The risk register is updated once a year; risk changes every day. Between two assessments, vulnerabilities pile up, the number of accounts without MFA grows, backup tests aren’t done for months, and the risk register still says “medium”. Key risk indicators (KRIs) exist to close this gap: measurable signals that warn the risk is growing before the register is updated. Below we separate KPIs from KRIs, choose leading and lagging indicators for information security, and tie down who makes which decision when a threshold is crossed.
The difference between KPI, KRI and KCI
The three abbreviations get mixed up all the time, because they’re often fed from the same data source. What separates them is the question the metric answers.
| Indicator | Question it answers | Example | Typical owner |
|---|---|---|---|
| KPI | Are we reaching the target? | Number of security requests closed per month | Process owner |
| KRI | Is the risk growing? | Number of critical vulnerabilities older than 30 days | Risk owner |
| KCI | Is the control working? | Rate of access reviews completed on time | Control owner |
The simple test we use in the field: if the first reaction when the indicator gets worse is “the team should work harder”, what you have is a KPI. If the first reaction is “our exposure has increased; are we going outside the level we accepted?”, it’s a KRI. A KCI sits between the two; it shows early that a control is faltering in operation, and is often a leading indicator for a KRI. If access reviews are skipped for two quarters, a few months later the unauthorized access KRI starts to move too.
Information security objectives define what the management system wants to achieve; we explained that side in the setting information security objectives guide. A KRI is independent of the objective. It tracks the direction of the risk profile.
Leading and lagging indicators
Lagging indicators are counted after an event happens: number of security incidents, data breach notifications, unplanned outage time. They’re valuable because they show real damage, but they come late. Leading indicators measure the conditions that increase the likelihood of an event; they leave room to intervene before damage occurs.
- Lagging: Number of accounts compromised through phishing in the last quarter.
- Leading: Number of active users who haven’t completed MFA registration; the rate of users who click the link in a phishing simulation and don’t report it.
Put at least one leading indicator on every significant risk. A dashboard built only from lagging indicators tells management about the past and leaves no time for decisions. Don’t throw lagging indicators out of the set either: they confirm whether the leading indicators were chosen correctly. If the leading side is green while the lagging side gets worse, you’re measuring the wrong thing. That’s also a picture you may come across.
Characteristics of a good KRI
Before putting any candidate indicator on the dashboard, run it through this list:
- Linked to a risk: It refers to a specific row in the risk register. An indicator that can’t be linked to a risk isn’t a risk indicator.
- Measurable and repeatable: It’s calculated from the same data source, with the same query, by the same method every period. The calculation method is written down.
- Can be collected automatically: An indicator compiled by hand falters in the first busy month. The data should come from a system report or a query.
- Has thresholds: Green, amber and red zones are defined in advance and consistent with the risk acceptance criteria.
- Triggers decisions: When a threshold is crossed, it’s clear who is informed and which decision is made.
- Resistant to manipulation: The number improving means the risk really decreased. “Number of open vulnerabilities” also improves when scanning scope is narrowed; that’s why you put the number of scanned assets next to it.
Sample information security KRI table
The set below is an example prepared for a mid-sized, Microsoft-heavy infrastructure. Thresholds are examples only; set them according to your own risk acceptance criteria.
| KRI and related risk | Data source | Example threshold (amber / red) | Escalation |
|---|---|---|---|
| Number of critical vulnerabilities older than 30 days — system compromise through external attack | Vulnerability scanner report | >5 / >15 | IT manager / ISMS committee |
| Rate of active accounts without MFA registration — credential theft | Entra ID registration report | >%2 / >%5 | Identity manager / CISO |
| Number of enabled accounts not signed in for 90 days — unauthorized access | Active Directory query | >10 / >30 | System administrator / CISO |
| Number of Domain Admins members — misuse of privileged accounts | Active Directory query | +1 / +3 over the approved list | CISO / general manager |
| Days since the last successful restore test — data loss | Backup test records | >90 / >180 | Infrastructure owner / ISMS committee |
| Number of expired policy exceptions — risk left outside control | Exception register | >0 / >3 | Risk owner / CISO |
The indicators on the Active Directory side are collected with a single PowerShell script. On a management server with the ActiveDirectory module installed:
Import-Module ActiveDirectory
# KRI-03: enabled user accounts not signed in for 90 days
$inactive = Search-ADAccount -AccountInactive -TimeSpan 90.00:00:00 -UsersOnly |
Where-Object { $_.Enabled }
# KRI-04: members of the Domain Admins group (nested groups expanded;
# -Recursive returns only user/computer objects and doesn't check Enabled status)
$da = Get-ADGroupMember -Identity "Domain Admins" -Recursive
[pscustomobject]@{
Date = Get-Date -Format "yyyy-MM-dd"
InactiveAccount = ($inactive | Measure-Object).Count
DomainAdmins = ($da | Measure-Object).Count
} | Export-Csv .\kri_ad.csv -Append -NoTypeInformation -Encoding UTF8
When you run the script monthly as a scheduled task and append the output to the same CSV, the time series needed for a trend chart builds up by itself. A field note: Search-ADAccount -AccountInactive bases its assessment on the lastLogonTimestamp attribute. With default settings, this attribute isn’t updated at every sign-in but within a 9–14 day window. Read the 90-day threshold with a deviation of about two weeks; inactivity thresholds under 14 days can’t be measured reliably this way. Also record which account the scheduled task runs under. If the task silently stops, so does the indicator.
