AI Strategic Compass Specification
This is the single reference for how the Compass works: the dimensions, the scoring anchors, the review bands, and the domain rules. If another page or ecosystem site differs from it, follow this spec.
01Dimensions and scoring anchors
Each initiative is scored 1 to 5 on all six dimensions. On every dimension, a higher score means more risk or impact. Use the anchor that best describes the initiative. Each card shows the short label and the field name used in saved scorecards.
Whether the initiative advances the institution's mission and current priorities, or is impressive but beside the point. Considers alignment with the strategic plan, the academic mission, and leadership goals.
What could go wrong for the people involved and for the institution. Considers bias and fairness, transparency and explainability, data privacy and security, and legal/regulatory compliance (FERPA, ADA, GDPR, and peers).
Whether the institution can afford it and whether the return justifies the spend. Considers up-front and ongoing cost, staff time and resources required, and short- and long-term ROI and sustainability.
Whether it actually improves how work gets done. Considers efficiency gains, added capacity, fit with existing workflows, and the degree of real innovation rather than novelty for its own sake.
Whether the AI will actually do what it promises. Scored twice: at proposal, on outside evidence plus the quality of the evaluation plan; at the scale decision, on campus pilot results. Scored like the others: 1 = strong evidence it works, 5 = little or none.
How the people affected will experience the initiative. Considers impact on students, staff, and faculty, equity for those most affected, and public trust and institutional reputation.
Two scoring points for Efficacy Risk. A proposal has no campus evidence yet. Score it on outside evidence plus the quality of the pilot evaluation plan (the proposal score). At the end of the pilot, re-score it on campus results (the scale score). A decision to scale needs the scale score. Keep both scores on the record.
Escalation rule. A 5 on Ethical, Legal & Compliance, Stakeholder Impact, or Efficacy Risk moves the decision up one band, whatever the total, at either scoring point. For consequential student or employee decisions, a scale score of 5 on Efficacy Risk blocks scaling until validation evidence exists.
02Totals and review bands
The total is the sum of the six scores, from 6 to 30. The total sets the starting review path, and the escalation rule can move it up one band. The score opens a discussion; it does not decide the outcome.
Every band follows the same six steps: Screen, Score, Select, Plan, Track, Reflect. Plan sets exit criteria and re-review triggers. An initiative that scales is registered in the AI Registry, then operated and monitored.
Per-initiative measures are KPIs (outputs), OKRs (outcomes), and efficacy measures (quality and safety). Maturity and Readiness are institution-level context from the Maturity Assessment and are not scored per initiative. The GRC grid rates the six dimensions against Governance, Risk, and Compliance (18 cells, Level 1 / 3 / 5).
| Total | Band and review | Decides | Target time | Documentation and monitoring |
|---|---|---|---|---|
| 6–12 | Low Risk / Impact Self-assessment, then proceed | Unit or department lead | Days | Self-assessment with the unit lead against the acceptable-use checklist; Scored rubric, short description, and a named owner; Monitor at the end of the pilot and at the annual review |
| 13–18 | Moderate Risk / Impact One cross-functional review | AI governance lead or committee delegate | About 2 weeks | One cross-functional review before launch, with written comments; Document data sources, access, and retention; Draft KPIs and OKRs up front; Monitor at mid-pilot and end of pilot |
| 19–24 | High Risk / Impact Full team review plus specialist reviews | AI governance council | About 4–6 weeks | Full team review plus specialist reviews; Efficacy plan with exit criteria; Risk mitigations with named controls; Bias / equity and privacy impact assessment; Stakeholder consultation (including those affected); Ongoing monitoring, reviewed at least annually |
| 25–30 | Critical Risk / Impact Extended review, or reconsider | Executive sponsors (e.g., CIO, provost, VP) on council recommendation | As long as needed; time-limited approval | First ask: does a lower-risk alternative achieve the goal?; Affected groups consulted; Alternatives analysis and algorithmic impact assessment; Legal, privacy, and compliance review; Phased pilot with explicit scale-or-stop gates; Documented rollback / contingency plan; Continuous monitoring, audit trail, and incident response |
03Application domains
Tag every initiative with one primary domain, chosen from these four:
- Teaching, Learning & Student Success
- Research & Innovation
- Operations & Infrastructure
- Advancement & Public Service
Optional: Clinical Care & Health Services. Patient care and clinical services under health-system governance. Use it only where a campus operates a health system; otherwise it is out of scope. Medical education stays in Teaching, Learning & Student Success, and clinical research stays in Research & Innovation.
- Test by who benefits, in order. A learner's progress goes to Teaching, Learning & Student Success; new knowledge goes to Research & Innovation; an external party or public benefit goes to Advancement & Public Service; everything else goes to Operations & Infrastructure.
- Place capabilities, never offices. One office can work in several domains.
- One primary domain per capability. Secondary tags support search and shared work but never count in rollups. Count each initiative and its spend once.
- Record each platform once. Link each use of a platform to its own outcome and risk review.
“Student & Community Engagement” is retired. Its enrollment and student-service work moved to Teaching, Learning & Student Success, and its external work moved to Advancement & Public Service. See Application Domains for capabilities and edge cases.
04Migrating older scorecards
Version 1.0 scored five dimensions with totals from 5 to 25. Because Efficacy Risk adds a sixth score, v1.0 totals cannot be compared directly with v2.0 totals or bands.
- Score Efficacy Risk for each v1.0 initiative using the anchors above.
- From v2.1: treat existing Efficacy Risk scores as proposal scores, and re-check them against the revised anchors. Re-score with pilot results before any decision to scale.
- From v2.0: convert each Efficacy score to 6 minus the old score (5→1, 4→2, 3→3, 2→4, 1→5). Totals and bands don't change. The online Compass does this once when it opens a v2.0 scorecard.
- If evidence can't be reviewed yet, enter a provisional Efficacy Risk of 3 and record that the evidence has not been assessed. The online Compass applies this default when it opens a five-dimension scorecard.
- Re-check the band on the new 6–30 total. Do not convert v1.0 totals by formula.
- Re-review consequential uses first, starting with initiatives that affect student or employee decisions.
- Keep the v1.0 record and note the version on every scorecard so decision history stays readable.
05Version history
| Version | Effective | Change |
|---|---|---|
| 2.2 | November 21, 2026 | Efficacy Risk is scored at two points: a proposal score (outside evidence plus the quality of the evaluation plan) and a scale score (campus pilot results). Anchors are revised so a new idea with a solid plan scores about 3. The scaling block applies to the scale score. |
| 2.1 | November 21, 2026 | Efficacy is scored as Efficacy Risk (Evidence Gap), in the same direction as the other five dimensions. The total is a plain sum. Escalation is on a 5 for Efficacy Risk, not a 1. |
| 2.0 | November 21, 2026 | Adds Efficacy & Evidence of Effectiveness as a sixth dimension, rated as evidence strength (5 = strong) and counted as 6 minus the score in the total. Adds one escalation rule covering ethics, stakeholder impact, and efficacy, and sets deciders, target times, and documentation for each band. Totals change to 6–30. |
| 1.0 | Before November 2026 | Five dimensions, totals from 5 to 25. |