← AI Adoption Cycle 🕐 9 min read
AI Adoption Cycle

Rewired Capability Gap Scorecard

Brandon Sneider · April 2026

Score each capability 1–5. Use the Current State Indicators to anchor the score — if three or more indicators in a row describe your organization, that is your score for that row.

One-page diagnostic for Fortune 500 clients. Fill in during a 30-minute working session with the CIO or transformation lead.

Framework: Lamarre, Smaje & Levin, Rewired 2nd ed., 2024. Evidence grounded in independent corpus data — see source notes below each row.

How to Use This Scorecard

Score each capability 1–5. Use the Current State Indicators to anchor the score — if three or more indicators in a row describe your organization, that is your score for that row. Then assess the Gap Signal column to confirm the score is calibrated against what the data says matters. Priority Score (1 = table stakes, 5 = highest strategic leverage) is pre-set from corpus evidence; it should not change unless your industry has a documented reason to resequence.

Completing this with a client: Score each row together. The most valuable part of the session is disagreement between the CIO and the business unit owner on the same row — that disagreement is the gap.

After scoring all six rows, total the scores (6 = minimum / 30 = maximum). Then read the Scorecard Interpretation at the bottom.

Scorecard

Rewired Capability Current State Indicators Gap Signal Target State Priority Score (1–5)
1. Bold Strategy & Roadmap (Ch. 1–7) Score 1–2: AI portfolio spans >10 initiatives; no named business domains; CIO or CAIO owns transformation brief; board has not reviewed AI strategy in 12+ months. Score 3: 5–10 initiatives; some domain framing in documents but not in budgets; strategy owned by a steering committee. Score 4–5: 1–3 named business domains (e.g., claims automation, customer acquisition, supply planning); each domain has a P&L owner who is a business leader, not IT; economic case per domain is written and tracked. McKinsey Manifesto (n=20 leading companies, Apr 2026): concentration on 1–3 domains with full workflow reinvention delivered 20% EBITDA uplift vs. incremental use-case accumulation. BCG AI Radar 2026 (n=640 CEOs): only ~15% of CEOs apply AI end-to-end. The “use-case implementation race” anti-pattern (Rewired, Introduction, pp.20–21) is the #1 failure mode preceding wasted transformation budgets. A named business leader (not IT) owns each of 1–3 business domains. Each domain has a P&L-attached economic case and a production deployment on a 12-month roadmap. Board reviews AI strategy quarterly. 5 — Domain concentration is the single most predictive strategic variable separating high performers from the 94% majority.
2. Talent Bench (Ch. 8–11) Score 1–2: No ML engineers, data engineers, or SREs in-house; IT fully outsourced; business leaders cannot read a model eval or interpret a confusion matrix; no AI literacy program. Score 3: Some technical talent in place; business leaders know the vocabulary but not the economics; training is voluntary and rarely completed. Score 4–5: Technical talent hired or under contract; business leaders in AI-adjacent domains have completed structured tech-literacy program; agentic-era roles (workflow designer, agent supervisor, eval engineer) are defined and staffed. Spencer Stuart / Stanford: <20% of Fortune 500 boards have two or more directors with operating AI experience. BCG AI at Work 2025 (n=10,600): employees with 5+ hours of AI training become regular users at 79% vs. 67% with less. Rewired Ch. 9 p.141: IT talent reshape is a triage, not a headcount reduction. Fully outsourced IT is incompatible with a Rewired transformation at any stage. Business leaders in AI-active domains have completed a structured tech-literacy curriculum. Technical roles are filled or contracted. Training completion is tracked, not voluntary. At least one person per AI pod is accountable for eval quality. 4 — Talent gaps take 6–18 months to close. Starting late on capability builds is the second-most-common reason transformations stall behind schedule.
3. Operating Model (Ch. 12–16) Score 1–2: IT owns all AI projects; business units are “consulted” but not accountable for outcomes; projects are handed off between IT and business on a requirements-delivery cycle; no stable cross-functional pods. Score 3: Some pod-like teams in place but not stable; business leaders involved but not accountable; governance reviews happen at project end, not throughout. Score 4–5: Stable, business-aligned, cross-functional pods own a workflow or product end-to-end (Ch. 12, p.199); engineering is embedded (Ch. 13, p.217); pod owns the P&L outcome, not just the deployment. McKinsey State of AI (n=1,993, Nov 2025): only 21% of organizations using gen AI have fundamentally redesigned any workflows. The organizational bottleneck, not model quality, is what separates the 6% high performers from the 94%. Stanford Enterprise AI Playbook 2026 (n=51 deployments): AI-primary architectures achieve 71% median productivity gain vs. 30% for human-primary — same tool, different operating model. Pod-based operating model with embedded engineering. Pods own workflow outcomes, not task delivery. Business leader, not IT leader, is accountable for P&L results. Stable team persists through production, not just deployment. 4 — Operating model is the structural prerequisite for Capability 5 (adoption at scale). Organizations that skip this produce the IT-handoff failure pattern.
