← All Metrics
AI & Automation

AI-Assisted Resolution Rate

Also known as: AI Intervention Rate, AI-Touched Outcome Rate

The percentage of CS-initiated interventions in a period where AI played a material role in the outcome, including drafting the response, surfacing the risk signal, generating the recommendation, or taking a bounded autonomous action. Measures how much of CS delivery is being augmented or automated by AI.

Formula

(CS interventions with documented AI involvement / Total CS interventions in period) × 100

Who Is This Metric For?

VP/Director of CS

Use to measure AI program ROI and set targets for expanding AI coverage across CS workflows.

CS Ops

Track and report monthly. Own the definition of 'material AI involvement' and ensure consistent tagging across workflows.

CS Manager

Compare AI-assisted vs. unassisted outcomes by CSM and account segment to identify where AI adds the most value.

Priority by Stage

Crawl low

Don't measure this yet. You're still deciding whether to use AI tools at all, and tracking a metric for a capability you haven't adopted creates noise.

Walk low

Start tracking informally once you have a pilot (call summarization, health scoring copilot). Establishes a baseline before scaling.

Run medium

Track formally. At this stage AI is embedded in multiple workflows, so you need to know the coverage and whether AI-assisted interventions produce better retention outcomes.

Benchmarks

SegmentGoodGreatWorld Class
SMB20–40%40–65%65%+
Mid-Market15–35%35–60%60%+
Enterprise10–25%25–50%50%+

Common Mistakes

  1. Counting AI tool usage (e.g., CSM opened the copilot) instead of AI-materially-influenced outcomes. Activity is not impact
  2. Including interventions where AI was used but the output was completely rewritten. Set a minimum threshold for 'material involvement'
  3. Optimizing this metric without measuring outcome quality. A high rate with declining CSAT or GRR means AI is creating noise, not value
  4. Not distinguishing between AI-drafted content, AI-surfaced signals, and autonomous AI actions. These have very different reliability profiles

Used in Playbooks

Related Metrics

Start typing to search the framework.