Skip to content

PRD - Metric store updates


Objective -

Reducing analytics effort required to understand experiment reads accurately by relying on experiment metric store
Export accurate metric reads for discovery / browsing experience experiments which are a function of Android / iOS app version

Current system context -

Metric computation -
Experiment store currently computes metrics for the last 10 app versions to reduce compute cardinality
Experiment dashboard -
Experiment reads are then populated in the following
Experiment ID, Experiment start and end date are kept configurable updates the metrics dynamically basis these inputs

Proposed changes -

Abacus to input the Android app version to be specified by teams while setting up experiments - This would be a non mandatory input
For discovery / BEx experiments, This will be a mandatory input to be passed on for all features tied with some app version update (Non SDUI driven experiments)
Abacus team will soon roll out a feature to specify the app version above which teams want to run experiments - The audience allocation would however happen only post the app adoption for the specified app version exceeds 50%
Experiment metric store to fetch the app version signal specified for such experiments from Abacus and publish metrics from app version >= app version mentioned as experiment input on Abacus
The metrics computed with the filter for app version to be directly indexed in the reads for experiment dashboard

Instrumentation -

P1 ask - DP team to instrument the app version above which metric has been computed to QC the reads (Can be parked off for now)
Want to print your doc?
This is not the way.
Try clicking the ··· in the right corner or using a keyboard shortcut (
CtrlP
) instead.