Frontier · Prompt library

Employee Wellbeing and Workload Signals Prompt

Employee Wellbeing and Workload Signals

← Back to Prompt Library

📋 Review the schema first. Before running this prompt, confirm your column names and the Copilot/agent action taxonomy against the Schema Documentation.

🗂️DataPerson Query 📅Min data4 complete weeks for exposure flag 💻LanguageR or Python 📦OutputStatic HTML report 🎯LevelIntermediate

Purpose

Create a static HTML report of elevated collaboration and connected-time exposure, group differences and possible associated work patterns.

Audience

People analytics practitioners, HR partners and leaders reviewing sustainable working practices.

When to use

When a Person Query contains collaboration and connected-time measures and you want to identify aggregate workload patterns worth reviewing. These measures cannot establish burnout, a clinical condition or how much personal time someone has.

Required inputs

• Person Query exports in the workspace’s data folder, with PersonId, MetricDate and any available organisational attributes.
• Primary raw headers: Collaboration hours, Active connected hours, After hours collaboration hours and Weekend collaboration hours.
• Configured date grouping, metric windows, population and working-hours settings where available.
• Four or more complete weeks per person for the proposed multi-metric exposure flag. Eight or more weeks are recommended for group comparisons and trends.
• Optional verified measures such as meeting, email, chat or focus metrics. Suitable survey outcomes are needed for direct statements about reported wellbeing.

If a primary measure is unavailable, continue with supported component summaries and omit the combined flag.

Assumptions

• High digital collaboration or connected time is an exposure signal, not a diagnosis of burnout or a measurement of total work.
• “Exceptionally high” is a transparent relative threshold with sensitivity checks, not a validated health cut-off.
• The four measures can overlap. They must not be added into total working hours.
• Associations with workload signals are descriptive and may reflect role requirements, schedules or unmeasured factors.
• This is a guided analysis prompt, without a completed end-to-end data trial or a verified wellbeing assessment.

Recommended output

A self-contained output/employee-wellbeing.html report, reproducible R or Python source, privacy-protected aggregates and a validation log. Start with measured workload signals, then explain group patterns, associations and limitations.

Quick prompt (short version)

Use the viva-insights-analysis skill to build a self-contained static HTML report about workload signals relevant to employee wellbeing. Read the Person Query from the workspace's data folder and save all generated files under its output folder, not filesystem-root paths.
Answer: how common are elevated collaboration and connected-time signals, which groups show them more often, and which other recorded work patterns are associated with them?
Prioritise Collaboration hours, Active connected hours, After hours collaboration hours and Weekend collaboration hours. Inspect actual headers, configured cadence, metric windows and working-hours settings. Use only supported measures and actual organisational attributes.
Never describe telemetry as diagnosed burnout, health status, total working hours or measured lack of personal time. Use "elevated workload exposure" for the descriptive flag.
For a complete weekly panel, calculate each person's mean weekly value over a common reference window, requiring at least four complete weeks with all four primary measures observed. Give each person equal weight when calculating each metric's population 90th percentile.
Flag elevated exposure when at least two of the four person-level means are strictly above their respective fixed reference thresholds. Equality does not count. This is a proposed descriptive rule, not a validated risk model. Show the thresholds, missingness and sensitivity at the 85th and 95th percentiles. If all four measures or sufficient weeks are unavailable, omit the combined flag and keep component views.
Do not add overlapping hour measures. Do not fill missing behavioural hours with zero. Keep ineligible or incomplete observations separate from unflagged people.
Compare sufficiently sized groups using prevalence with explicit denominators and component distributions. Use other metrics only for exploratory associations, excluding the flag's defining metrics and their transformations from any claimed driver analysis. Do not present associations as causes or predictions of burnout.
Use the stricter organisational minimum or 10 distinct people for every published cell and its complement, including flagged cohorts, embedded data and downloads. No individual rankings or employee-level risk lists.
Produce a summary-first HTML report with group comparisons, compatible time trends, caveats, an evidence ledger and reproducible source. Embed all assets, verify offline desktop/mobile rendering, and record executed versus skipped checks.

Prompt

Create an Employee Wellbeing and Workload Signals report using the viva-insights-analysis skill and its shared analysis policy. Use installed R or Python tools and prefer supported package functions after checking their actual signatures.

PURPOSE AND CLAIMS
Answer three questions:
1. How common are elevated collaboration, connected-time and out-of-hours exposure signals in this population?
2. Which organisational groups show those signals more often?
3. Which other observed work patterns are associated with elevated exposure?
Do not answer "who has burnout". Digital activity cannot diagnose burnout, establish exhaustion or measure personal time. Use "elevated workload exposure" and explicitly distinguish the descriptive rule from a validated health assessment. Do not infer a healthy state when telemetry is low.

