What Key Performance Indicators (KPIs) does Cal report?

We co-define KPIs for your domain. Common examples: review minutes, cycle time, policy/Standard Operating Procedure (SOP) adherence, time-to-publish communications, incident/near-miss recurrences, and time-to-resolution. Results are reported as medians with interquartile ranges (IQR) and sample size (n).

How do you measure impact on pilots?

We use a 3–6 month baseline or a recent sample, then run either randomised A/B tests or phased rollouts by team/site/case type. We instrument workflows, capture timestamps, and report medians with IQR, plus inclusion/exclusion rules and a methods note you can audit.

Why report medians with interquartile ranges (IQR) instead of averages?

Medians and IQR are robust to outliers and better reflect typical performance in operational data. Where distributions are symmetric, we may also show means—always alongside sample size (n) and study design notes.

How do you protect privacy and governance requirements?

Personally Identifiable Information (PII) is redacted by default; reversible tokens are used only when follow-up needs identity and are protected with role-based access control (RBAC). High-impact actions require Human-in-the-Loop (HITL) approval. Retrieval-Augmented Generation (RAG) answers cite authorised sources, and all prompts/actions are logged for audit.

What exactly do you publish from a pilot?

A single conservative headline median on product pages and a downloadable methods note (PDF) detailing timeframe, sample size (n), inclusion/exclusion rules, statistics reported, and reproducibility details. Public case studies are optional.

Impact methodology

Here we explain what we measure, how we measure it, and how results are published, so you can trust our impact claims.

Talk to us about your KPIs

Large language model with Retrieval-Augmented Generation (RAG), audit trail, and approvals to help regulated teams make faster, safer decisions.

Security and privacy

Awards and recognition

Case studies

What we measure (KPIs)

We publish median improvements. Our core KPI families are (pick those relevant to your domain):

Operations & throughput
Task review minutes, cycle times, first‑time‑right rate, mean time to decision mean time to resolution/recovery.

Quality & compliance
Exception rate, defect/leakage rate to audit, risk assessment and method statement (RAMS)/Standard operating procedure (SOP) adherence where applicable, policy conflicts detected.

Safety & risk
Incident/near‑miss review minutes, recurrence of top themes, control‑gap closures, leading‑indicator scores.

Customer/stakeholder communications
Time‑to‑publish notices/updates, re‑open rate/second‑touch rate, template deviation errors.

We agree the exact KPI list and definitions up‑front.

The KPIs we track: review minutes, cycle time, policy/Standard Operating Procedure (SOP) adherence, communications speed, and more—reported as medians.

Our method: baselines, A/B or phased rollout, medians with interquartile ranges (IQR), Retrieval-Augmented Generation (RAG) citations, Human-in-the-Loop (HITL).

How we measure

Baseline window: Last 3–6 months of comparable work or the most recent items.
Design: Randomised A/B or phased rollout by team/site/case type.
Statistic: We report medians and sample size (n); means only if distribution is symmetric.
Definition of done: Documented acceptance criteria per sprint; Only completed, reviewed items count.
Exclusions: Non‑comparable records (e.g., missing timestamps, training data) are disclosed and excluded from primary stats.

Privacy and governance

Personally Identifiable Information (PII) is redacted by default; reversible tokens used only where follow‑up requires identity.
Human‑in‑the‑Loop approvals for high‑impact actions.
Retrieval‑Augmented Generation (RAG) answers with citations to your authorised sources; all actions and prompts are logged.

Privacy by default: Personally Identifiable Information (PII) redaction, audit trails, governed actions with approvals, and trustworthy Retrieval-Augmented Generation (RAG).

What we publish

A single, conservative headline median on product pages (once validated). A downloadable methods note (PDF) per claim covering timeframe, n, inclusion/exclusion rules, statistic choice, and reproducibility details.

We publish median outcomes with methods notes, sample sizes, and timeframes—downloadable PDFs with citations for procurement and compliance teams.

Frequently asked questions

What KPIs does Cal report?
We co‑define KPIs for your domain; common ones include review minutes, cycle times, adherence to policies/Standard Operating Procedures (SOPs), time‑to‑publish communications, incident/near‑miss recurrences, and time‑to‑resolution.

Why medians and interquartile ranges (IQR)?
They’re robust to outliers and reflect typical performance; we include sample size (n) and design notes.

How do you protect privacy?
Personally Identifiable Information (PII) is redacted by default; when follow‑up is required, we use reversible tokens under role‑based access control (RBAC).

Do you guarantee results?
No. We publish measured outcomes from your environment and methods notes you can audit.

Can we run this on‑premises or in a Virtual Private Cloud (VPC)?
Yes – hosting options include dedicated tenants; expect separate commercials and security review.

Product

Cal AI Agent

Maps & Visualisations

Internal reports

Integrations

Industries

Pharma & Life Sciences

Transport & Infrastructure

Trust

Company

"SIRV helped us move beyond basic reporting into a system that actively supports decision-making". Les O'Gorman, Director of Facilities, UCB - Pharma and Life Sciences

Privacy LinkedIn Medium Cookies Accessibility Terms of use