The acx report - Chapter 09

The new metrics.

The measurement frameworks most organizations are using to evaluate their AI investments were designed for a different era. NPS was built for human-paced relationship management. Conversion rate was designed for funnel thinking. Cost-per-click optimizes for the wrong lever entirely in an agentic world.

Access Full Report

ACx requires a new measurement language, not because the old metrics are wrong, but because they are insufficient. They measure the outputs of human-scale operations. ACx operates at a different scale, at a different speed and with different levers. The measurement framework has to match what is actually being built.

Three criteria, six metrics

The measurement framework mirrors the three criteria of ACx success.

Chapter Four named the three criteria that separate ACx from AI deployment: Strategy — connecting capability to a defined customer outcome; Delivery Execution — moving from brief to production at the speed the market demands; and Industry Depth — the vertical judgement that makes agentic output commercially meaningful. The metrics that follow are organized the same way. Two for each criterion. Six in total.

01 - Strategy

New metric

90-Day ROI Validation

Revenue lift, cost reduction or efficiency gain that is measurable and attributable within 90 days of deployment. If impact cannot be demonstrated within 90 days, the deployment design was wrong, not the timeline expectation.

Replaces: Annual review / post-project retrospective

New metric

Experience Continuity Score

Measures whether the system carries customer context across sessions, channels and time. A customer who must re-introduce themselves is experiencing a continuity failure. This metric makes the failure visible and attributable.

Replaces: Per-touchpoint NPS / session satisfaction

02 - Delivery Execution

New metric

Time to Value

From project kickoff to working system in production. Not time to recommendation. Not time to presentation. Time to a system operating in a live environment and generating measurable output.

Replaces: Time to recommendation / time to deck

New metric

Velocity Multiple

AI-augmented output volume divided by human-only baseline at equivalent quality. Tracks how much faster the organization produces as AI capability matures. The 10× practitioner is real and measurable.

Replaces: Headcount growth / resource utilization rate

03 - Industry Depth

New metric

AI Contribution Rate

The percentage of customer-facing output — content, recommendations, personalization — generated or materially enhanced by AI systems. Tracks the structural shift from human-first to AI-augmented delivery within a vertical context.

Replaces: AI adoption rate / number of active users

New metric

AI Delivery Rate

Complex customer interactions resolved by agentic AI without human escalation while maintaining or improving satisfaction. Not deflection as cost-cutting. Deflection as proof the agent is genuinely capable of the task it was built for.

Replaces: Cost-per-contact / ticket volume reduction

Adopting this framework changes what organizations ask of their AI partners. It shifts the conversation from capability demonstration to outcome accountability; from "show me what your platform can do" to "here is what you will be measured on in 90 days." That is a harder conversation. It is the only one worth having.

 

We respect your privacy

We and selected third parties use cookies or similar technologies for technical purposes and, with your consent, for other purposes as specified in the cookie policy. Denying consent may make related features unavailable.