Security Metrics for Boards: What Directors Ask in 2026

Six security metrics a board can act on, each tied to a cost, a downtime figure or evidence a director can check. Includes a method for putting dollar ranges on cyber risk.

By the Total 360 Security team14 min read
Closed printed board pack on a stone windowsill beside folded reading glasses, with a blurred North Texas skyline behind it

A November board pack that opens with blocked firewall connections gets the same response from directors every time: they want to know what it means for the business. Directors reject numbers that do not connect to a decision. Every security metric worth putting in front of a board should answer one of three questions: what could this cost us, how long could a critical process be down, and can we prove we did what we said.

If your draft has a slide of blocked emails, a count of scanned devices or a patch tally, you are not short of data. You are short of translation. The rest of this article shows how to turn the numbers your IT team already produces into statements a director can act on, and how to present them in a way that holds up at the audit committee.

Why do boards reject a slide of blocked emails?

Why do boards reject a slide of blocked emails?, Total 360 Security

A blocked-email count measures attack volume. It does not measure how well your organization performs. A director looking at a number in the hundreds of thousands can reasonably conclude one of two things: either the company is under heavy attack, which is alarming, or the company is blocking things, which is reassuring but tells them nothing about whether anything bad got through or who would respond if it did. Either way, the slide invites a question the presenter cannot answer from the slide.

The scale of the volume is real. Texas Woman's University's information security program reports that of roughly 150 million emails delivered to university accounts in a single year, 62 million were identified as threats, and about 72,000 threats were quarantined on university-owned devices. Those figures make the point well. They show the pressure the institution is under. None of them says how many threats reached a person who could be deceived, whether any credentials were entered, how long a compromise would have run before someone noticed, or what it would have cost. That is the information a board needs.

A director needs a statement about exposure and response instead. Here is the shape, with your own figures and dates in place of the brackets: "[Number] phishing messages reached staff in [function] last quarter. [Number] were reported within [time], none led to a payment, and our tested response time for a suspected payment-diversion attempt is [time], verified in the [month] exercise." That sentence carries a decision (is the response time acceptable?), an evidence date and an owner. The blocked-email count can stay in the appendix for anyone who wants it.

Which six metrics answer what directors are actually asking?

Search results for this topic often list metrics as they are named in security operations centers. Those names are useful, and the ones below keep them so the terms match what your team already reports. What changes is the second column: each metric is restated as the sentence the board hears. Use the table to build your pack. If a metric does not map to one of the three director questions, drop it.

Metric Operational number (what IT produces) Board statement Director question answered
Time to detect (MTTD) Average hours from initial compromise to detection, from the last tested incident A compromise on a critical system could run for this many hours before we see it, based on the last exercise dated [date] How long could a critical process be exposed before we know?
Time to respond (MTTR) Average hours from detection to containment, measured against a named responder Once detected, a named person with authority can cut access within this many hours How long until someone who can act is acting?
Critical vulnerabilities fixed within SLA, with exceptions and oldest critical age Percentage remediated on time, count of approved exceptions, age in days of the oldest open critical item We fixed this share of critical issues on time, we carry this many exceptions, and the oldest open critical item has been exposed for this long Are we keeping up with the risks that matter most?
Hours a critical process is down Logged outage hours for the production or service systems, from the operations log Our most important process was unavailable for this many hours last year, at an estimated cost of this amount What could an outage cost us, and how long could it last?
Annualized loss expectancy (ALE) Single loss expectancy multiplied by the annualized rate of occurrence, using the inputs in the risk register Our expected yearly loss from this scenario is within this dollar range, based on the inputs and sources shown What is this risk worth in money, and is the spend justified?
Dated control evidence against the applicable framework or insurance requirement Number of controls evidenced in the last 90 days, out of the total mapped to the requirement, with open gaps listed We can show evidence for this many of the controls our insurer and regulators expect, and these gaps have dates to close Can we prove we did what we said we would do?

How long could an attacker go unseen, and how long until someone stops it?

Mean time to detect and mean time to respond are real measures, and your team should track them. Directors hear them as questions about speed, and that is the right way to read them. The board's question is how long a compromise could sit unseen, and how long it would take before a named person can contain it.

A current case makes the question concrete. On 27 September 2026, CISA issued an alert about active zero-day exploitation of Citrix NetScaler ADC and Gateway, the enterprise gateways many organizations use for remote access. Whether or not your organization runs that product, the board can ask the same thing of its own perimeter: if a remote access gateway were compromised this week, how long before we would know, and how long before someone with authority could cut the access?

Stating detection time as a board sentence

Take the average from your last real or simulated incident, not a vendor benchmark. Then write the sentence the director will repeat. For example: "For a compromised remote access account, our most recent test found detection in [hours] during business hours and [hours] overnight. That test was run on [date]." The date matters. A number from an exercise eighteen months ago is a weaker claim than one from last quarter, and directors will ask. If you have not tested the claim, say so and put a test on the roadmap with a date.

