Stop chasing noise
Most metric changes mean nothing. We show you the ones that do.
Mostly Stable brings statistical process control to your product metrics — learning the normal range and flagging only the moves that break it, so your team stops investigating routine variation.
One project free forever · connect your first metric in 5 minutes
How to read it
The shaded band is the range your metric normally moves within. As long as points stay inside it, nothing has really changed — that's just routine variation.
The highlighted point breaks outside the band. That's a real signal worth investigating — and the only thing you need to act on.
What is Mostly Stable?
A metric monitor that knows the difference between a real shift and normal day-to-day noise.
Every metric you track — signups, revenue, latency, churn — moves up and down a little every day for no real reason. Mostly Stable applies statistical process control — the same control-chart method factories have used for decades to tell a real defect from random variation — to your product metrics. It learns the band your metric normally lives in and flags the moment something genuinely breaks the pattern, all simple enough to read at a glance.
The method behind it
We compute control limits from your own history — the statistically expected range of normal variation. A point inside the limits is common-cause noise and safe to ignore; a point outside is a special-cause signal worth investigating. That single distinction is what stops your team from reacting to every random wiggle.
How it works
From a raw stream of numbers to a clear signal
Three steps to knowing exactly when to pay attention.
Connect your metric
Pull in a CSV, Google Sheet, Mixpanel, or Amplitude source. Mostly Stable starts charting immediately — no formulas, no setup wizard.
We calculate your control limits
Using statistical process control, we derive the expected range from your own history and draw those control limits right on the chart — the band your metric should stay inside.
Get alerted only on real signals
When a point breaks the pattern in a way that isn't just noise, we flag it. Everything else stays quiet — no more chasing routine wiggles.
Why it matters
Less time reacting, more time on what changed
- Stop investigating noise
- Skip the standups spent debating a 3% dip that means nothing.
- Catch real shifts early
- See a sustained change the moment it breaks trend, not weeks later.
- No statistics degree required
- The math runs in the background. You just read inside or outside the band.
Everything you're wondering
Do I need a stats background? +−
No. We calculate the control limits for you. You just read whether a point is inside the normal range or worth a look.
How long does setup take? +−
About five minutes. Connect a CSV, Google Sheet, Mixpanel, or Amplitude source and your first chart builds automatically.
What if I just want to try it? +−
One project is free forever with no credit card. Upgrade only when your team needs more.
What changes on a team plan
Free is great for solo analysts. Growth is built for teams making decisions together.
- Projects 1 10
- Shared Slack alerts when a metric signals
- Comment threads on each chart
- Weekly digest of what actually moved
- Priority onboarding support
Free
For solo analysts
$0/mo
- 1 project
- CSV, Sheets, Mixpanel, Amplitude
- Unlimited charts
- No credit card, free forever
Growth
For teams
$49/mo
- 10 projects
- Shared Slack alerts on signals
- Comment threads on charts
- Weekly 'what moved' digest
Spend your reviews on what actually changed
Sign up for freeOne project free forever · 5-minute setup