Most articles on key risk indicators stop at the definition. This one is the library: the indicators practitioners actually put on dashboards in regulated firms, grouped first by risk category and then by sector, each with a metric definition, example threshold logic and the system the number normally comes from.
If you are new to the concept, read how to define, monitor and act on KRIs first. This article assumes you know what a KRI is and need to choose some.
How to read the tables
Every row has four parts:
- Metric definition — the exact calculation. Half of all KRI disputes are two people computing "the same" indicator differently.
- Threshold logic — illustrative green / amber / red bands. They are starting points for calibration, not benchmarks; the firm's own history and risk appetite decide the real numbers.
- Data source — where the figure comes from without manual effort. A KRI nobody can produce monthly without a spreadsheet exercise will be abandoned within two quarters.
- Lead or lag — whether the indicator warns before the loss (leading) or confirms it afterwards (lagging). A good set has both; a set that is all lagging is a loss report with colours. Most rows below are leading indicators of control health; the regulatory ratios are lagging by construction.
KRI examples by risk category
Credit risk
| KRI | Metric definition | Green / Amber / Red | Data source |
|---|---|---|---|
| Watch-list exposure | Exposure on internal watch list ÷ total credit exposure | < 5% / 5–10% / > 10% | Credit system, monthly |
| Stage 2 migration rate | New IFRS 9 Stage 2 balances in period ÷ opening performing book | < 1% / 1–3% / > 3% | Provisioning engine |
| Single-name concentration | Largest group exposure ÷ eligible capital | < 10% / 10–20% / > 20% (large-exposure limit is 25%) | Exposure aggregation |
| 30-day arrears roll rate | Accounts moving from current to 30+ days past due ÷ current accounts | < 0.5% / 0.5–1.5% / > 1.5% | Collections system |
| Unwaived covenant breaches | Facilities with a covenant breach not formally waived | 0 / 1–2 / 3+ | Covenant monitoring log |
| Collateral revaluation lag | Secured exposures with a valuation older than policy allows ÷ secured exposures | < 5% / 5–15% / > 15% | Collateral register |
Market risk
| KRI | Metric definition | Green / Amber / Red | Data source |
|---|---|---|---|
| VaR limit utilisation | 1-day 99% VaR ÷ approved VaR limit | < 70% / 70–90% / > 90% | Risk engine, daily |
| Backtesting exceptions | Days in the trailing 250 where actual loss exceeded VaR | ≤ 4 / 5–9 / ≥ 10 (Basel traffic-light zones) | Risk engine vs P&L |
| Net open FX position | Sum of net open positions by currency ÷ own funds | < 5% / 5–10% / > 10% | Treasury system |
| EVE sensitivity | Change in economic value of equity under ±200bp ÷ Tier 1 capital | < 10% / 10–15% / > 15% (supervisory outlier test) | ALM model, quarterly |
| Stale prices | Positions priced from a quote older than 5 business days ÷ total positions | < 1% / 1–3% / > 3% | Valuation control |
Liquidity risk
| KRI | Metric definition | Green / Amber / Red | Data source |
|---|---|---|---|
| Liquidity coverage ratio | High-quality liquid assets ÷ 30-day net stressed outflows | > 130% / 110–130% / < 110% (floor 100%) | Regulatory reporting |
| Survival horizon | Days until the liquidity buffer is exhausted under the firm's severe stress | > 90 / 30–90 / < 30 | ALM stress model |
| Depositor concentration | Top-10 depositors ÷ total deposits | < 15% / 15–25% / > 25% | Core banking |
| Intraday buffer usage | Peak intraday overdraft or collateral use ÷ available intraday liquidity | < 50% / 50–80% / > 80% | Nostro monitoring |
| Asset encumbrance | Encumbered assets ÷ total assets | < 15% / 15–25% / > 25% | Treasury |
Operational risk
| KRI | Metric definition | Green / Amber / Red | Data source |
|---|---|---|---|
| Operational loss vs budget | Year-to-date gross operational losses ÷ pro-rated annual loss budget | < 80% / 80–110% / > 110% | Loss event database |
