[mitchm]

This all started with a simple question: when metrics are down, how do you know part of your application isn’t broken for a subset of users?

Production for the frontend is the user’s device, and we all know things can go wrong without an explicit error. With the current state of frontend observability tools, how do you tell when something should have happened but didn’t?

Too bad we can’t just have our users run `it('opens checkout modal on click', …)`, right? What if we could..

Maybe the world doesn’t need more HTML attributes, but it’s my best attempt so far at improving frontend observability. I think the ends justify the means.

[mitchm]

I should add, this is the instrumentation layer. You still send all the events generated to your event collection backend.