Live numbers

The metrics we run on, in public

The same definitions, refresh cadence, and units we use in our own war rooms. Refreshed continuously where possible, never adjusted for the audience, every metric linked to the surface that owns it.

Metrics published live

20

Across 11 areas of operation

Refresh ≤ 15 min

16

Including reliability and security

Open negative signals

0

Zero hidden, everything linked

Definitions on

/open-metrics

Plain language, with formulas

Spend on platform

MetricCurrent
Trailing 30d ad spend processed€247.4M
Active workspaces342

Recovered budget

MetricCurrent
Recovered to customers (trailing 30d)€3.84M

Pacing

MetricCurrent
Median pacing accuracy across workspaces98.6%
Pacing decisions / s peak (last 24h)1,284

Reliability

MetricCurrent
Composite uptime (trailing 30d)99.981%
Open SEV-1 incidents0
Open SEV-2 incidents1

Forecast

MetricCurrent
Median forecast MAPE (trailing 30d)3.7%

Security

MetricCurrent
Privileged standing access0
Open critical CVEs in production0
Tenant-isolation tests passed (trailing 24h)2,924 / 2,924

Treasury

MetricCurrent
Payouts awaiting quorum11
FX exposure hedged94.2%

People

MetricCurrent
On-call paged in last 24h3

Customers

MetricCurrent
NPS (trailing 90d, verified responders)67
Open escalations to ombudsman0

Engineering

MetricCurrent
Release train deployments (trailing 7d)84
Error budget burn (trailing 7d)11.2%

Audit trail

MetricCurrent
Audit events appended (trailing 24h)≈ 244M

Why this page exists

  • We do not believe trust is built by annual reports alone. Live metrics make claims falsifiable in real time.
  • Numbers are unadjusted. If something looks bad, it is bad; the linked surface tells you what we are doing about it.
  • Definitions never change quietly. Methodology updates are pre-announced on /change-notice-policy.