Northbench builds, ships, and runs one standardized workflow for one chosen vertical. The point is standardization, not novelty — six to twelve clients in the same vertical, the same lead-routing shape, the same booking flow, the same escalation rules. The unusual part of the proposition is reporting: every month, every retainer receives an outcome report on the same five metrics, and the same metrics are aggregated here, in public, every year.
Northbench · 2026 annual report
State of the Workflow.
Twelve months of aggregate outcomes from every Northbench retainer, anonymized into six cohorts across five verticals. No forecast, no synthetic data, no marquee client — just the numbers the workflows actually produced, measured in the time-tracking fields your operations lead already uses. Read it as the alternative to AI hype: things that shipped, that held a cadence, that you can defend in a monthly finance meeting.
Updated annually. Same five-vocabulary format every month for every retainer.
Loading aggregate figures…
What the numbers mean, in order
Four metrics. Four short stories. The numbers do the rest.
Each section below walks one of the four cards above, so a reader does not need to scroll back to match prose to data.
Hours saved is the metric operators ask about first. The aggregate is 11.4 hours per client per month, median across the eight-quarter trailing window. We count it from the workflow owner’s time-tracking field, not from inference — a Northbench retainer does not "credit" automation time, only the manual hours the workflow retired. Clients who started close to the baseline still end the year at the same number; the gain is cumulative, not novelty-driven.
Lead-response and error-rate are the two quality bars. The first-touch latency delta of minus 38 minutes is a median against each client’s own pre-Northbench baseline — small clients do not pull it either way, because the comparison is per-client, not per-cohort. The error-rate delta is the boring one: it falls slightly each month, accumulates over the year, and is reported against the same booking flow on every cohort. If a cohort introduced a new workflow mid-period, we re-base the metric from the month the workflow went live.
Retention is the proof. 96.2 percent of active retainer clients renew the next month, trailing twelve. The handful who exit do so on scope mismatch — a retainer trying to extend into a vertical Northbench does not run — which is the system working. Quiet churn, the other kind, is the one we guard against; the monthly outcome report exists in part to surface a quiet-churn signal before it becomes an off-ramp.
Loading the aggregate.
Take the next step
The same vocabulary, in your vertical.
The methodology is the audit. The audit is the first step on a retainer. Read each one in the order that fits the moment.
How we measure it
Methodology
The monthly outcome report format, the documented rollback procedure, and the human-review safeguards on consequential outbound messages — spelled out, not promised once.
Run it on your own workflow
Book a paid audit
Two weeks. A written brief on one chosen workflow in your vertical: what to automate, what to leave alone, what it should cost to run monthly. $250 deposit, fully refundable.