Date of slack thread: 5/14/24
Anonymous: I’m currently composing several custom metrics and I wanted to share some feedback in terms of ease of use. For additional context, I’ve been migrating from Mixpanel, and writing relatively straightforward custom metrics is much more difficult in Statsig.
I’m trying to calculate a revenue metric which is the summation of two fields in the same event:
mile_transaction(event) = {
transaction_type: "shop" / "voucher",
payout_amount: 123,
spend_amount: 456
}
In mixpanel, I can simply write the custom property out with IFS.
In statsig, I would need to define payout_amount as a metric, and spend_amount as a metric and then create a metric that allows me to sum them. (Thankfully this should work for my case)
A few immediate things that don’t seem to work out of the box. Summing is the only way to go.
(3rd photo) rollup mode shows: total experiment and custom attribution window, mismatched with tool tip.
Cooper Reid (Statsig): You’re taking the correct approach today by creating 2 metrics and summing those. Most experimentation platforms don’t even allow for multi-metric aggregations, let alone applying metric filters and selecting specific metadata fields for their aggregations.
I think that formula expression is a useful feature for this use-case and we’ll consider your feedback.
Anonymous: Thanks! Mixpanel definitely does a great job at this. Really flexible product analytics. Or… maybe something like the meta del2oid tool.