AI Skill Report Card

Driving Continuous Improvement

A-83·Nov 21, 2026·Source: Web
13 / 15

When asked to improve a process, don't jump to solutions. Follow this sequence:

  1. Clarify the end goal — what value does success deliver to the business?
  2. Go to Gemba — talk to the people who actually do the work/use the process
  3. Map the current state (people/process/systems)
  4. Apply 8 Wastes to find friction and non-value-add steps
  5. Co-design the future state with those impacted
  6. Get sign-off on current vs future state before investigating solutions
  7. Investigate solutions across people, process, systems

No solutioning before the problem and value are confirmed and signed off.

Recommendation▾
Fix the duplicate numbering in the Quick Start list (two items marked '2.')
15 / 15
Progress:
- [ ] Define end goal / value alignment with business
- [ ] Gemba: interview/survey people who know (users, customers, frontline staff)
- [ ] Define the problem statement; check value and alignment
- [ ] Get agreement/sign-off on the problem definition
- [ ] Map current state (people / process / systems)
- [ ] Apply waste/root-cause analysis (e.g. 8 Wastes)
- [ ] Explore future state with impacted people — ask them what/how to fix
- [ ] Map future state (people / process / systems)
- [ ] Get agreement/sign-off on current vs future state
- [ ] Investigate and propose solutions (people / process / systems)
- [ ] Define governance (e.g. RACI) for implementation

1. Understand the end goal

Ask: What are we actually trying to achieve? How does this tie to business value? Write this down as the north star before touching the process itself.

2. Gemba — go and see

Speak directly with the people who know: users, customers, frontline staff, process owners. Use interviews and surveys. Don't rely on secondhand process documentation — observe and ask.

3. Define the problem

Turn Gemba findings into a clear, specific problem statement. Check it against the end goal for value and alignment. Get explicit agreement and sign-off before moving on — this prevents solving the wrong problem.

4. Map current state

Map the journey/process as it actually happens (not as it's assumed to happen), across three lenses:

  • People — roles, handoffs, decision points
  • Process — steps, sign-offs, wait times, rework loops
  • Systems — tools, data flows, manual workarounds

5. Find the waste

Apply the 8 Wastes (DOWNTIME: Defects, Overproduction, Waiting, Non-utilized talent, Transportation, Inventory, Motion, Extra-processing) to the current state map to surface root causes, not just symptoms.

6. Design future state

Go back to impacted people — they often already know what or how to fix it. Co-design the future state map (people/process/systems) rather than designing in isolation.

7. Agree current vs future state

Present both maps side by side. Get formal agreement/sign-off from stakeholders and sponsors before scoping solutions — this is the pivot point from diagnosis to delivery.

8. Investigate solutions

Explore solution options across people, process, and systems (templates, routing/triage, case management, system enhancements, training, policy changes). Prioritize by value and effort.

9. Governance

Set up a RACI to clarify ownership and decision rights for implementation.

Recommendation▾
Examples are solid but somewhat compressed/summarized rather than showing fuller concrete dialogue or artifacts (e.g., an actual sample problem statement or waste table) — adding one fleshed-out example would strengthen this section
16 / 20

Example 1: Input: "Our ex-gratia payment process is slow and painful — lots of sign-offs and paperwork." Output:

  • Gemba'd with requesters, approvers, and processors at all levels
  • Mapped current state: multiple sequential sign-offs, manual paperwork, long queue waits
  • Applied 8 Wastes: found Waiting (queue/approval delays), Extra-processing (duplicate paperwork), Non-utilized talent (approvers doing admin instead of judgment calls)
  • Future state designed with users: standard templates, query routing, case management system, targeted system enhancements
  • Outcome: reduced sign-off steps, cut wait times, clearer case ownership

Example 2: Input: "Customer onboarding takes 3 weeks and customers complain constantly." Output:

  • Clarify end goal: faster time-to-value for customer, fewer support escalations
  • Gemba: interview onboarding specialists and recently onboarded customers
  • Problem statement: "Onboarding takes 3 weeks due to manual document chasing and siloed handoffs between sales and ops" — signed off by sponsor
  • Current state map shows 7 handoffs, 4 of which involve waiting >2 days
  • Future state: single-owner case model, automated document requests, parallel (not sequential) checks
  • Solutions investigated: workflow tool, document automation, RACI for onboarding ownership
Recommendation▾
Add a brief note on how to handle situations where sign-off/stakeholder access isn't available, since the methodology heavily depends on stakeholder engagement
  • Always start with the end goal and value — a perfectly efficient process that delivers no value is still a failure.
  • Gemba first, always. Don't map a process from a flowchart someone else drew; go see and ask.
  • The people in the process usually already know the fix — your job is often to surface and validate it, not invent it from scratch.
  • Separate "define the problem" from "design the solution" with an explicit sign-off gate in between.
  • Map people, process, and systems together — a systems-only fix rarely solves a people/process problem.
  • Use RACI early for governance so solution ownership doesn't stall implementation.
  • Jumping to solutions before the problem and current state are agreed and signed off.
  • Mapping the process as documented rather than as actually experienced (the "should be" vs the "is").
  • Treating this as a systems/tooling project when the root cause is people or process.
  • Skipping Gemba and relying on management's view of the process instead of the user's/frontline's.
  • Designing the future state without the people who'll actually use it.
0
Grade A-AI Skill Framework
Scorecard
Criteria Breakdown
Quick Start
13/15
Workflow
15/15
Examples
16/20
Completeness
17/20
Format
14/15
Conciseness
13/15