I think there's a lot more overlaps between the observability space and modern product analytics than ppl realize

the "standard" analytics workflow involves 1) framing a question, 2) figuring out the data you need/have, & 3) "answering the question" or making it answerable (1/n)

..there's a lot of crunching and transforming and prep. But the approach isn't optimized for asking and answering *new* questions and/or easily exploring the data.

Contrast this with a team I was chatting with recently ... (2/n)

Paraphrasing ... "we develop good habits around measurement and telemetry, but don't get too bogged down in the questions. If we do the former correctly, we'll be able to do the latter 95% of the time"

In my mind this is the "self service" that matters...(3/n)

A team with awareness of the customer domain and the product can figure out what to instrument *without* worrying about questions ahead of time. It takes a bit of practice but it is completely doable...

The important point is that we worry about crunching at query time...(4/n)

This is a different approach, and one with some parallels in observability. The observability challenge has extra layers of complexity -- way more data/dimensions - but the desire to observe -> frame questions -> explore - frame questions/hypotheses --> explore is similar (5/end)