4. Technology & Data Backbone (Ch. 17–29) Score 1–2: Data lives in 5+ systems with no integration layer; no API access to production data; AI pilots ran on curated sample data; no data architecture investment; tech debt is unpriced. Score 3: Some data modernization underway; 1–2 data domains accessible via API; platform exists but is not reusable across use cases; tech debt is known but not in the AI business case. Score 4–5: Governed, reusable data architecture in place (Ch. 25, p.391); data products built once and reused (Ch. 26, p.401); 3+ AI use cases running on shared infrastructure; tech debt priced and in the business case. Gartner (2026): 60% of AI projects predicted to be abandoned through 2026 due to lack of AI-ready data. Only 7% of enterprises say their data is completely ready for AI (Cloudera/HBR, n=230, Mar 2026). IBM IBV (n=1,300, Nov 2025): 18–29% of total AI implementation cost through 2027 absorbed by tech-debt remediation. Mid-market adaptation: for organizations below Fortune 500 scale, the Rewired “build the platform” prescription translates to purchasing Snowflake+Cortex / Databricks / Vertex and investing in data products on top (see Tension 3 in synthesis). Reusable data architecture covers the 1–3 priority domains. Data products are built once and consumed by multiple AI workloads. Tech debt for each priority workflow is priced and included in the deployment business case. Platform selection decision is made and documented. 5 — “No data architecture, no AI advantage” (Rewired Ch. 25, p.391). This is the bottleneck that produces the Gartner 60%-abandonment figure. It is also the slowest capability to build — 9–24 months for mid-market Path C resets. Assess now.
5. Adoption & Scaling (Ch. 30–33) Score 1–2: No workflow redesign mandate; AI tool rollout is a top-down mandate; no training beyond how-to-use the tool; success metric is “hours saved” or “licenses deployed”; no reuse architecture. Score 3: Workflow redesign discussed but not completed for most deployments; training covers tool mechanics but not when not to use the tool; some KPI tracking. Score 4–5: Adoption is engineered (Ch. 30, p.457): explicit design of what AI handles, what humans handle, what is eliminated. Training covers tool + workflow + failure modes. Reuse architecture designed from day one (Ch. 31, p.469). KPIs tied to P&L line items, not activity counts. McKinsey (n=1,993): 55% of high performers fundamentally redesigned workflows vs. 18% of others — the single most predictive behavioral gap across 25 tested attributes. MIT CISR (n=721): Stage 1 (tools without redesign) = −12.6 pp growth vs. industry average; Stage 3 (with redesign) = +11.3 pp. Futurum Group (n=830, Feb 2026): financial ROI nearly doubled YoY as primary success metric; “hours saved” fell 5.8 points. A 2026 CFO will not accept hours-saved as a return metric. Workflow redesign precedes or runs parallel with tool deployment. Training covers when not to use the tool, not just how. KPIs are tied to specific P&L lines. Reuse architecture is explicit in the design: which data products and infrastructure from deployment #1 serve deployment #2 and #3. 5 — This is the highest-leverage near-term capability for most organizations. It does not require new technology; it requires organizational decisions about what changes and who owns the change.
6. Governance & Risk (Ch. 34–35) Score 1–2: No named individual whose performance review includes AI governance outcomes; governance framework has no review cycle; shadow AI operating without approval in most departments; risk and compliance are not involved until a problem surfaces. Score 3: Governance framework exists on paper; committee owns it but no single accountable person; some shadow AI acknowledged; review cycle is annual rather than quarterly. Score 4–5: Named owner of AI governance with performance accountability. “Right-to-deploy” framework in place (Ch. 34, p.511): legal, risk, compliance, and security sign off as a parallel stream to engineering. Living governance with review cycle matched to deployment velocity. Psychological safety for experimentation embedded in culture (Ch. 35, p.523). McKinsey Responsible AI benchmark (n=~500, 2026): organizations with named RAI ownership score 2.6/4.0 vs. 1.8 without — a 0.8-point gap that maps directly to the financial-performance divide. EY Technology Pulse (n=500, Jan–Feb 2026): 52% of department-level AI initiatives operate without formal approval or oversight. Grant Thornton (n=950, Feb–Mar 2026): 78% of all organizations cannot pass an independent AI governance audit in 90 days. MIT CISR FinCo case: comprehensive governance without a living review cycle produced more shadow AI than before. Named governance owner with P&L accountability for AI risk outcomes. Governance framework has explicit review cycle (quarterly minimum). Less than 10% of AI deployment surface is operating without formal approval. Culture allows deployment teams to flag problems without career risk. 3 — Governance is a Stage 3 scaling prerequisite but should not be the first investment at Stage 1. Sequence it after data architecture is underway and workflow redesign is committed to — not before, or it becomes the FinCo failure pattern.

