How it works
From observation to a measured improvement.
Nine stages, one workflow at a time. Each stage produces a record you can check, and nothing moves forward on an assumption that has not been written down.
The loop
Observe, map, baseline, compare, pilot, measure, learn.
The method in brief. The nine stages below are the same loop, with what happens at each step and what it asks of your team.
- 01ObserveWatch the work as it happens today, with the people who do it.
- 02MapDraw the real workflow: steps, handoffs, systems, exceptions.
- 03BaselineMeasure how long, how often and how costly — before changing anything.
- 04CompareWeigh a process change, plain software, AI, or doing nothing.
- 05PilotRun one small, approved change with stop conditions and a rollback.
- 06MeasureCompare against the baseline and say honestly what it shows.
- 07LearnKeep what worked, drop what didn't, document what can be repeated.
The nine stages
What happens, what you provide, what you receive
Stage 1: Understand the business
A short conversation about how the business makes money, where it hurts and who owns the problem.
From you
A leader who owns the workflow and can decide.
From us
Questions, not a pitch. We agree which single workflow is worth studying first.
Stage 2: Observe the workflow
Interviews and shadowing with the people who do the work, on real cases.
From you
Access to the team and permission to observe.
From us
Structured sessions covering triggers, steps, tools, handoffs, exceptions and judgment calls.
Stage 3: Capture evidence
Each observation is recorded with its source and how far it can be trusted.
From you
Consent for what may be observed and stored.
From us
An evidence log that keeps what people said separate from what we saw and from our own inferences.
Stage 4: Map the current process
The workflow as it really runs — not as the process document says it should.
From you
A review to confirm the map is right.
From us
A step-by-step map with actors, systems, time estimates and friction points, validated with you.
Stage 5: Establish a baseline
One or two measures that matter, recorded before anything changes.
From you
Access to the data that exists, or help counting it.
From us
A baseline with its period, sample size, source and known confounders. Missing data is marked missing, not guessed.
Stage 6: Compare improvement options
Several ways to address the friction, including doing nothing.
From you
Your judgment on cost, risk and fit.
From us
A side-by-side comparison of process changes, deterministic software and AI, with assumptions and costs written down.
Stage 7: Run a controlled pilot
One small change, on a defined scope, for a defined period.
From you
A named approver, and the people who will try it.
From us
A pilot plan with acceptance criteria, stop conditions and a rollback, approved before it starts.
Stage 8: Measure the outcome
The same measure, after the change, compared with the baseline.
From you
Honest feedback from the team.
From us
A result labelled for what it is — observed, or not enough evidence — with confounders stated.
Stage 9: Document what can be repeated
If it worked, a playbook. If it didn't, a clear record of why.
From you
A decision: scale, revise or stop.
From us
A report and, where justified, a reusable playbook another team can check against their own process.
How long it takes
It depends on the workflow, how often it runs and how much data already exists. Stages one to six form the Operational X-Ray. A pilot runs for as long as it needs to produce a usable comparison — agreed in advance, not stretched until it looks good.
Where it can stop
At any stage. Sometimes the X-Ray shows the problem is smaller than it felt, or that the best option is a simple process change you can make yourselves. That is a useful result, and we will say so.
Start with stages one to six.
An Operational X-Ray covers one workflow from first conversation to a scoped pilot proposal.