Work Understanding - Concepts and key terms
Objective. Build the mental model behind Work Understanding and define its key terms plainly, so a rep can reason about any screen from first principles rather than memorising each one, and can explain the inference-and-confidence idea that the whole app rests on.
The core idea of Work Understanding is that a process is not a diagram someone drew - it is a shape you infer from observed work, carrying a confidence, against which you can localise where time is lost. This module gives the mental model and a plain-language glossary of the terms that recur across every screen (discovered workflow, variant, conformance, reference path, step time, waiting, friction, rework, detection confidence, coverage, opportunity vs released time), then chains them together and names the one concept people most often get wrong.
The core concept the app is built on
Every other Time Champ app answers a 'what' or 'how much' question - how many paid hours, how much output, how many denials. Work Understanding answers a 'why' question: why does this work take as long as it does, and which change should we validate first?
To answer that, it does something distinctive: it infers the process from captured activity signal rather than reading a documented process. It discovers how work actually moved, exposes the variants a whiteboard never captured, and localises where the time goes. Because the process is inferred, every output carries a confidence and is bounded by coverage - the honest unit is 'observed contribution with a confidence', never 'proven cause'.
Hold two ideas together: the app is powerful because it sees reality instead of a self-reported snapshot; and it is honest because that reality is inferred, so findings are candidates to validate, not verdicts. The whole product is a disciplined way of turning observed signal into a change worth testing - and then measuring whether the test worked.
Key terms defined plainly
- Discovered workflow - the shape of a process inferred from how work actually moved, not a hand-drawn map. An evidence-based draft to confirm with the customer.
- Variant - an alternative path through the process that the discovery found. Variants can be legitimate handling of different cases, not necessarily errors.
- Reference path - the intended path a process is supposed to follow. Customer-owned. Conformance is measured against it.
- Conformance - how often reality follows the chosen reference path. The reference choice drives the number, so conformance is never an absolute grade.
- Step time - how long a step takes, split into active handling and waiting.
- Waiting - elapsed time where nothing is actively being done (for information, a payer, an approval). In most RCM processes this is where the days sit.
- Friction - a detected point where time is lost. A hypothesis with a confidence, not proven waste.
- Rework / handoff - work redone, or passed between people or systems. Localises possible time loss; some is necessary.
- Detection confidence - how sure the app is that observed activity was a given piece of work. Everything downstream inherits it.
- Coverage (validated work coverage) - how much of the real process is observed with adequate confidence. Low coverage means partial findings.
- Opportunity - the modelled time a change might release. A hypothesis.
- Released time - the verified reduction measured by a before/after. The result. Never the same as opportunity.
How the concepts chain together
The terms are not a flat list - they form a chain where each link's trustworthiness depends on the one before it:
Capture signal -> detection (with confidence) -> taxonomy mapping -> discovered workflow -> variants + conformance (against a reference path) -> step time + waiting -> friction + rework (hypotheses) -> an experiment in the improvement workspace -> released time (the result).
Read left to right, confidence only ever decreases or holds - it never improves downstream. A friction finding cannot be more trustworthy than the detection it rests on, and a conformance figure cannot be more meaningful than the reference path it was measured against. Coverage bounds the whole chain: a process only partly observed produces only partial findings everywhere downstream.
The chain also tells you where honesty lives at each stage: discovery and variants are inferred (confirm with the customer); conformance is reference-dependent (state the reference); friction and rework are hypotheses (validate, do not sum); and only the final link, released time from a measured before/after, is a result you can report as a saving.
The one concept people most often get wrong
The single most common error is collapsing opportunity into released time - treating the modelled 'this change might release X hours' as if it were the achieved saving.
They are different kinds of number. Opportunity is a hypothesis derived from inferred findings; released time is a measured result from a before/after on the same scope and metric, against the customer's own baseline. Reporting opportunity as result is the exact error the improvement workspace exists to prevent, and it compounds with the overlap trap: opportunities from overlapping findings can be summed on a slide, but the time they claim may be the same time counted twice.
The discipline that fixes it: opportunity frames which change to validate first; released time is what you report once you have validated it. If you only remember one thing about the app, remember that the headline number is always a measured before/after, never a modelled opportunity - and never a sum of overlapping ones.
Interpretation check
A process in Work Understanding is inferred from observed signal with a confidence, not read from a document - so discovery and variants are confirmed with the customer, conformance depends on the chosen reference path, and friction and rework are hypotheses. Confidence only decreases down the chain, and the only reportable saving is measured released time, never modelled opportunity.
In the field
KnackGlobal RCM: the concept chain explains why the review focuses on waiting in the claim process - detection and discovery surface the real flow, step time separates minutes of data entry from days of waiting, and the only number that goes to the customer is a measured before/after on a validated waiting-step change, not a modelled opportunity total.
Practical task
Draw the concept chain (capture -> detection -> taxonomy -> discovery -> variants/conformance -> step time/waiting -> friction/rework -> experiment -> released time) for one demo process, and at each link write the one honesty rule that applies there. Then write a two-sentence explanation of opportunity vs released time you could say to a customer.
Self-check
In one sentence, what distinctive question does Work Understanding answer, and how is its method different?
It answers 'why does this work take as long as it does, and which change should we validate first' - by inferring the process from observed activity signal with a confidence, rather than reading a documented or self-reported process map.
Explain the difference between opportunity and released time.
Opportunity is the modelled time a change might release - a hypothesis from inferred findings. Released time is the verified reduction measured by a before/after on the same scope and metric against the customer's baseline - the result. Only released time can be reported as a saving.
Why can a friction finding never be more trustworthy than the detection beneath it?
The concepts form a chain from capture signal through detection, taxonomy, discovery and conformance to friction. Confidence only decreases or holds down the chain - a finding inherits the confidence and coverage of the detection it is built on, so weak detection means a weak finding.
What does the choice of reference path have to do with a conformance number?
Conformance is how often reality follows the chosen reference path, so the reference choice drives the number. Change the reference and the same process scores differently - which is why conformance is never an absolute grade and the reference must be stated and agreed.
What is the single concept people most often get wrong, and the discipline that fixes it?
Collapsing opportunity into released time - reporting a modelled estimate as an achieved saving. The fix: opportunity frames which change to validate first; released time, from a measured before/after on de-duplicated findings, is the only number you report as a saving.
Verified vs demo. The screens that illustrate these concepts are demo-shown on seeded data; figures are SME-validation-pending. The concepts and their honesty rules are the app's own discipline; the actual detection method, confidence thresholds and coverage for a customer are confirm with product.