Scorecard Interpretation

Total Score Stage Equivalent Recommended Action
6–12 Stage 1 (Experimenting) Stop adding use cases. Identify your single highest-value domain (Capability 1), run the data reset decision tree on that domain’s workflows, and commit a production path. Do not expand pilot portfolio.
13–20 Stage 2 (Scaling Pilots) You have working pilots and no production momentum. The gap is almost always Capability 5 (adoption/workflow redesign) or Capability 4 (data architecture). Run the 30-minute workflow readiness assessment on your highest-value pilot.
21–26 Stage 2→3 Transition You are doing most things right. The bottleneck is usually one specific capability scored 1–2. Fix that one before adding new domains.
27–30 Stage 3+ (Industrialized) Run this scorecard again in 6 months against the next domain. Scale what is working before building new infrastructure.

Evidence Summary

Source Stat Capability
McKinsey Manifesto (n=20 leaders, Apr 2026) 20% EBITDA uplift from domain concentration 1
BCG 10/20/70 (Mar 2026) 70% of AI value in people, process, org design 3, 5
MIT CISR (n=721) Stage 1: −12.6 pp growth; Stage 3: +11.3 pp 5
Stanford Enterprise AI Playbook (n=51, 2026) AI-primary: 71% productivity gain vs. human-primary: 30% 3, 5
Gartner (2026) 60% of AI projects abandoned due to data unreadiness 4
IBM IBV (n=1,300, Nov 2025) 18–29% of AI implementation cost absorbed by tech debt 4
McKinsey RAI benchmark (n=~500, 2026) Named RAI owner: 2.6/4.0 maturity vs. 1.8 without 6
NBER (n=5,867, Feb 2026) 89% of firms report zero productivity impact from AI All
BCG AI at Work (n=10,600, Jun 2025) 5+ hrs training → 79% regular user rate vs. 67% with less 2

Brandon Sneider | brandon@brandonsneider.com April 2026