Prepare to facilitate the reasoning

This guide supports the Colabmo course and its four application-coaching sessions. The facilitator leads inquiry and learning; the SMEs provide business knowledge and develop their own solutions. Use it with the learner course and participant workbook. Model responses are illustrative reasoning, not the only acceptable designs.

For practice, use groups of three to five; a suggested cohort is six to sixteen participants. These are design choices that make role rotation and feedback manageable. With a larger group, add capable facilitators or divide into smaller sessions. Ensure the relevant process, customer, frontline and control perspectives can be represented.

The facilitator should be able to distinguish claim types, trace an inference, recognize a disguised solution, handle disagreement, and explain the limits of a test. Rehearse the two cases and capstone before delivery. Familiarity with these materials alone does not establish prior FPT delivery experience.

Inputs and preparation

• Read the article and lessons; complete the capstone independently.
• Confirm the learning audience, accessibility needs and process examples.
• Identify a process owner, decision boundary and permitted use of data for real-work application.
• Prepare blank W1–W11 sheets, a shared writing surface, visible timer and a place for unresolved claims.
• Request a non-sensitive standard transaction, an incomplete transaction and an exception. Where unavailable, use the provided fictional cases.
• Explain that course completion is a learning record, not external certification or demonstrated business improvement.

Opening script

“Our job is to understand what the process must accomplish and which premises justify its design. You own the business solutions. I will ask you to separate evidence from assumptions, make the reasoning visible and test the points that could change your decision. We will challenge claims respectfully and record what remains uncertain.”

Ask each person for one assumption they expect to examine. Establish the right to ask for a source, revise an opinion and describe an exception. Avoid rewarding confidence or novelty more than defensible reasoning.

Run the lessons as practice sessions

Each row gives a complete timing allocation in minutes. Pause between lessons as needed; breaks are additional. The 670-minute core can be distributed across shorter learning sessions. Delivery location and commercial logistics are outside this curriculum.

If learners cannot distinguish observations from assumptions, repeat a short classification exercise before moving into principles. If they cannot generate distinct mechanisms, return to the function list and vary who, when and what evidence. If their experiment cannot fail, require a disconfirming result before proceeding.

Recognize and interrupt conventional thinking

Listen for the move in reasoning, not just a keyword. “Best practice” can introduce a useful comparator or substitute for evidence. “We need a portal” may be a hypothesis or an unquestioned requirement. Ask the speaker to identify which role the statement is playing.

Role-play cards

SME card A: Insist that the director must approve everything because a serious error occurred years ago. When asked for detail, reveal that the error involved an exception category, not standard cases. Do not concede that the director review is unnecessary; ask how the risk will be controlled.

SME card B: Insist that a new platform will solve the queue. When challenged, reveal that no one has defined the queue’s age distribution or the completeness of incoming work. Ask the facilitator to recommend a product; observe whether they return the group to functions and evidence.

Observer card: Record the exact question, whose contribution it enabled, the claim reclassified and the next evidence action. Give one concrete improvement. Do not score charisma or agreement with your preferred solution.

Debrief questions

Which intervention changed the reasoning? Which sounded like a disguised answer? What made challenge respectful? What evidence would resolve the remaining disagreement? Who owns the solution now?

Use the worked responses to assess reasoning

Outcome and evidence exercises

For “replace the approval spreadsheet with AI,” accept an outcome such as “give an eligible requester a correct authorized decision with a visible next action.” It should specify start and end events, include quality and effort, and identify who may decide. Do not require a particular speed target when the baseline is missing.

For the claim-classification exercise, “the system cannot route by category” is an unverified capability claim until tested or documented; “customers prefer email” is a hypothesis unless supported by scoped evidence; “policy requires independent authorization” is an obligation claim whose source and applicability must be checked; “18 of 60 lacked a start date” is a sample observation if records support it; “use a single queue” is a proposal. Mixed answers are acceptable when participants explain what evidence changes the classification.

Constraints and principles

The weekly supplier committee exercise should surface eligibility, authority, risk judgment and evidence retention. The meeting cadence and attendee list are mechanisms. A good conditional alternative preserves the required decision rights and handles exceptions. A poor answer assumes every approval is unnecessary.

For a principle, accept a logically supported statement such as “Before activation, the decision must be associated with the required identity and authorization evidence.” Reject “all approvals must disappear” or “one portal is always best.” A learner should be able to explain both what follows from a premise and what does not.

Onboarding case

Expected sample arithmetic: 12/40 = 30% returned and 4/40 = 10% exceptions. The combined count is unknown because overlap is unspecified. Learners should identify selection of completed cases, missing effort data and unverified independence of finance and operations. Accept any coherent alternative that satisfies the fictional policy and exposes its new assumptions.