| Near-miss ratio | Near misses reported ÷ (near misses + loss events) | > 60% / 40–60% / < 40% — a low ratio means people stopped reporting | Incident register |
| Overdue control tests | Control tests past due date ÷ tests scheduled | < 5% / 5–15% / > 15% | Controls module |
| Aged reconciliation breaks | Open reconciling items older than 5 business days | 0–5 / 6–20 / > 20 items | Reconciliation tool |
| Overdue high-severity audit findings | High-rated findings past their agreed closure date | 0 / 1–2 / 3+ | Audit tracking |
| Undocumented manual workarounds | Critical processes relying on a manual step with no written procedure | 0 / 1–3 / > 3 | Process inventory |
IT and cyber risk
| KRI | Metric definition | Green / Amber / Red | Data source |
|---|---|---|---|
| Critical patch lag | Mean days from vendor release to deployment for critical CVEs on internet-facing systems | < 7 / 7–14 / > 14 days | Vulnerability scanner |
| Phishing click rate | Users clicking in simulated phishing ÷ users tested | < 3% / 3–8% / > 8% | Awareness platform |
| MFA coverage | Accounts with MFA enforced ÷ accounts with remote or privileged access | 100% / 97–99% / < 97% | Identity provider |
| Unrecertified privileged accounts | Privileged accounts not reviewed within 90 days | 0 / 1–5 / > 5 | IAM access review |
| End-of-life systems | Production systems past vendor support with no approved exception | 0 / 1–2 / 3+ | CMDB |
| Backup restore test failures | Failed or untested restores of Tier 1 systems in the quarter | 0 / 1 / 2+ | Backup tooling, test log |
| Critical system downtime | Unplanned minutes of Tier 1 unavailability per month | < 30 / 30–120 / > 120 | Monitoring, incident tool |
Third-party and outsourcing risk (DORA, EBA outsourcing guidelines)
| KRI | Metric definition | Green / Amber / Red | Data source |
|---|---|---|---|
| Critical providers without a tested exit plan | Providers supporting critical or important functions with no exit plan tested in 24 months | 0 / 1 / 2+ | Third-party register |
| Overdue due-diligence reviews | Critical providers whose periodic assessment is past due ÷ critical providers | 0% / 1–10% / > 10% | Third-party module |
| Provider SLA breaches | Material SLA breaches by critical providers in the quarter | 0 / 1–2 / 3+ | Vendor performance reports |
| Provider concentration | Critical or important functions dependent on one provider or cloud region ÷ all such functions | < 25% / 25–40% / > 40% | Register of information |
| Unnotified sub-outsourcing | Sub-contractors supporting critical functions discovered without prior notification | 0 / 1 / 2+ | Contract reviews, attestations |
| Register of information completeness | Mandatory DORA register fields populated ÷ required fields | 100% / 95–99% / < 95% | Register of information |
Compliance and AML
| KRI | Metric definition | Green / Amber / Red | Data source |
|---|---|---|---|
| Overdue periodic KYC reviews | Clients past their risk-based review date ÷ active clients | < 2% / 2–5% / > 5% | Onboarding system |
| Alert backlog aged 30+ days | Transaction-monitoring alerts open more than 30 days | 0 / 1–25 / > 25 | Monitoring system |
| Suspicious-report filing time | Median business days from internal escalation to filing with the FIU | < 5 / 5–10 / > 10 | Compliance case log |
| High-risk client share | High-risk and PEP relationships ÷ active clients | Board appetite band, e.g. < 10% / 10–15% / > 15% | Client risk rating |
| Unresolved screening hits | Sanctions or adverse-media hits open more than 2 business days | 0 / 1–5 / > 5 | Screening tool |
| Mandatory training completion | Staff completing AML and conduct training by deadline ÷ staff required | 100% / 95–99% / < 95% | Learning system |
| Regulatory breaches notified | Breaches notified to the regulator in the rolling 12 months | 0 / 1 / 2+ | Breach register |
People and HR risk
| KRI | Metric definition | Green / Amber / Red | Data source |
|---|---|---|---|