Stating response time with a named owner

Response time is only credible when someone is named. Write the role, not just the team: "Name the person who can authorize isolation of a remote access gateway, then state the time it takes them to do it: [role] can authorize isolation of a remote access gateway within [minutes] at any hour." Then state what that person can actually do, such as disable accounts or take a system offline, and whether they have the authority to do it without waiting for a committee. The incident response plan should name these people. If it does not, fix the plan before the board meeting. A useful reference for that work is our guide to a cybersecurity incident response plan, which covers who isolates systems, who decides on ransom and who calls the insurer.

Why can a high patch percentage hide the system that matters most?

Percentage of critical vulnerabilities remediated within SLA is the metric most likely to mislead a board. A figure of 95 percent sounds excellent. It does not say whether the remaining 5 percent includes the one unpatched internet-facing system that a remote attacker could reach. It also does not show how many exceptions were approved to miss the SLA, or whether those exceptions were reviewed.

Report the percentage with three numbers beside it. The first is the exception count, with the reason and approver for each. The second is the age in days of the oldest open critical item. The third is how many of the open items sit on internet-facing systems. Those three numbers together show whether the high percentage reflects a program or a lucky quarter.

Vulnerability volume is also rising regardless of anyone's effort. Microsoft's 2026 Digital Defense Report counts nearly 40,000 published common vulnerabilities and exposures in the first half of 2026, which puts the year on pace to roughly double prior volumes. A board that sees a rising raw count of open findings will reasonably ask whether the company is falling behind. The answer is to show exposure that matters to the business: how many open items are reachable from outside, how many touch systems that run critical processes, and how many have a known exploit in use. A bigger number without that filter invites alarm and no decision. For how a team ranks findings and closes them, our guide on vulnerability management tools goes into the workflow.

What does an outage and a ransomware blast radius cost in dollars?

What does an outage and a ransomware blast radius cost in dollars?, Total 360 Security

Directors do not need a precise figure from you. They need a method they can check, with inputs that someone accountable has supplied. Give them the method below and let the finance and operations leaders fill in the numbers.

  1. Downtime cost per hour. Take the hourly value of the critical process. For a production line, that is roughly the margin lost per hour of stoppage plus fixed costs that keep running. For a service business, it is the revenue or contract penalty per hour. Estimate this with the people who run operations and finance, because the business owns the cost of an outage and IT does not. Write down who supplied the figure and when.
  2. Blast radius. List the systems, sites and customer data sets that could be encrypted or taken offline together if one credential or one remote access gateway were compromised. Ask whether any plant floor or operational technology system shares a network path with corporate IT. Globally, industrial and manufacturing organizations accounted for 31 percent of recorded ransomware incidents in August 2026, according to Industrial Cyber, so this question is not hypothetical for many mid-market firms in the Dallas–Fort Worth corridor.
  3. Single loss expectancy (SLE). Multiply the asset value by the exposure factor. Asset value is what the affected system or data set is worth to the business. Exposure factor is the share of that value lost in one event, expressed as a percentage. A two-day outage of a production line that is worth a given amount might carry an exposure factor you estimate with operations, and the product is the single-event loss.
  4. Annualized loss expectancy (ALE). Multiply the SLE by the annualized rate of occurrence, meaning how often you expect the event in a year. This is the hardest input and the one most likely to start an argument. Use your own incident history, your insurer's loss information where available, or a low and high range rather than a single point estimate. State which one you used.
  5. Show the range and the inputs. Present the ALE as a range, with each input and its source on the same slide. A finance director will ask where every input came from before looking at the result, and a range built on sourced inputs is more persuasive than a single confident number built on guesses.

These formulas are standard risk-quantification practice, and the method is only as good as the inputs. Keep the register that holds them current, and review it when the operations plan or the systems change.

Downtime cost per hour

Pick the one or two processes whose outage would hurt most, and calculate their hourly cost with operations and finance in the same meeting. Record the hours of unplanned downtime in the last year from operations logs. Multiply. Then ask whether the figure matches what the board thinks an outage costs. The gap between the two is often the most useful thing the exercise produces.

Blast radius and the plant floor

Draw one diagram of what shares credentials, network paths and backup storage. Mark the places where corporate IT can reach operational systems, because those are the paths that turn an office ransomware event into a plant stoppage. The board does not need every host. It needs to know whether a single compromised account could reach the systems that run production, and what the company would do about that. Our manufacturing security page is a starting point for that work.

Single and annualized loss with your own inputs

Build a one-page register with one row per scenario: a ransomware event on corporate systems, a remote access compromise, a supplier outage, a physical intrusion into a server room. For each row, record the asset value, the exposure factor, the single loss, the rate of occurrence (low, high), the annualized loss range and the source of each input. Review it each quarter. The first version will be rough. The fourth version will be the one the board trusts.

How should compliance and insurance evidence be tracked quarter by quarter?

