About
PostObs started from a simple frustration: every product team I know is drowning in behavioral data they don't have time to read. Dashboards tell you that something happened. Replays show you what happened. Almost nothing tells you why.
That gap — from observation to explanation — is what we're building toward: an engine that reads your behavioral data and tells you what the patterns mean, where they probably came from, and what to try next. Useful to PMs who decide the product, and to engineers who implement the fixes.
What we're trying to learn
Right now PostObs is a smoke test. The goal isn't to ship a finished product — it's to answer three questions, honestly:
- Do growth PMs actually want this enough to share their data?
- Which kinds of insights make someone say “huh, I want this every week”?
- What price feels obvious vs. what feels painful?
Until we have clean answers, we're generating diagnoses by hand. It scales worse, but it teaches us more. Every email you get back was written by a human reading your data — not a hallucinating model.
How we're different
Replay tools
Hotjar, FullStory — show you what happened, frame by frame. You still do the thinking.
Analytics
Amplitude, Mixpanel, Heap, PostHog — they store the numbers. You still build the funnels and stare at the charts hoping the answer falls out.
PostObs
Reads exports from any of the above and tells you the why. Diagnosis, not dashboard — plus the instrumentation gaps you'd need to close to make the next diagnosis sharper.
Where we're going
Today PostObs is laser-focused on one slice of the user journey: onboarding. We start there because it's where the gap between “the dashboard says X% drop here” and “I know what to do about it” is widest, and where the data is cleanest to reason over.
The long arc is bigger. Every product generates behavioral data across the full journey — the first ad click, the landing page, signup, activation, the subscription decision, and everything that happens after. Almost none of it gets read in depth. PMs glance at funnels. Engineers ship more telemetry. Nobody actually understands what users were thinking at each step.
The end goal is an engine that reads that full arc — clickstream, funnel events, and signals from session replays like rage clicks, dead clicks, hesitation, and scroll depth — and explains it the way a senior PM would: what happened, where users hesitated, what they were probably trying to do, and what would unblock them.
The wedge is onboarding. The vision is every user journey.
Who's behind this
Built by Raphael Balbous. I'm the founding engineer at Koncile and have spent the last few years watching small teams struggle to get clean signal out of their product data. PostObs is the thing I kept wishing existed.
If you want to talk shop — book a 30-min call.
Want to try it on your own funnel? Get a free diagnosis.