| Key-role turnover | Voluntary leavers in designated key roles ÷ key-role headcount, rolling 12 months | < 10% / 10–20% / > 20% | HR system |
| Critical vacancy duration | Days a control-function or licensed role has been vacant | < 30 / 30–90 / > 90 | HR system, recruitment |
| Single points of failure | Critical activities with only one trained person | 0 / 1–3 / > 3 | Skills matrix |
| Overdue mandatory training | Staff with any mandatory module overdue ÷ headcount | < 3% / 3–10% / > 10% | Learning system |
| Whistleblowing and grievance volume | Cases opened in the quarter per 100 staff | Trend-based: a sustained zero is amber, a step change is red | Case management |
| Sustained overtime | Staff averaging more than 10 overtime hours a week for 8+ weeks | 0 / 1–5 / > 5 | Time recording |
Strategic risk
| KRI | Metric definition | Green / Amber / Red | Data source |
|---|---|---|---|
| Client revenue concentration | Top-3 clients ÷ total revenue | < 25% / 25–40% / > 40% | Finance |
| Budget variance | Actual vs budgeted operating result, year to date | within 5% / 5–15% / > 15% adverse | Finance |
| Strategic project slippage | Board-sponsored initiatives more than one quarter behind baseline ÷ initiatives | 0% / 1–25% / > 25% | Portfolio tracker |
| Client attrition | Clients lost in the trailing 12 months ÷ opening client count | < 5% / 5–10% / > 10% | CRM |
| Overdue product reviews | Live products whose post-launch review is past due | 0 / 1–2 / 3+ | Product governance log |
ESG and climate risk
| KRI | Metric definition | Green / Amber / Red | Data source |
|---|---|---|---|
| Emissions vs pathway | Scope 1 and 2 tCO2e year to date ÷ transition-plan trajectory | ≤ 100% / 100–110% / > 110% | Sustainability data |
| Transition-sensitive exposure | Lending or investments in high-transition-risk sectors ÷ portfolio | Appetite band, e.g. < 10% / 10–20% / > 20% | Portfolio tagging |
| Physical-risk asset share | Collateral or own assets in high flood, heat or wildfire zones ÷ total | < 5% / 5–15% / > 15% | Geospatial overlay |
| ESG data coverage | Counterparties or holdings with verified ESG data ÷ total | > 90% / 70–90% / < 70% | Data vendor, own collection |
| Reporting datapoint readiness | Required CSRD/ESRS datapoints with a named owner and a source ÷ required | 100% / 85–99% / < 85% | ESG reporting module |
KRI examples by industry
The category tables apply everywhere. The sector tables add the indicators a supervisor in that sector will expect to see on your dashboard.
Banks
| KRI | Metric definition | Green / Amber / Red | Data source |
|---|---|---|---|
| CET1 headroom | CET1 ratio minus overall capital requirement including Pillar 2 guidance | > 300bp / 150–300bp / < 150bp | Regulatory reporting |
| Loan-to-deposit ratio | Net loans ÷ customer deposits | < 90% / 90–110% / > 110% | Core banking |
| Cost of risk | Annualised impairment charge ÷ average gross loans | within plan / plan + 25bp / plan + 50bp | Finance |
| Complaints per 1,000 accounts | Complaints received in the quarter ÷ active accounts × 1,000 | Trend-based: 20% quarterly rise is amber, a regulator-escalated complaint is red | Complaints system |
Insurers (Solvency II)
| KRI | Metric definition | Green / Amber / Red | Data source |
|---|---|---|---|
| SCR coverage ratio | Eligible own funds ÷ Solvency Capital Requirement | > 160% / 130–160% / < 130% (floor 100%) | Solvency II model |
| Combined ratio | (Claims + expenses) ÷ earned premium | < 95% / 95–100% / > 100% | Actuarial, finance |
| Prior-year reserve development | Movement in prior-year claims reserves ÷ opening reserves | Release or < 2% strengthening / 2–5% / > 5% strengthening | Reserving |
| Reinsurer concentration | Recoverables from the largest reinsurer ÷ own funds | < 20% / 20–35% / > 35% | Reinsurance system |
| Lapse rate | Policies lapsed in period ÷ policies in force, annualised | within pricing assumption / + 2 points / + 5 points | Policy administration |
Investment firms (MiFID II, IFR/IFD, ICARA)
