The Implementation Lifecycle
Diagnose → Prototype → Evaluate → Adopt. The third axis of the program's three-axis map — and the difference between shipped and adopted.
V1 · Pedagogy
✅ PASS (5/5)
Sweller, Wiggins & McTighe, revised Bloom's, retrieval practice, Vygotsky's ZPD — confirmed and reused from prior lessons.
V2 · Framework lineage
✅ PASS (3/3)
Four-stage FDE Lifecycle anchored to Six Sigma's DMAIC (Define-Measure-Analyze-Improve-Control), confirmed via ASQ. Manufacturing-quality 2026 context anchored to publicly reported AI-vision-inspection production deployments.
V3 · Quantitative claims
✅ PASS (3/3)
Lenovo (Hannover Messe 2026): 85% faster lead times, 42% lower logistics costs, 58% productivity gain. Collins Aerospace: inspection time 30 → 10 min, defects halved, output +14%. Market $1.62B in 2024 growing at 13.8% CAGR.
V4 · Logical consistency
✅ PASS (4/4)
Section times sum to 60 min; four lifecycle stages defined non-overlappingly; Apply outcome assessed by an authentic card-sorting artifact; Bloom progression Remember → Understand → Apply.
Cardinal Polymer Systems makes injection-molded automotive components across six lines in two Tennessee plants. Quality inspection is mostly manual. After watching a competitor publish a press release about AI-driven inline inspection, the VP of Operations hired an FDE to deploy computer-vision defect detection on Line 3.
Three months later, Line 3 still uses the manual process. The vision model has been quietly turned off. Operators say it "flagged too many things they had to dismiss." Nobody can produce a number for how often it caught a real defect. The Day-2 owner was never named. The supervisor who was meant to be trained as a super-user has not opened the tool in five weeks.
By the end of this lesson you'll have a four-stage lifecycle that tells you the difference between shipped and adopted — and you'll be able to look at twelve anonymized artifacts from a real engagement and place each one in its correct stage, including the three that are genuinely hard to call.
By the end of this lesson, you will be able to:
The §7 sort with twelve real-style artifacts is the assessment. The model in §4 is the lens. Naming failure modes (Understand) is the bridge between Remember and Apply — the part most lifecycle lessons skip.
Today's lesson is the third axis of the three-axis map. Lesson 1.2 told you who the FDE is. Lesson 1.3 told you what kind of team they sit inside. Lesson 1.4 tells you what they do on Tuesday — which stage of which engagement they are in.
The FDE Implementation Lifecycle has four stages. The model adapts Six Sigma's DMAIC (Define-Measure-Analyze-Improve-Control) — the canonical quality-improvement cycle maintained by the American Society for Quality (ASQ) since the 1990s. DMAIC has five stages; the FDE lifecycle compresses to four because in FDE work, the Measure stage folds into both Diagnose (measure before you build) and Evaluate (measure what you built).
Click any stage to see its detail, dominant activity, DMAIC analog, and characteristic failure mode:
1. Diagnose
Is there a problem worth solving?
2. Prototype
Can we build the narrow path?
3. Evaluate
Can we prove it works AND is safe?
4. Adopt
Day 2 — without us in the room.
Stage 1 · Diagnose
Is there a problem worth solving, and where does the work actually live?
Why the lifecycle is not linear
It's drawn as four arrows in a row, but in the field it loops. Discoveries surface in Adopt that send you back to Diagnose. Evaluation findings send you back to Prototype. A senior FDE expects to revisit each stage at least once per engagement — and treats a clean linear run as evidence she stopped looking, not as evidence that the engagement was easy.
Below are four artifacts pulled at random from the Cardinal Polymer engagement (the one from the hook). For each, the senior FDE has named the lifecycle stage and the single piece of evidence she relied on.
Four more artifacts from the same engagement. For each, click the stage you'd assign. The hint above each tag row identifies the decisive feature.
Brief
Below are twelve artifacts drawn from a different Cardinal Polymer engagement — same customer, different FDE, six months later. Sort each into its lifecycle stage. Three of them are genuinely ambiguous; they are flagged so you can spot them.
Diagnose
Prototype
Evaluate
Adopt
Justify the three hardest calls
Identify the three artifacts you found hardest. For each: which stage you chose, what the artifact's verb was, and what the alternative interpretation would have been.
A9 → Adopt. Although it ends in a code change, the verb is "Marco's new operator pinged me twice today" — an operator-using-the-tool signal. The fix is downstream; the artifact is an Adopt signal that prompted the fix.
A11 → Prototype. A planning artifact for the next prototype iteration. Could be argued as Diagnose; the test is whether the document constrains the build (Prototype) or the problem framing (Diagnose). "Narrow path v2" with named defect classes is build-side.
A12 → Adopt. Although it cites a metric (+12%), the verb is "ran the model on Line 5 for the whole week with no FDE intervention" — a Day-2 adoption signal. The metric is Evaluate-shaped; the artifact is Adopt because of who produced it and what it proves.
Senior FDE habit: classify by verb, not by noun.
Treating the lifecycle as linear
"We finished Evaluate; on to Adopt." Adopt surfaces a Diagnose error and nobody goes back.
Correction
Expect to revisit each stage at least once. A clean linear run is evidence you stopped looking — not evidence the engagement was easy.
Skipping Evaluate because "it looks fine"
The demo went well; the team is scheduling rollout. No eval suite, no red-team, no security stance.
Correction
"Looks fine" is the most expensive sentence in FDE work. Evaluate is the only stage that produces evidence the customer's compliance, security, and ops teams will accept after you're gone.
Confusing shipped with adopted
"Line 3 is in production!" Model deployed; operators still using the spreadsheet. No Day-2 owner named.
Correction
Adoption is a person, a metric, and a 30-day window — not a deploy. Three Adopt tests: (a) named Day-2 owner; (b) measurable adoption signal moved; (c) operators using the new workflow without an FDE in the room.
Bring to mind one project of your own — current or recent.
The lifecycle is the spine of the capstone. M1–M2 map to Diagnose; M3–M4 to Prototype + Evaluate; M5–M6 to Adopt. Today's reflection is your draft of which stage you're about to enter.
Which lifecycle stage do most FDE projects actually fail in, and why?
Success criterion: Adopt. Failure mode = confusing shipped with adopted: demo works, no Day-2 owner, no super-users, operators continue with the old process. Technology rarely the problem at this stage; the organization is.
Next lesson
You can now read any engagement on three axes (Trinity, Maturation, Lifecycle). Lesson 1.5 turns inward: which behaviors most distinguish FDEs who ship from those who circle.
Capstone hook
The Lifecycle is the spine of the capstone. M1–M2 = Diagnose. M3–M4 = Prototype + Evaluate. M5–M6 = Adopt. Save today's §7 sort — you'll return to it before each milestone.
Optional reading
asq.org/quality-resources/dmaic — the canonical Six Sigma cycle today's lifecycle adapts.
Companion OS
File today's §7 sort under this tag. Before each capstone milestone, classify your current artifacts to know which stage you're actually in.