The model test begins with replay to investigate dependency and evidence availability, then proposes a separately authorized live pilot. It must not treat replay elapsed time as a real service benefit. A proposed pilot threshold is a design choice, not a result or industry norm.

Purchasing transfer case

Expected sample arithmetic: 18/60 = 30%. The threshold and delegation rules are missing, so independent-review applicability cannot be fully resolved. Strong participants propose a useful test that retains current authority while investigating completeness, or keep a delegated option conditional. Weak responses invent a threshold or transfer the onboarding design unchanged.

Capstone model response

Outcome: a requester receives a correct authorized resolution and clear status, with restricted work protected and urgent cases continuously owned. Retain first-receipt-to-completion as a measure and separately observe waiting for information. Include reopening, permission failures, effort and urgent-case ownership as quality and operating measures.

Evidence: 20/50 = 40% returned; 10/50 = 20% urgent; 6/50 = 12% both; (20 + 10 − 6)/50 = 48% in either category; 5/50 = 10% reopened. These are observations in completed cases. Hands-on logs omit clarification, so recorded effort is incomplete. The backlog, abandonment, causal drivers and tool capability are not established by the initial data.

Functions: capture need, verify identity and permissions, assess urgency, assign an accountable owner, determine required expertise, perform authorized work, confirm resolution and communicate closure. Some functions can overlap; restricted execution must not precede the required verification. Reassignment must preserve urgent-case ownership.

Principles: required permission evidence precedes restricted work; urgent cases have an accountable owner across handoffs; completion is represented by a defined resolution condition and communication rather than simply removal from the inbox. The last principle must be agreed within the outcome definition; do not present it as an external policy supplied by the case.

Alternatives: A retains manager assignment but improves evidence preparation and visible urgent ownership. B delegates assignment to qualified coordinators with documented limits and human permission checks. C uses structured requester preparation and specialist triage, potentially aided by an advisory classifier whose output cannot authorize restricted work. Each needs exception handling and workload assessment.

Response to update: B is no longer blocked merely by a presumed manager-only rule, but the team must verify who is qualified and named. The absence of automated permission verification rules out designs relying on that nonexistent capability. Human verification can remain a valid mechanism. AI is optional and would require its own accuracy, data-handling and human-oversight test within the organization’s applicable rules.

Next test: replay standard, urgent, incomplete and restricted cases through the preferred candidate. Check correct routing, permission evidence before restricted work and continuous urgent ownership. Compare staff actions and missed decisions with the current route. After resolving replay gaps, propose a limited authorized live comparison with matched case types, a defined time measure, workload capture including clarification, zero tolerance for proceeding without required permissions, and a stop-and-restore action for control failure.

Decision: recommend a bounded test, name the owner, state the conditions and record what would cause revision. Do not claim the queue is eliminated or that the pilot has succeeded. An alternative recommendation can earn full credit if its reasoning is equally defensible.

Score the capstone consistently

Score eight dimensions from 0 to 2 for a maximum of 16. A completion recommendation requires at least 12, no zero, and correction of any proposal that bypasses required authorization or permission. Keep the group packet score separate from each participant’s reflection and facilitation observation.

Before assessing a cohort, have two reviewers score one sample and discuss differences against the anchors. If only one facilitator is available, mark uncertain judgments and seek a second review before issuing a borderline completion decision. Record specific evidence for each score; do not infer an individual’s competence from the group’s best speaker.

For remediation, return the participant to the missing dimension. An evidence gap requires revised claims and sources; a weak alternative set requires new mechanisms; an unfalsifiable test requires a failure result and decision rule. Reassess the revised work with a changed scenario where practical.

Adapt and evaluate the course

Preserve the reasoning sequence, claim distinctions, reconstruction from functions, testability and learner ownership. Adapt terminology, case complexity, accessibility supports and practice duration. For beginners, use the supplied cases before real work. For experienced teams, shorten explanations only after checking the relevant skill.

For distributed learning, use one shared worksheet per team and ask participants to contribute independently before discussion. For self-directed learning, pair participants for the challenge exercises; reading alone cannot demonstrate facilitation skill. Provide printed or accessible text equivalents of all prompts, allow additional response time where needed and avoid making visual mapping the only way to participate.

Evaluate the first delivery using four evidence types: completed learner outputs and scores; facilitator observations of recurrent confusion; participant feedback about relevance and usability; and later application records where available. Record the actual delivery context. Course production checks establish that the material exists and works as a local artifact; they do not establish instructional effectiveness.

Suggested evaluation prompts: Which distinction changed your thinking? Where was the evidence requirement unclear? Which exercise lacked enough information? Can you explain a changed decision? What would help you apply the method independently? Use the answers to revise the owning course source and preserve the prior baseline.