Skip to main content

Exporting Pulse Reports

How to Export Pulse Data

Finding Export Report

You can export your Pulse Results for Feature Gates and Experiments. Simply navigate to the relevant "Pulse Results" page, and click "Export Report". Exporting results can take up to 10 minutes. A notification and an email will be sent when the report is ready, and a link will be available under under Project Settings -> Reports.

Export Pulse Report Menu

Report Types

There are three types of export:

  1. Exposures - A table of all exposed users and their first exposures. This is useful for joining on your own internal data, and running custom queries within your own data warehouse. This can also be used to verify who was in the experiment, what group they were assigned to, and when they were first exposed (around 1-25MB). This will contain:

    1. <experiment\>_first_exposures.csv - contains a list of users and their first exposure to the experiment.
  2. Pulse Summary - This provides precomputed summary experimental data for all metrics and test groups including everything that's visible on Pulse (around 10-100 kb). This will contain:

    1. <experiment\>_pulse_summary.csv - contains Pulse aggregate metrics computed over the duration of the experiment.
  3. Raw Data - This provides raw exposures and metrics data at the user-day level. This is best used for manually inspecting data, or recomputing your own statistics (around 10MB-1GB). This will contain:

    1. <experiment\>_first_exposures.csv - contains a list of users and their first exposure to the experiment. If this is the only file you are interested in, you can get this by exporting an "Exposures" report which will be much smaller in size.
    2. <experiment\>_user_metrics.csv - contains a list of experimental users, and their calculated metrics for each day they were enrolled in the experiment.

Pulse Summary File Description - For Feature Gates

Column NameDescription
nameName of the Experiment or Feature Gate
ruleName of the Feature Gate Rule.
metric_typeCategory of the metric. Different metric_types are computed differently, including how they're computed in Pulse.
metric_nameThe name of the metric. For event metrics, this is the name of the event.
metric_dimensionThe subcategory of the metric. For example, if you log value in LogEvent, then value will show up as a subdimension. dimension = !statsig_topline indicates that this row reflects an aggregate across all dimensions.
start_dateThe start date for this measurement
end_dateThe end date for this measurement
test_unitsThe number of users in the test group
test_meanThe average value of this metric across test users (or participating units when applicable)
test_stderrThe standard error for the estiamte of the mean for test users. This can be used to compute confidence intervals.
ctrl_unitsThe number of users in the control group
ctrl_meanThe average value of this metric across control users (or participating units when applicable)
ctrl_stderrThe standard error for the estiamte of the mean for control users. This can be used to compute confidence intervals.
abs_deltaThe absolute difference between the test and control mean (test_mean - ctrl_mean)
abs_stderrThe estimated standard error of abs_delta
rel_deltaThe relative difference between test and control mean, sometimes referred to as lift (test_mean - ctrl_mean)/ctrl_mean
rel_stderrThe estimated standard error of rel_delta (abs_delta/ctrl_mean)
z_scoreThe calculated Z-score

Pulse Summary File Description - For Experiments

Column NameDescription
nameName of the Experiment or Feature Gate
ruleName of the Feature Gate Rule.
experiment_groupThe group of users for which this metric is computed for. For a feature gate, this is pass/fail. For an experiment, this is the variant name.
metric_typeCategory of the metric. Different metric_types are computed differently, including how they're computed in Pulse.
metric_nameThe name of the metric. For event metrics, this is the name of the event.
metric_dimensionThe subcategory of the metric. For example, if you log value in LogEvent, then value will show up as a subdimension. dimension = !statsig_topline indicates that this row reflects an aggregate across all dimensions.
start_dateThe start date for this measurement
end_dateThe end date for this measurement
unitsThe number of users included in this metric estimate.
meanThe average value of this metric across units (or participating units when applicable)
stderrThe standard error for the estiamte of the mean. This can be used to compute confidence intervals.

First Exposures File Description

Column NameDescription
user_id / stable_id / <custom_id>Refers to the unit identifier used in the experiment
nameThe name of the gate/experiment
ruleFor gates, this refers to the rule name
experiment_groupThe group the user was assigned to
first_exposure_utcThe UTC timestamp when the user was first assigned to the experiment
first_exposure_pst_dateThe date in PST when the user was first assigned to the experiment
as_of_pst_dateThe date this data was generated
user_dimensionsJSON-formatted key-value pairs describing the user's attributes at the time of first exposure

Unit Metrics File Description

Column NameDescription
pst_dsThe 24hr window the the data refers to. All dates are anchored from 12:00a -> 11:59p PST.
user_id / stable_idRefers to the unit identifier used in the experiment
metric_typeThe category of the metric
metric_nameThe name of the metric
metric_dimensionThe name of the metric dimension. '!statsig_topline' is the overall metric with no slicing.
metric_valueThe numeric value of the metric
numeratorFor some metrics, we track the numerator
denominatorFor some metrics, we track the denominator