Should we let AI score employee performance using emails, calls, and task activity?

Question

Tyler Smith · Accepted Answer

Do not turn communication activity into an automatic judgment of employee performance. AI can help organize evidence, but an inferred score can hide missing context and make a questionable conclusion look objective.

EOS places importance on clear roles, accountability, and individual contributions. Start with the outcomes a role owns and the expectations already communicated to the person. Email volume, speaking time, and task counts are not substitutes for those outcomes.

Before using AI in a performance process, establish boundaries:

- Specify which information may be used and why it is relevant.
- Explain the process to employees before collecting or analyzing information for this purpose.
- Restrict access and set retention rules for sensitive material.
- Require a manager to verify factual claims against the underlying evidence.
- Give the employee an opportunity to correct errors and explain context.
- Keep consequential employment decisions with accountable people.

A safer starting point is an employee-reviewed summary of agreed goals, completed work, and unresolved obstacles. The manager can use that summary to prepare for a conversation, not to avoid one. Obtain appropriate HR and legal review before introducing monitoring or automated evaluation, since requirements vary by jurisdiction and use.

For a Batteries Included build, request traceable evidence and correction controls rather than an opaque performance leaderboard.

The recommendation is to use AI to improve preparation and factual accuracy. Do not use it to outsource fairness, interpret someone's motives, or replace the direct conversations that build a healthy organization.