Thresholds and escalation
There are two ways to set thresholds. The first is to work backwards from the risk acceptance criteria: the red threshold is the point where the risk moves into the unacceptable zone. The second is to start from a few periods of historical data: today’s normal level is set just below amber and tightened over time toward the target. In practice, starting with the second in the first period and tying it to the first as the data settles creates less friction. The link between thresholds and risk appetite is detailed in the risk acceptance criteria and risk appetite guide.
The escalation matrix should define at least these three situations:
- Amber zone: The risk owner is informed, the root cause is investigated and a corrective action is set for the next period. Reporting is routine.
- Red zone: The CISO or ISMS committee is informed within a set time. The related row in the risk register is reassessed; if needed, the risk treatment plan is updated and additional budget or resources are requested.
- Consecutive red: An indicator that stays red for two periods in a row is taken to top management as a separate decision item: either additional resources are provided, or the increased risk is formally accepted and recorded.
The record of the decision made when a threshold is crossed is as valuable as the indicator itself. The auditor’s question will be “what did you do when the indicator turned red?” The answer should be a meeting note, a ticket or an updated risk row. A verbal explanation isn’t evidence.
Indicator dashboard and reporting
The management version of the KRI dashboard is a single page. Each row shows the indicator name, current value, the values of the previous two periods, trend direction, threshold status and the risk it’s linked to. The calculation method and technical detail sit on a separate annex page; management doesn’t read it, the auditor does.
Choose the reporting rhythm according to how fast the indicator changes:
- Monthly: Vulnerability, account and patch indicators; discussed in the IT management meeting.
- Quarterly: A trend summary to the ISMS committee, and the decisions made on threshold breaches.
- Yearly: As an input to management review: the trend over the year, the adequacy of the indicator set and proposed threshold revisions.
It’s tempting to show only red rows on the dashboard. Don’t. Indicators that stay green should be visible too: when management sees where risk is under control, the case for the resources requested for the red rows gets stronger.
Maintaining the indicator set
Indicators go stale over time. Once a year, ask these questions for each KRI: Did this indicator trigger any decision last year? Is the risk it’s linked to still in the risk register, and has its priority changed? Has the data source changed (for example a move from on-premises AD to a cloud identity system)? An indicator that never crosses a threshold either has too loose a threshold or no longer represents a significant risk. An indicator that stays red constantly either carries an unrealistic threshold or points to a risk management isn’t deciding on. Both should be discussed in management review.
As a general framework, KRI design is consistent with the risk monitoring principles of ISO 31000. To set up the indicators together with the risk register and tie them to management review, our management review and KPI design work can be the starting point; if the risk register itself needs revisiting, our information security risk assessment work.
Frequently asked questions
Can a KRI and a KPI be the same metric?
Yes. The patch compliance rate is a performance target for the IT team; the same rate dropping below a certain threshold is a risk signal. The difference isn’t in the metric but in which question it answers and who decides what when the threshold is crossed.
How many KRIs should be defined?
There’s no fixed number. A good measure is that the set going to management fits on one page and that each indicator is linked to a specific risk in the risk register. An indicator that doesn’t trigger any decision comes out of the set.
When are KRI thresholds updated?
When the risk acceptance criteria change, when the related control structure changes, or when the indicator stops producing information by never crossing a threshold for a long time or by crossing it constantly. Do this review at least once a year, together with management review preparation.
Are KRIs mandatory in an ISO 27001 audit?
The standard doesn’t use the term KRI. However, clause 9.1 requires monitoring and measurement; clauses 6.1 and 8.2 require risks to be assessed at planned intervals. Risk-linked indicators are strong evidence that these requirements are met.