Diagnose before you automate
Automating an undocumented process encodes its defects at speed. The diagnosis is the intervention that makes every later decision cheaper.
Executive summary
- Most disappointing transformation results come from good technology applied to a workflow nobody observed.
- A two-to-four week diagnosis produces the baseline, the constraint and the ranked opportunity list that make implementation defensible.
- The output of diagnosis is not a recommendation deck — it is a measurable starting point.
Executives rarely buy the wrong software. They buy reasonable software for a process that has never been described accurately. The tool is configured around the documented workflow, the organization keeps running the real one, and the gap between them becomes the cost of the program.
Three questions a diagnosis must answer
- What does this workflow actually cost the business today, in numbers finance recognizes?
- What is the binding constraint — capacity, information, decision latency, or handoffs?
- Which intervention moves the constraint at the lowest risk and shortest time to value?
None of these can be answered from a systems diagram. They require observation of the real path: the tabs open on the operator's screen, the spreadsheet that holds the truth, the exception nobody logged.
Why the sequence matters commercially
Diagnosis is the cheapest phase of any transformation and the one that determines the return on every phase after it. A wrong build decision costs months. A wrong diagnosis costs the program.
The diagnosis sequence
- 01
Objective
Agree the business measure that defines success before touching the process.
- 02
Observation
Watch the real work performed by the people who perform it.
- 03
Baseline
Instrument volume, cycle time, touches, exceptions and unit economics.
- 04
Constraint
Identify what actually limits throughput or quality.
- 05
Options
Rank interventions on value, feasibility, risk and adoption effort.
- 06
Business case
Model the financial impact against the baseline so finance can challenge it.
A practical example
A services firm believed its bottleneck was proposal production and planned to buy a proposal automation platform. Two weeks of observation showed proposals took four hours; the delay was upstream, in an intake process where qualification waited on a single partner's availability, often for three days.
The intervention that moved the number was structured intake with routed qualification — not proposal automation. The platform purchase was deferred, and the same budget produced a measurable conversion improvement instead.
Action checklist
- Name the single business measure the change must move.
- Observe at least five live instances of the workflow end to end.
- Record volume, cycle time, touches, exception rate and unit cost.
- Identify where information is re-keyed between systems.
- Ask the operators where the process fails, and write down their answer verbatim.
- Rank candidate interventions before evaluating any vendor.