Compliance and cyber insurance deserve their own metric family. A single pass or fail tells a board very little, and insurance renewals in particular turn on evidence that can be shown on a date. Track a score or a control count each quarter using the same method, so directors see direction over time rather than a snapshot.

Before you choose the scoring, check which framework or requirement applies to your data and your customers. Texas state entities report against the Texas Cybersecurity Framework, and Texas Woman's University's factbook shows the kind of comparable score that framework produces: 3.02 for the university, against a 2.54 average for all state agencies and 2.83 for higher-education agencies. A private mid-market firm should confirm with counsel or its customers which framework fits before adopting one, since the state framework is written for state entities. Texas SB 2610 is another requirement to check if you hold sensitive records; our Texas SB 2610 guide sets out the readiness questions.

Choosing the score a board can follow

Pick a method you can repeat without changing the rules. A control count against a named framework works, as does a maturity score on a fixed scale. What matters is that the same controls are assessed the same way each quarter, and that the board sees the trend and the list of open gaps. Changing the method every quarter makes the trend meaningless. If you need a training component, the same discipline applies to employee cyber security training, where insurers look for evidence that behavior changed, not just that a course was completed.

Evidence an insurer will ask for at renewal

Before the renewal application arrives, prepare a short list of the controls you described on the last one. For each, record the date it was last evidenced, the artifact that proves it (a configuration export, a test result, a signed policy, a training record) and any open gap. Then ask your broker which gaps would change renewal terms, and fix those first. A board that sees this list each quarter will not be surprised by the renewal conversation in the fourth quarter.

How should the quarterly report run for the audit committee?

Use one quarterly template, in the same order, every time. Consistency across quarters builds more trust than any single sophisticated dashboard. The sections below are a sound default for the deck.

  • An executive summary that lists the decisions the board must make this quarter, not a list of activity.
  • What changed in exposure since the last report, in the three-question frame: cost, downtime and proof.
  • A risk matrix with dollar ranges from the register, showing the inputs and sources.
  • A peer benchmark only when the source is named, comparable and dated. Benchmarks taken from vendor marketing are not benchmarks, and a director who spots one will discount the whole deck.
  • A remediation roadmap with an owner and a date for every item, and a note on anything that slipped since the last report.
  • A budget justification that ties each dollar requested to the specific risk it retires.
  • Physical and operational risk in the same report. When a director asks about a power outage, a warehouse break-in or a disrupted shipment, that is a security question, and a cyber-only report leaves it unanswered.

Before you spend time on visuals, confirm two things. First, that someone is named as the owner of each metric. Second, that the IT report and the finance report reconcile. If either fails, the problem is governance rather than the dashboard, and no slide will fix it. Texas Cyber Command's "Texas First. Cyber Ready." campaign ran through October 2026, so directors and insurers have a state readiness message in front of them going into the fourth quarter. That is a good reason to have the pack ready before the meeting rather than during it.

This is the work a Virtual CSO does in practice. A vCSO runs the security program across its risk domains, owns these metrics, and reports them to the board in the same frame each quarter, so the executive presenting does not have to translate IT activity into risk on the day. If you want to test whether your current report answers the three director questions, a 30-minute risk discussion is a practical start. Our Virtual CSO program covers how that reporting is run, and you can book a risk discussion to review your current pack.

Frequently asked questions

What are the most important security metrics for a board of directors?

The six metrics in the table above answer the director questions most often: time to detect, time to respond, critical vulnerabilities fixed on time with exceptions and oldest critical age, hours a critical process is down, annualized loss expectancy, and dated control evidence against the applicable requirement. Any metric that does not lead to a decision should be dropped from the pack.

What is the difference between MTTD and MTTR?

MTTD, or mean time to detect, is the average time from a compromise starting to the team detecting it. MTTR, or mean time to respond, is the average time from detection until a responder acts to contain it. Boards hear them as how long something could sit unseen and how long until a named person can stop it.

How do you calculate annualized loss expectancy?

Single loss expectancy equals asset value multiplied by exposure factor. Annualized loss expectancy equals single loss expectancy multiplied by the annualized rate of occurrence. Use your own incident history, your insurer's loss information or a low and high range for the rate, and state which one you used.

How do you present cybersecurity risk to a board?

Open with the three director questions: what could this cost us, how long could a critical process be down, and can we prove what we did. Then present the decisions needed, a risk matrix with dollar ranges, a benchmark only if it is named and comparable, a remediation roadmap with owners and dates, and a budget tied to the risk each dollar retires. Use the same quarterly template each time so directors can follow the trend.

Total 360 Security provides virtual Chief Security Officer, vCISO, and Enterprise Security Risk Management programs for mid-market organizations. They design, run, and report security programs covering all 13 risk domains rather than just offering advisory decks or point solutions. The firm is Texas DPS-licensed with CPP, CISSP, and CISM credentials.

Serving Texas, California, Nevada, New Mexico.