The six dimensions
Value potential contributes 25% of the score; process stability 20%; data readiness 15%; risk controls 20%; integration readiness 10%; and adoption capacity 10%. Each is scored from the answers you provide. These weights express an editorial framework, not a validated prediction of business performance.
How inputs map to scores
Weekly active labor under 0.5, 2, 5, and 10 hours maps to value scores of 10, 35, 60, and 80; 10 or more hours maps to 100. Changing, mostly stable, documented, and unknown rules map to 15, 60, 100, and 25. Structured, mixed, unstructured, and unknown data map to 100, 55, 30, and 20.
Risk, integration, and adoption
Risk uses the lower of sensitivity and error-impact scores. Public/internal/personal/financial/regulated/unknown sensitivity maps to 100/80/45/25/0/15; low/moderate/high/critical/unknown impact maps to 100/60/15/0/15. Native/supported/manual/unknown integration maps to 100/85/25/20. Adoption uses the lower of owner capacity (100/45/0/10 for available/limited/none/unknown) and pilot capacity (100/60/0/15 for ready/some/none/unknown).
Operating controls limit each dimension
Each dimension takes the lowest relevant score, so one missing control cannot be hidden by stronger answers. Value is limited by baseline evidence and sample coverage. Process stability includes trigger, completion criteria, handoffs, documentation, and exception routing. Data includes quality checks and identifiers. Risk includes permission, data minimization, and an available reviewer when review is required. Integration includes access, limits, monitoring, recovery, and duplicate handling. Adoption includes team involvement, cost coverage, pilot cases and boundaries, acceptance and stop rules, rollout, training, and a review date. Full machine-readable mappings are available at /assessment-framework.json.
Decision rules take priority
First, material or critical consequences, regulated judgment, blocked data permission or system access, a required but unavailable reviewer, or an uncontrolled live pilot put execution on hold. Unclear boundaries, frequently changing rules, missing documentation, or unbounded exceptions call for process mapping. An adequate existing feature calls for an existing-tool plan. No owner, no pilot capacity, less than half an hour of weekly effort, or no new-software budget defers purchase. Open unknowns, unfinished before-pilot actions, unavailable integration, financial sensitivity, or a score below 70 require validation. Only the remaining cases support a bounded trial.
Time is an estimate, not a promise
Weekly effort equals quantity multiplied by total human minutes per unit, divided by 60. Monthly quantities are divided by 52/12. The illustrative net range subtracts new review and maintenance from 25–50% of current labor. Edit those assumptions to suit your process. A negative result means the assumed new work outweighs the time released.
Optional cost scenario
Monetary estimates are produced only when labor value per hour, setup hours, and monthly software cost are all entered. Monthly capacity value equals weekly net released hours × 52/12 × labor value, minus additional monthly software costs. Setup labor value equals setup hours × labor value. Modelled setup recovery divides setup labor value by positive monthly capacity value; a non-positive scenario has no positive recovery period. These figures value capacity, not guaranteed cash savings. They exclude exchange-rate conversion, taxes, financing, other one-time fees, and unmodelled repair work. They do not override the readiness decision.
Reported evidence and limitations
Five evidence checks are shown separately: a measured or sampled baseline; checked input quality and a representative baseline; testing of existing capabilities; tested connection behavior with verified access; and tested recovery with representative cases. Zero or one reported checks means Limited coverage; two to four, or any unknown answer, means Partial coverage; all five with no unknowns means Broader coverage. Nothing is independently verified: no documents are uploaded and no business systems are inspected. More questions and a higher score are not a certification or a probability of success.
Review, records, and older assessments
The final step shows all 15 input sections with direct editing. A change requires renewed review before a report is generated. The copied and downloaded report includes the answer record, evidence coverage, prioritized actions, time assumptions, and any optional cost scenario. Previously saved six-step answers are retained and reopened as a draft; the added questions must be completed before a new result is shown.
Commercial independence
The result is calculated before software candidates are matched. Candidates require an eligible result, confirmed data permission, and verified system access, and are limited by workflow type and audience. They are research starting points, not verified integration matches. Commercial approval cannot override any control. All provider profiles currently use ordinary official links. Framework version: A02-2.0.0.