| KRI | Metric definition | Green / Amber / Red | Data source |
|---|---|---|---|
| Own-funds headroom | Own funds ÷ the higher of fixed-overhead, permanent-minimum and K-factor requirement | > 150% / 120–150% / < 120% | ICARA, regulatory reporting |
| Client-money reconciliation differences | Unexplained differences in the daily internal client-money reconciliation | 0 / resolved within 1 day / unresolved after 1 day | Client-money reconciliation |
| Best-execution exceptions | Orders flagged outside tolerance by execution monitoring ÷ orders sampled | < 1% / 1–3% / > 3% | Execution monitoring |
| Suitability file exceptions | Advised or managed accounts with incomplete or stale suitability assessment ÷ accounts | < 2% / 2–5% / > 5% | Periodic review workflow |
| Wind-down trigger proximity | Liquid resources ÷ wind-down cost estimate from the ICARA | > 150% / 110–150% / < 110% | ICARA |
Payment and e-money institutions
| KRI | Metric definition | Green / Amber / Red | Data source |
|---|---|---|---|
| Safeguarding reconciliation gap | Relevant funds owed to customers minus safeguarded balance | 0 / shortfall covered same day / shortfall carried overnight | Safeguarding reconciliation |
| Fraud rate | Value of fraudulent transactions ÷ total transaction value, in basis points | < 5bp / 5–13bp / > 13bp (13bp is the PSD2 reference rate for the lowest exemption band) | Fraud monitoring |
| Chargeback ratio | Chargebacks ÷ transactions, monthly, per card scheme | < 0.5% / 0.5–0.9% / > 0.9% (scheme monitoring programmes start near 1%) | Acquirer, scheme reports |
| Authorisation success rate | Approved authorisations ÷ attempted | > 95% / 90–95% / < 90% | Processing platform |
| Monitoring alert backlog | Transaction-monitoring alerts unactioned after 5 business days | 0 / 1–50 / > 50 | Monitoring system |
Fund managers (AIFMD, UCITS)
| KRI | Metric definition | Green / Amber / Red | Data source |
|---|---|---|---|
| NAV errors | Pricing errors above the materiality threshold in the quarter | 0 / 1 / 2+, or any requiring investor compensation | Fund administrator |
| Liquidity mismatch | Days to liquidate 50% of the portfolio under stress vs redemption notice period | within notice period / up to 2× / > 2× | Liquidity model |
| Investment restriction breaches | Breaches of prospectus or regulatory limits in the quarter | 0 / 1–2 passive / any active breach | Pre- and post-trade compliance |
| Leverage vs limit | Commitment-method leverage ÷ disclosed maximum | < 70% / 70–90% / > 90% | Risk system |
| Depositary queries outstanding | Depositary oversight queries open more than 10 business days | 0 / 1–3 / > 3 | Depositary correspondence log |
Fiduciary and corporate service providers
| KRI | Metric definition | Green / Amber / Red | Data source |
|---|---|---|---|
| Overdue client file reviews | Client structures past their annual or risk-based review date ÷ active structures | < 3% / 3–10% / > 10% | Client administration system |
| Incomplete beneficial ownership | Structures where UBO identification or verification is incomplete | 0 / 1–5 / > 5 | KYC records |
| Missed statutory deadlines | Annual returns, financial statements, FATCA/CRS or UBO-register filings missed in the quarter | 0 / 1–2 / 3+ | Compliance calendar |
| Unresponsive clients | Structures where the client has not answered information requests for 90+ days ÷ structures | < 2% / 2–5% / > 5% | Correspondence log |
| Fees receivable over 90 days | Receivables aged 90+ days ÷ total receivables, a proxy for disengaged or distressed clients | < 10% / 10–20% / > 20% | Finance |
Leading versus lagging: getting the mix right
A lagging indicator measures the consequence: losses booked, breaches reported, complaints received. It is reliable and easy to source, and it tells you nothing you could have acted on. A leading indicator measures a cause or a precondition: overdue reviews, patch lag, staff turnover, alert backlogs. It is noisier, but it moves first.
