Insights
Governed Action Rate: the AI metric your board will ask for
Mark Tack, Founder of RealMandates and author of Marketing Rebuilt · October 31, 2026 · 5 minutes readWhy boards need one number
Boards are starting to ask about AI the way they ask about cybersecurity and financial controls. They do not want a tour of the technology. They want to know whether it is under control. Most companies answer with a list of projects and policies. That does not tell a board whether the AI acting on the company's behalf is accountable.
The Governed Action Rate turns that question into a number.
What it measures
Governed Action Rate = In-scope observed consequential actions that meet defined governance requirements ÷ All observed consequential actions in that same scope × 100.
Governance coverage describes how much of the organization's identified consequential agent activity is included in that governed measurement scope. A 100% GAR with limited coverage is not equivalent to 100% enterprise-wide governance. These are proposed management metrics and should be validated against actual system instrumentation before being represented as measured product outputs.
A consequential action is any agent action that reaches a customer, spends money, changes a system of record, or speaks for the company.
A governed action is one taken under a mandate, with a named person accountable for it, and with a human decision first where the mandate requires one.
Routine actions that stay inside an agent's mandate and carry no consequential effect are not counted, so the metric stays focused on what matters.
An illustrative calculation
Illustrative example
In one month, a marketing team's agents take 4,000 actions. Of these, 1,200 are consequential. 1,140 were taken under a mandate with a named owner. 60 were not, mostly from an agent a regional team set up without one. The fix is not to stop the regional agent. It is to give it a mandate and an owner.
In this illustrative example, 1,140 of 1,200 observed, in-scope consequential actions met the governance requirements, producing a GAR of 95%. The organization would also need to assess whether the observed population includes all relevant agents and workflows. The figure does not establish enterprise-wide coverage.
Why the goal is 100 percent
Most metrics have targets below perfect. This one should not, because the target is not about how often agents are right. It is about whether a person is accountable for every consequential action. Agents will make mistakes, just as people do. The goal is that every mistake has an owner who can fix it.
100 percent is the standard RealMandates sets, not an industry benchmark.
How to start reporting it
- List the agents acting on your company's behalf, including ones teams set up without central review.
- Define which of their actions are consequential.
- For each agent, confirm a mandate and a named owner.
- Report the rate monthly, with the ungoverned actions listed by agent and team.
- Treat any number below 100 percent as a short list of agents that need mandates.
The bottom line
A board does not need to understand how AI agents work to govern them. It needs to know that every consequential action has an owner. The Governed Action Rate tells it.
Frequently asked questions
Written by Mark Tack. Examples labeled illustrative are composites created to explain an idea; they do not describe a specific company.