Not promised.Proven.

What happens

Verdict connects read-only to your tools before the first workshop, so we judge against data and not against memory. Workshops run one discipline at a time with the people who do the work. Their managers stay out of the room, because what actually happens changes when they are in it.

We ask to be shown, not told: the last release, the last incident, the oldest branch. One real piece of work is followed from request to production with every queue timed. Flow efficiency, active time divided by elapsed time, is usually below 15% when we first measure it. Anything only an interview supports is labelled a hypothesis until the data confirms it.

Seven kinds of bottleneck

• Unrefined work
• Review latency
• Build and test friction
• Environment contention
• Manual change control
• Rework
• Dependency waits

Every bottleneck in the report carries one of these types, its measured cost and the capability that fixes it, so reports compare across teams, customers and engagements.

What happens

Reveal starts on day one of Profile, when Verdict connects, and its four-week baseline runs underneath the workshops. Within a day the Four Keys and the flow metrics begin to fill from your own history, and you watch the scorecard build in real time.

Your sponsor sees the scorecard first. No leader should learn their tier in a room full of their engineers. Anything Verdict cannot connect to is a finding too. A team that cannot produce a deployment timestamp has a release problem before it has a metrics problem.

What happens

We sequence the work by its effect on the bottlenecks against the effort, with prerequisites drawn. Progressive delivery does not come before infrastructure as code. Mutation testing does not come before there is a test suite worth mutating.

The work runs in waves: one discipline, one team, four to eight weeks. Wave one always holds a quick win that shows in Verdict within a month, such as batch-size limits, review turnaround or quarantining flaky tests. What is deliberately out of scope is written down, with the trigger that would bring it in.

What happens

In an existing product, most teams already do most of the right things, informally. The first move is to turn them into gates: the review that already happens becomes required, the build that already runs becomes a required check. That costs days and changes no tool.

Then every wave pairs our engineer with one of yours who will own the capability. Gates go in enforcing from day one, never advisory, because an advisory gate teaches everyone that gates are advisory. Old paths are deleted, not disabled.

What happens

Every quarter the report is generated from Verdict. We do not write it. If a number needs explaining, we explain it in the meeting and leave it as it is in the document. Customers keep old reports and check.

Where time is converted into money, the assumption sits next to the number. An honest smaller number beats an impressive one your CFO takes apart.

What happens

At handover we switch Verdict to steady state together and watch the first drift alert reach your team. From then on your team owns it.

Every metric has a rolling baseline, usually the trailing 90-day median, and drift rules, such as cycle time up more than 25% over two sprints, or change failure rate above the tier ceiling for two weeks. When a rule fires, the alert names the discipline, the likely gate and the playbook. A tier drop is an event, not a footnote.

We read the portfolio view with you for thirty minutes a month. An alert open for more than thirty days comes up in that read. Twice a year we run a one-hour guild session.