Framework alignment

Track B — Traditional Change Management

Kotter’s 8 Steps, Prosci ADKAR, and McKinsey 7S — the standard enterprise playbook for transformation, executed well.

01

The phases

UrgencySPRINT 1
Competitive analysis. “AI is transforming claims — we must lead or follow.”
CoalitionSPRINT 2
Steering committee formed. Consultants engaged. Center of Excellence chartered.
VisionSPRINT 3
Future-state operating model designed. New roles defined.
CommunicateSPRINT 4
Town halls. Roadmap published. Change champions recruited.
EmpowerSPRINT 5
Vendor selected. New roles posted. RACI updated. Training developed.
WinsSPRINT 7
Early-adopter stories. Self-reported time savings. Momentum.
ConsolidateSPRINT 9
Lessons learned. Phase 2 planning. ROI dashboard.
AnchorSPRINT 13
AI in performance reviews. CoE permanent. Innovation Award.
02

The mechanics — what these phases now actually do

Track B is not a narrative. From Empower onward it commits and rolls out real workflow mutations, so it pulls the same levers as Track A and the winner is decided by conditions, not construction.

A-priori seam
standardization
Target what’s gappy first. Track B ranks the seams its Target Operating Model predicts are risky and standardizes them in rollout waves — re-attempting any it missed or that reverted. Success on each seam is design_accuracy_base × (1 − seam_tacitness): accurate on codifiable seams (schemas, thresholds, payment rails), blind on tacit ones (negotiation, liability). Kotter Step 5 — Empower Broad-Based Action, made mechanical
Regression under
change fatigue
Shallow, imposed change reverts under load. Every sprint after a wave, each completed seam rolls for reversion scaled by actual turbulence + change exhaustion — not a flat timer. Calm quarters hold; turbulent ones wash the fix off. This is the honest cost of big-bang. Prosci Reinforcement + Kotter Consolidate & Anchor; ~70% of transformations fail to sustain
03

Framework alignment Kotter · Prosci · McKinsey

Kotter step 1 —
Create urgency
CEO presents competitive analysis showing GEICO/Allstate AI investments. CFO models 15% cycle-time improvement as annual savings. Industry-trends deck distributed. Without urgency, the organization defaults to the status quo. John Kotter, “Leading Change” (1996) — 8-step framework
Kotter step 2 —
Build coalition
Steering committee with executive sponsorship (Kathryn as VP Claims), operations (Ron), IT (Mike), and compliance (Leslie). External consultants engaged. AI Center of Excellence chartered. The coalition must have enough power to overcome inertia. Kotter: “A strong guiding coalition is always needed — right composition, trust, shared objective.”
Kotter step 3 —
Form vision
Vision statement: “Lead the industry in AI-enabled claims processing, achieving 40% cycle-time reduction and 25% cost-per-claim improvement through enterprise-wide deployment.” Target operating model designed: new roles (AI Claims Supervisor, AI QA Lead, Prompt Engineer), new reporting lines (AI CoE → CIO, dotted line to Claims VP), new governance (Ethics Review Board, Decision Appeal Process). Kotter: “A clear, compelling vision people can see and understand.”
Kotter step 4 —
Communicate
Multi-channel communication: all-hands town halls, an intranet site (“Our AI Journey”), department briefings, an FAQ document, and a change-champion network on a volunteer basis. Communication must be frequent, powerful, and two-way. Kotter: “Transformation is impossible unless hundreds of people are willing to help.”
Kotter step 5 —
Empower action
Obstacles removed: vendor selected (Microsoft Copilot + custom claims agent, $8M contract), new roles posted, RACI updated for all claims workflows, training curriculum developed (4 modules, 8 hours), pilot group identified (200 users). The key is removing barriers, not just announcing change. Kotter: “Remove obstacles to the vision. Encourage risk-taking.”
Kotter steps 6–8 —
Wins, consolidate, anchor
Wins: early-adopter success stories, self-reported time savings. Consolidate: Phase 1 lessons learned, Phase 2 use cases prioritized, ROI dashboard live. Anchor: AI proficiency in performance reviews, CoE made permanent, onboarding includes AI training. Kotter: “Plan for visible improvements. Consolidate gains. Anchor new approaches in the culture.”
Prosci ADKAR —
individual change
Awareness (why AI?) → Desire (what’s in it for me?) → Knowledge (how do I use it?) → Ability (can I do it?) → Reinforcement (is it sticking?). Applied through the training curriculum, change-champion outreach, and resistance management. Each individual must pass through all five stages for change to hold. Prosci ADKAR Model — the most widely used change framework in enterprise transformations
McKinsey 7S —
alignment
Strategy: AI-enabled claims leadership. Structure: new AI CoE, reporting lines. Systems: Copilot + custom claims agent. Shared Values: “innovation with accountability.” Style: executive-driven, data-informed. Staff: upskilled through training. Skills: prompt engineering, AI QA, AI supervision. All seven must align for transformation to succeed. McKinsey 7S Framework — organizational-alignment diagnostic for large-scale change
04

What Track B measures8 concerns

License activation
rate
% of employees with activated AI tool licenses. Baseline 0% (sprints 1–2), organic pre-rollout growth (sprints 3–4, ~5–15%), active rollout ramp (sprints 5–8, 15–95%), steady state (sprint 9+, ~95%). Measures access, not value.
Training completion
rate
% of employees who completed AI training modules. Lags license activation — training is harder to scale than license distribution. Training completion ≠ capability; people complete training and change nothing.
AI-assisted
claims %
Claims where AI touched any step. Progressive already at 55% baseline from existing AI pipeline; target 80% after enterprise rollout. Maps to volume, not quality — AI touch ≠ AI value.
Milestone hit
rate
Roadmap phase completion on schedule: urgency → coalition → vision → communicate → empower → wins → consolidate → anchor. Each phase has a target sprint; the steering committee reviews milestone status monthly.
Adoption
dashboard
Weekly active users, departments onboarded, low-adoption teams, resistance flags. Reviewed by the steering committee each sprint. Low-adoption teams get more training and change-champion outreach.
Budget &
ROI
$15M total program budget: consultancy $2M, vendor contract $8M, new roles $150K/year each, training $200/head × 70K. Steering committee reviews burn rate vs. plan. ROI estimated from cycle-time and cost-per-claim projections.
Employee AI
sentiment
Survey-based measure of employee attitudes toward AI. Captures the Desire and Reinforcement stages of ADKAR. Resistance is flagged as a “standard adoption curve” and met with training and communications.
Resistance
interventions
Count and type of resistance-management actions. Diagnosis: “resistance to change — standard adoption curve.” Intervention: more training and change-champion outreach. Tracked per department; escalated to the steering committee if persistent.
05

What Track B does NOT do

Does not deploy
without a plan
AI is not deployed into existing structures without first designing the target state. The operating model is designed, then implemented — not discovered through implementation. A deliberate choice, not a failure: it reduces uncertainty, aligns stakeholders, and creates accountability.
Does not grant bounded
authority to one owner
Decisions flow through the steering committee, not an individual accountable owner. This ensures cross-functional alignment and prevents local optimization at the expense of enterprise coherence. The trade-off is speed.
Does not let evidence
gate the next step
The roadmap determines what happens next, not sprint observations. Phases advance on schedule, not on evidence thresholds. This provides predictability for budgeting, staffing, and vendor management. The risk is advancing before the organization is ready.
Does not expose informal
control points
The transformation works through formal structures — reporting lines, RACI matrices, role descriptions. Informal controls (like Diana’s pre-review) stay invisible unless they surface as resistance or adoption lag. The process doesn’t systematically look for them.