SARPI scores are structured self-assessments, not audits. This page explains exactly how answers become pillar scores, an overall score, and a maturity band — so every number in a report can be traced and reproduced.
Last updated 11 August 2026
Each assessment is organised into pillars (for example governance, data foundations, talent, and operating model). Every pillar contains questions answered on a five-level maturity scale. SARPI measures the maturity of the practices an organisation reports having in place — it does not verify evidence, inspect systems, or certify compliance.
1 — Initial: ad hoc, individual-led, no repeatable practice.
2 — Developing: practices exist in pockets but are inconsistent.
3 — Defined: documented, repeatable practices with clear owners.
4 — Managed: measured practices with monitoring and review cadence.
5 — Optimising: continuously improved against outcome metrics.
The same scale is used across every SARPI assessment so results stay comparable across editions and over time.
Each answered question contributes its level to the pillar it belongs to. A pillar score is the weighted mean of its answered questions, normalised to a 0–100 scale. The overall score is the weighted mean of pillar scores; weights reflect the relative contribution of each pillar to safe, value-generating AI adoption and are fixed per assessment version.
Unanswered questions are excluded from the mean rather than scored as zero. Assessments below the required completion threshold are marked incomplete and excluded from benchmark cohorts.
All scoring is recomputed on the server from the raw answers. Client-submitted scores are never trusted.
Overall scores are mapped to maturity bands using fixed thresholds: Initial (0–20), Developing (21–40), Defined (41–60), Managed (61–80), Optimising (81–100). Bands are presentational — the underlying score and pillar detail remain the authoritative output.
Every stored assessment records the methodology identifier, methodology version, and assessment version used at the time of submission. Reports render against the stamped version, so a report generated today reproduces the same numbers next year even after the methodology evolves.
sarpi.ai-readiness · methodology 1.0.0 · assessment 2026.1
sarpi.agentic-readiness · methodology 1.0.0 · assessment 2026.1
sarpi.custom-template · methodology 1.0.0 · assessment 2026.1
Benchmark comparisons only use complete, eligible, tagged submissions, and a cohort is published only once it reaches the configured minimum size. Demo and test records are excluded. Cohort statistics show median and quartiles rather than raw peer records, and no individual organisation is identifiable.
SARPI results are self-reported and therefore reflect perception as much as practice. Scores are not a legal, audit, or certification opinion, do not guarantee regulatory compliance, and do not predict financial outcomes. ROI figures are scenario models based on inputs you supply. Use SARPI outputs to prioritise and direct work — validate with evidence before making irreversible investment or compliance decisions.
Material changes to weighting, thresholds, or question sets increment the methodology version. Historical assessments keep their original stamp and are never rescored retroactively. Re-run an assessment to obtain a score under the current version.
Reach the AIgilityX team at hello@aigilityx.com or through the contact form.