The useful question for each top risk is: what would have been visible three months before this risk crystallised? Most answers are leading indicators of control health — things being late, piling up or going untested. Pair each with one lagging indicator so you can check, after the fact, whether the leading one actually warned you. If a leading indicator has never moved before an event, replace it.
A workable ratio for a board-level set is roughly two-thirds leading, one-third lagging. Regulatory ratios (LCR, SCR coverage, own-funds headroom) are a special case: lagging in construction but carrying mandatory consequences, so they stay on the dashboard regardless.
Setting thresholds that mean something
Three methods, in order of preference:
- Appetite-anchored. The board has said how much of something it will tolerate — a concentration cap, a maximum loss, a minimum capital buffer. Red is the appetite limit; amber is a buffer below it that gives management time to act, typically 10–20% of the limit; green is everything else. This is the only method that makes the KRI a genuine appetite monitor.
- Regulatory-anchored. Where a supervisor sets a floor, red should sit above the floor, not on it. Breaching a regulatory minimum is not an early warning; it is the event. Treat the floor plus the time needed to restore it as red.
- History-anchored. Where neither applies, use the firm's own distribution: amber above the 75th percentile of the last 24 months, red above the 90th. Revisit after a year; percentile thresholds drift as the business grows.
Whichever method you use, write down who may change a threshold and how. A threshold a risk owner can quietly loosen the month it turns red is not a threshold. Make amber mean something specific — usually "the owner investigates and reports at the next committee" — and red mean escalation to a named forum within a defined number of days. Some indicators need two-sided thresholds: a near-miss ratio or a whistleblowing count that drops to zero is as much a warning as one that spikes.
Reporting cadence and ownership
| Indicator type | Measured | Reported to |
|---|---|---|
| Market, liquidity, client money, safeguarding | Daily | Treasury or finance; breaches escalated the same day |
| Cyber, third-party SLAs, alert backlogs, reconciliations | Weekly or monthly | Operational risk committee |
| Credit, operational loss, compliance, HR | Monthly | Executive risk committee |
| Strategic, ESG, capital headroom | Quarterly | Board risk committee |
Report trends, not points. A KRI at 8% means little; a KRI that has moved from 3% to 8% over four months while staying green is the story. Each indicator needs one accountable owner, a documented calculation, a linked risk in the risk register and a linked control where one exists, so a red light tells you which control to go and look at. A platform such as ActiveERM holds the KRI, its thresholds and its history against the risk it monitors, so the dashboard and the register move together rather than being reconciled by hand before each meeting.
Starting a KRI library
- Start with 15–25 indicators covering the top ten risks, not 70. The tables above are a menu, not a target.
- For every candidate, confirm somebody can produce the figure monthly from a system. If not, automate the source or drop the indicator.
- Pilot thresholds for two quarters before presenting them to the board as final.
- Review the whole set annually and whenever a loss event occurs that no indicator predicted. That gap is the most valuable information the programme produces.
Supervisory expectations for indicators sit inside the broader risk frameworks summarised on our risk regulations page.
Frequently asked questions
How many KRIs should a small regulated firm have?
Between 15 and 30 at board level is typical. Below that, coverage of the top risks is thin; above it, the dashboard stops being read. Operational teams can track more detailed indicators beneath them.
What is the difference between a KRI and a KPI?
A KPI measures how well an objective is being achieved; a KRI measures how likely it is that something will prevent it. Staff turnover is a KPI for HR and a KRI for operational risk — the same number, read for a different purpose. Many KRIs are KPIs with a risk threshold attached.
Should thresholds be the same across business units?
The calculation should be identical; the thresholds may differ if the units have genuinely different risk profiles and the board has accepted that. Document the reason. Unexplained differences are usually negotiation, not calibration.
How often should KRI thresholds be reviewed?
Annually as a set, plus whenever the appetite statement changes, a regulatory limit changes, or a loss event shows the indicator did not warn. Recalibrating individual thresholds more often than that usually means the indicator was poorly chosen.
Can a KRI be qualitative?
Yes, if the scoring is defined. A quarterly 1–5 assessment of the regulatory relationship, scored against written criteria, is a legitimate KRI. An unexplained "amber" from a manager is not.