For Engineering leaders

Catch regressions before users tweet.

Your monitoring watches error rates. Northbeam watches behaviour, which is where a bad deploy shows up first.

The problem

Catch regressions before users tweet.

A deploy that breaks nothing technically can still stop people finishing onboarding. Error rates stay flat, dashboards stay green, and the first signal is a support ticket. Northbeam alerts on the behaviour itself, with the cohort and the two most likely causes attached to the message.

“Yesterday Northbeam caught a 14% drop in onboarding completion two hours before our oncall did.”

YTYannis TheodorouCEO, Plum Fintech
#product-alerts
nb
Northbeam APP 08:14

Onboarding completion dropped 14% in the last 2 hours. Cohort: signups from the pricing page.

completion %, hourly
Start here

Three pulses engineering leaders clone first

Every one of these is in the public pulse library. Clone it, point it at your warehouse, edit the SQL if you disagree.

Pulse

p95 page load by route

Latency where it affects a funnel step, not averaged across the whole app.

Pulse

Failed signups last 24h

Attempts that never became accounts, grouped by the step they died on.

Pulse

Error rate by build

Behaviour deltas lined up against your deploy markers.

Other teams

Northbeam is used by more than engineering leaders

Twelve places in cohort #2.

Founder-led setup, a private Slack channel with Maya and Lior, and founding pricing for life.