# Risk Scoring Overview > How Risk Scoring rates each customer's risk with a template of weighted factors and rules, and how scores, risk levels, and history flow into your Workflows. Source: https://help.withpersona.com/articles/Pf3414g87aCZCtIXWxsgR/ Section: Risk Scoring > Introduction > Key Concepts ⚠️ Risk Scoring is currently in beta. For more information and to request access, please reach out to your Persona account team. ## Overview **Risk Scoring** is Persona's solution for rating the risk of your customers. You configure a **risk score template** — the weighted factors and rules that reflect your compliance program — and Risk Scoring calculates a score and risk level for each customer, keeping a complete record of every calculation and the exact template version that produced it. Use Risk Scoring to rate customers during onboarding (KYC/KYB), re-rate them on a schedule for periodic reviews, and route customers down different paths based on their risk level. ![The risk score template editor showing the weighted sum calculation method, factor sections for Country of incorporation, Years in business, and Cross-border product exposure, and the rules within a factor](https://assets.withpersona.com/f_auto,q_auto/help-center/articles/risk-scoring/risk-score-template-factor-rules.jpg '2026x950') _A risk score template with weighted factors and the rules that score each one._ ## Where can you find Risk Scoring? Risk Scoring lives in the Persona Dashboard under **Apps → Risk Rating → Risk Score Templates**. Because Risk Scoring is in beta, you won't see it until it has been enabled for your organization — reach out to your Persona account team to request access. ## Key concepts ### Building a template - **Risk score template** — the scoring rulebook: the inputs you collect, the factors you evaluate, the rules within each factor, and how factor results combine into an overall score and risk level. - **Input** — a piece of customer data the template evaluates, such as country of residence, industry, or expected transaction volume. Each input has a type (for example, a number, a country, or a yes/no value) and a missing-data policy that decides what happens when no value is supplied. - **Factor** — one dimension of risk, evaluated from one or more inputs. Each factor has a weight that determines how much it contributes to the overall score. Factors can be organized into categories (for example, Geography or Business Profile). - **Rule** — a condition within a factor. Rules are evaluated in order, and the matching rule sets the factor's risk level. A rule can also be marked **auto-high**, which raises the customer's overall rating to high regardless of the weighted score. - **Risk level** — the final rating (for example, low, medium, or high). The overall score falls into threshold bands you define, and each band maps to a risk level your Workflows can branch on. - **Reference set** — a reusable list the template looks values up in, such as a country risk classification. Reference sets are frozen into each published version, so a later list change never silently changes historical scoring. ### Results and history - **Version** — a published snapshot of a risk score template. Published versions never change. Every score records the version that produced it. - **Calculation run** — one scoring event: the inputs that were evaluated, the resulting score and risk level, and when the calculation happened. - **Override** — a reviewer's manual adjustment to a score, recorded as a new result that points back to the calculated one. The original calculation is never modified. - **Simulation** — a dry run that previews what a customer would score under a draft or published version. Simulations are never saved to history. ## How Risk Scoring works 1. **Build a risk score template.** Define your factors, the rules within each factor, their weights, and what should happen when a piece of customer data is missing. 2. **Publish a version.** Publishing locks the template into an immutable version. To change how scoring works, publish a new version — existing scores keep pointing at the version that produced them. 3. **Score customers.** Add the **Run customer risk rating** step to a Workflow, choose your published template, and map customer data to the template's fields. The step returns the score and risk level as Workflow variables. 4. **Branch on risk level.** Route customers based on their rating — for example, approve low-risk customers automatically and open a Case for high-risk ones. 5. **Review and override.** Reviewers inspect the full calculation history and, when justified, record an override that becomes the customer's current rating. ## Who uses Risk Scoring? **Builders** design and maintain risk score templates: they define factors and rules, test drafts with simulations before publishing, and publish new versions as your risk policy evolves. **Reviewers** work with results: they see each calculation's inputs, score, and the version that produced it, and can record an override with a documented justification when a rating needs manual judgment. ## What gets recorded Risk Scoring is built for auditability: - Every calculation run stores its inputs, its result, and the exact template version — including an integrity hash that proves the version hasn't changed since it was published. - Overrides never replace calculated results. An override is a new result linked to the calculation it adjusts, so the machine-calculated score is always preserved. - There is always exactly one **current** result, so your Workflows and reviewers never see an ambiguous rating. Older results remain in history. - An audit log records template publishes and overrides. ## FAQ ### Can I edit a published version? No. Published versions are immutable. To change scoring, edit the draft and publish a new version. Scores calculated under earlier versions are unaffected and keep their reference to the version that produced them. ### What happens when customer data is missing? Each factor has a missing-value policy that you choose when building the template — for example, treat the factor as highest-risk, use a default rule, or fail the calculation so it can be retried with complete data. ### Do simulations affect a customer's rating or history? No. Simulations are previews only. They can run against drafts and published versions alike, and nothing is saved. ### Does an override erase the calculated score? No. The override becomes the current rating, but the calculated result it adjusts stays intact in history, linked to the override.