hours or days.
Sources: Mixpanel Docs, Export all your data, Raw Event Export API, Pricing and plans, Data Pipelines and the data retention changelog. Steps checked against these pages in November 2026.
05
What stays private when Mixpanel data is licensed?
The short answer
The company decides which projects are in scope, and the data vendor removes or replaces person IDs, profile fields and personal event properties before an AI lab sees anything. The company keeps ownership of its event history, and the license covers defined records for a defined use.
You choose what is in scope
A company can license the events of one product and leave other projects out. It can also decide that user profiles stay out of the dataset entirely. The data vendor deletes records too sensitive to clean safely instead of anonymizing them.
Person IDs and personal details are replaced first
Anonymization means replacing every detail that identifies a person, such as an ID or an email address, with a stand-in that is the same every time it appears. The sequence of actions still reads in order, so a lab can learn from it, but it no longer says who acted. The diagram below shows one event before and after.
The method, including what is deleted outright and how you review samples, is in how company data is anonymized for AI training.
Ownership and payment
Licensing gives the buyer limited rights to a copy, while the company keeps ownership of the originals. The terms that matter, such as exclusivity, permitted use and deletion, are covered in data licensing. micro1 pays in full within 30 days of accepting the data, usually as a single payment.
Questions
Frequently asked questions
Can the Data Pipelines add-on export all of our Mixpanel history?
No. A pipeline can copy history from at most 6 months back, and a trial pipeline copies only one day. For anything older, run the Raw Event Export API, within your retention period.
How long does Mixpanel keep our events?
Two years by default, counted from the event date. Projects created before September 1, 2025 keep events for five years until the project changes plan or moves to Free, and longer retention needs Mixpanel's account team.
Can any team member see the event totals in Plan Details & Billing?
No. Plan Details & Billing sits in Organization Settings and needs the organization Owner or Billing Admin role, so ask one of them if your own role does not show the tab.
How long does a full Mixpanel export take?
It depends on volume. The API allows 60 requests an hour, so a large project exported in daily or monthly slices can take hours or days.
Do we need to export Mixpanel before we apply?
No. The application takes about two minutes and shares no data. You only need your approximate event count or size, your years of history and your seat count for the intro call, and you can read the event count from Data Usage. The raw export is needed later, when the records are transferred after you sign the license.
Are Mixpanel user profiles included in the event export?
No. The Raw Event Export API returns events only. User profiles come from the Engage API, and a company can choose to leave them out of the dataset.
What is Mixpanel data worth to AI labs and data vendors?
Mixpanel history is priced as part of a whole dataset rather than on its own. Companies are paid $100K to $2M or more per dataset, according to micro1, and the price depends on the years of history and how completely the records show real use. What your data is worth explains the factors.
Bottom line
Know what is in Mixpanel, then see what it is worth
You don't need to export anything to apply. Apply in two minutes, describe your records on the intro call, and micro1, the data vendor we typically refer companies to, prices the whole dataset and sends an offer 1–2 days later.
Not legal, tax or financial advice. Figures come from micro1's published pages and public reporting as of November 2026.