Open methodology · WMI v1
How AOWLab measures the quality of work management: five dimensions, two evidence channels, one score. This page is the public layer of the methodology.
The starting point
AI performs to the structure it is given. An organisation where work is undefined, without owners, without deadlines, with work living in chat, cannot delegate that work to AI, because there is nothing structured to delegate. WMI quantifies exactly that gap. The quality of work management is a leading indicator of AI readiness, which is why we measure it first.
No single channel of evidence is sufficient on its own. Telemetry shows what the system records. Surveys show what people experience. WMI reads both. Where the channels disagree, the disagreement is measured and reported as a finding, not averaged away. That distance between belief and data is the Perception Gap.
Channel A · AoW System Analysis
Read-only API telemetry over a 60-day window. Shows what the system records: ownership, deadlines, flow, concentration. Metadata only, never content.
Channel B · AoW WMS
A short, anonymous pulse survey. Shows what people experience: coordination load, meetings, tool fragmentation, informal networks. Reported only in aggregate.
What we measure
The quality of work management decomposes into five dimensions. Each is scored 0–100, higher is better, and each is read from the evidence channel that can actually see it. Together they cover 23 metrics.
Structure
Is work defined? Owners, deadlines, context, goals.
Flow
Does work reliably reach done?
Load
How much energy leaks into coordination instead of work?
Resilience
Does work survive the absence of individuals?
Adoption & Readiness
Do people actually work in the system, and are they ready for AI?
The metric definitions live on work objects: tasks, owners, deadlines, projects, goals. Not on the features of any single tool. Asana is the first connector, not the definition.
How we interpret
The five dimensions compress into two axes: how well work is designed, and how efficiently it runs. An organisation can score well on one and poorly on the other, and that asymmetry is the diagnosis. The centre of the map is the market benchmark, so every position reads relative to the market.
low design · low efficiency
Work is undefined and it leaks. Held together by the heroics of individuals. Highest urgency, biggest upside.
low design · high efficiency
Runs on informal networks and small scale. It works today; it breaks with growth, departures, or an AI rollout.
high design · low efficiency
The architecture exists on paper; execution flows around it. Typically: a system in place, the work in chat.
high design · high efficiency
Designed and flowing. The only quadrant where AI adoption compounds instead of amplifying chaos.
Maturity
The WMI Score maps to five maturity levels, but a level is not just a score band. Each level above the first is guarded by qualitative gates: conditions that must actually hold in the organisation, such as an established measurement cadence or repeated processes being identifiably defined. An organisation receives the lower of its score band and the level its gates support. Levels are a stable vocabulary for boards, partners and procurement; the Matrix is the working picture.
Ad hoc
Work lives in heads, chat and e-mail. No shared system of record.
Tool-present
A work management system exists and a meaningful share of people are genuinely active in it.
Structured
Work is defined: owners, deadlines, context. Repeated processes are identifiably defined.
Managed
A measurement cadence is established and goals are wired to execution.
Architected
Sustained over consecutive measurements. Sanctioned AI use exceeds shadow use, and teams share one architecture of work.
Honesty principles
A diagnostic is only worth what its honesty allows. These principles are not compliance decoration; they are structural rules of WMI, and every report is built on them.
Every WMI output carries a computed confidence grade: high, medium or low. It is printed, never omitted. Low confidence is reported with explicit caveats.
Every report contains an explicit section listing what was outside the observation window or below the evidence threshold. Ranges instead of false precision.
The centre of the Matrix is a benchmark population, and its version is printed on every report. When the benchmark moves, the version changes. No silently shifting goalposts.
The diagnostic is allowed to conclude that no intervention is needed. That outcome is codified in the methodology and it ships like any other result.
WMI measures how work is designed and how it flows, not people. The data model contains no individual performance scores, and results are reported in aggregate.
Processing and storage within the EU. Only work-object metadata is read, never message bodies, file contents or the text of communication. Raw pulls are deleted once metrics are computed.
WMI v1.0 · July 2026. The computational layer of the methodology (weights, formulas, the metric catalogue, question wording) is licensed to certified partners and is not published.
Read-only · no installation · results in minutes