Traces that survive a bad day
Every request keeps its own timeline, so an incident reads as a sequence of causes instead of a wall of red numbers.
1.4sMedian trace query
General-purposeMeridian
A landing page that has to sell something
Navigation, a hero built around the product's own surface, feature storytelling, social proof, pricing that changes with the billing period, an FAQ that opens, and a lead form that validates before it thanks you.
- Built from
- ButtonCardBadgeProgressAccordionFieldInputNativeSelectCalloutRating
- What the browser test drives
- Switch the billing period and watch the prices change; open an FAQ row; follow any call to action down to the walkthrough form; submit it empty, then with a broken address, then correctly.
- Appearance
- Cobalt with the technical mono display face, at the loaded (layered) depth.
Preview
Observability for teams on call
Know which deploy broke it, before the pager tells you
Meridian keeps the deploy, the trace and the alert in one timeline, so the first question of an incident already has an answer on screen.
No card, no collector to install, and a 40 ms p99 budget you can measure first.
production3 of 4 healthy
- healthyedge-routerdeployed 6m ago99.99%
- healthyingest-apideployed 1h ago99.94%
- degradedbilling-workerburn 2.4× budget98.20%
- healthysearch-indexdeployed 3h ago99.97%
Error budget this week2%
2,400services instrumented
31%shorter incidents
40msadded p99 overhead
Error budgets the team can see
Burn rate, remaining budget and the freeze date sit on one line, so the argument about reliability has a number in it.
0.02%Budget spent this week
Roll back from the alert
The alert carries the deploy it came from, so the fix is a button next to the evidence rather than a separate tab and a guess.
18sAlert to rollback
The freeze window did more for our release cadence than a year of process documents.
An alert arrives with the deploy that caused it attached. That one detail removed most of our triage.
We moved three teams onto it in a fortnight and nobody asked for the old dashboard back.
Priced per seat, on call or not
Anyone who can change an alert is a seat. Anyone reading an incident from a shared link never is.
Monthly
Starter
$0forever, for one service
One service, one person on call, and enough history to debug it.
- 7 days of trace retention
- 1 alert route
- Community support
Team
Most teams$24per seat / month
The whole on-call rotation, with budgets and a rollback that takes seconds.
- 90 days of retention
- Error budgets and freeze windows
- Unlimited alert routes
Scale
$79per seat / month
Multiple regions, an audit trail, and a support agreement with a number in it.
- 13 months of retention
- SSO and audit log
- 99.95% response agreement
Questions we get before the trial
Answers are short on purpose. Anything longer belongs in a conversation.
See it against your own traffic
We will point Meridian at one of your services, run it beside what you have, and leave you the comparison.
Interactive, not live: every action below changes local state in this page. No network request is made, no market data is fetched and no message is sent anywhere.