A quarterly skills review needs a change-triggered exception

The 2027 Skills Intelligence Report, released by Docebo on 14 October, says 43% of responding HR and learning leaders review skills data quarterly. Its public research page describes reviewing and updating data on that cadence. The operational question is not whether quarterly is inherently too slow. It is which changes should invalidate a particular assessment before the next scheduled review.

The sponsor reports that Centiment surveyed more than 1,000 HR and L&D leaders across Europe and North America on behalf of Docebo and 365Talents. Additional responses came through Brandon Hall Group’s research community; 11 interviews complemented the quantitative work. In the same study, 75% said they were prepared to reskill or upskill their workforce.

What the denominator can tell us

These are leader reports, not observations of employee proficiency or audited data pipelines. The opened public materials do not provide fieldwork dates, country-level sample sizes, response rates, weighting or a joint table connecting readiness answers with review frequency. Do not subtract the two percentages, infer how many confident leaders have quarterly reviews, or convert a cadence into an expiry rate. The report’s 2027 title is not a 2027 fieldwork date.

A scheduled review could coexist with timely changes to individual records. Conversely, a live integration might repeatedly refresh a weak inference. Neither collection speed nor a populated profile establishes that a person can perform a task. The survey identifies a useful question for local investigation; it does not answer it for an employer.

Separate three clocks

For a small pilot, record three dates separately: when evidence was produced, when it was assessed, and when it was last reconsidered for the current decision. A new database timestamp should not reset the first clock. Nor should moving an employee to another role silently preserve the relevance of the old assessment.

Choose one consequential workflow and predeclare a short list of invalidating changes: a different tool or model version, a materially changed procedure, a new quality threshold, or reassignment to a task with different constraints. These are proposed editorial controls, not findings tested by the survey. Their purpose is to ask for a targeted review when context changes, rather than rescore everyone continuously.

Assign the trigger to a process owner and the reassessment to a qualified reviewer. Until that happens, show the evidence as needing review—not as evidence that the employee has lost ability. A stable task with strong recent evidence may need no extra collection. Where the evidence is missing, record uncertainty rather than a negative skill rating.

Freshness has a cost and a privacy boundary

CIPD’s people-analytics guidance frames analytics around business problems and warns that passive monitoring can be intrusive even where lawful. It recommends transparency and consultation. This is independent practice guidance, not a replication of Docebo’s survey or proof that any refresh cadence works.

Keep the pilot bounded to necessary work evidence, consult affected employees, and provide correction and challenge routes. Do not quietly repurpose communication or activity data for employment decisions. Relevant legal and employment review remains necessary before implementing such a process.

Compare decisions made from the scheduled snapshot with decisions made after triggered review. Record disagreements, unsupported inferences, missed changes and reviewer time. More updates without fewer consequential errors are not a successful pilot. The existing analysis of the skills-visibility confidence gap explains why evidence matters; this new survey adds a narrower question about when its relevance should be reconsidered.

The next decision is a limited refresh-trigger experiment, with a stop rule if extra collection adds cost or intrusion without decision value—not a procurement mandate for a real-time platform.