DATA AND FILES
Read the Person Query from data relative to the workspace root. Save all generated files under output relative to that root, including employee-wellbeing.html, reproducible source, privacy-protected aggregate tables and a validation log. Do not modify the input files.
Inspect the files and actual headers first. Log row counts, distinct people, dates, configured grouping, metric windows, native key uniqueness and missingness. Treat PersonId as text. Use raw export headers in the report and retain mappings to imported names.
Prioritise Collaboration hours, Active connected hours, After hours collaboration hours and Weekend collaboration hours. Add other verified measures only when relevant.
Resolve duplicate person-period keys before analysis. Confirm weekly grouping, complete boundary weeks, comparable metric definitions and working-hours settings. Do not infer cadence from date gaps or reconstruct weekly values from monthly or overlapping-window totals. If weekly analysis is unsupported, report compatible component summaries and omit the proposed weekly flag.

DEFINE THE DESCRIPTIVE EXPOSURE RULE
Use a stated common reference window. For each person, retain weeks where all four primary measures are observed and complete. Require at least four such weeks and report observation counts. Missing behavioural hours are not zero, and incomplete people are not unflagged.
For eligible people, calculate mean weekly hours separately for each primary measure over those weeks. Then calculate each measure's 90th percentile across these person-level means, giving each person one equal weight.
Define elevated workload exposure as at least two of the four person-level means strictly exceeding their respective reference thresholds. Equality at the threshold does not count. A measure with no variation cannot discriminate exposure. Publish the exact reference population, dates, numerical cut-offs, tie rule and eligibility rule.
This is a proposed descriptive screening rule, not a validated predictor, clinical threshold or burnout score. Measures may overlap and correlate, so two flags are not two independent pieces of health evidence.
If any primary measure is absent, or too few eligible people remain, omit the combined flag. Continue with supported component distributions and explain the limitation without fabricating a reduced "burnout" score.
Repeat the calculation at the 85th and 95th percentiles as labelled sensitivity analyses. Keep the primary thresholds fixed when comparing later windows. Relative thresholds do not establish an absolute safe workload or permit direct comparisons between separately calibrated populations.

SUMMARISE POPULATION AND GROUPS
Show the number of people in scope, eligible for the rule, incomplete and flagged, subject to privacy protection. Define prevalence as flagged eligible people divided by eligible people, with distinct-person denominators.
Report component distributions and the share above each threshold. Do not add Collaboration hours, Active connected hours, After hours collaboration hours and Weekend collaboration hours into a total. Digital connected time is not total time spent working.
Discover available organisational attributes and choose a small set of useful comparisons. Report group denominators, observation coverage, prevalence and differences from the reference population. Do not imply that a role-expected working pattern is a health problem.
Use observed weeks for the person's period summary, and state how organisational moves are assigned in window-level views. For weekly trends, use dated groups and distinguish population composition changes from within-person changes. Show a stable-cohort sensitivity where feasible.
Do not show a trend from a single period. Exclude partial periods consistently, disclose exclusions and avoid attributing low activity to holidays or leave without supporting evidence.

EXPLORE ASSOCIATIONS
Where relevant additional measures are available, explore associations with the exposure flag or its components. Potential context includes meeting patterns, focus opportunities, role and organisational attributes.
Start with clear group comparisons and distributions. Do not call the flag's four defining measures, their transforms or close algebraic equivalents independent "drivers" of the flag. That would restate its construction.
Do not force a predictive model. If an exploratory model is used, state its outcome, population, adjustment set and repeated-person handling, assess stability and label it descriptive. It cannot demonstrate causes or predict clinical burnout.
Treat survey wellbeing outcomes as separate measures with their own consent, dates, coverage and grain. Do not invent survey scores or infer them from collaboration hours.
Offer proportionate questions for teams to investigate, such as meeting load or out-of-hours expectations, rather than individual interventions or employment decisions.

PRIVACY, PRESENTATION AND CHECKS
Publish only aggregates meeting max(the organisational minimum, 10 distinct people). Apply that threshold to flagged and unflagged cells, sensitivity cohorts, intersections, embedded data and CSVs. Suppress complementary cells or entire breakdowns if subtraction reveals a small group. Never publish person-level risk lists, identifiers or rankings.
Create a single self-contained HTML report with embedded assets and no CDN or running server. Lead with a short factual summary and the measurement limitation, then component views, two-column group comparisons, compatible trends, exploratory associations and methodology details.
Attach units, populations, dates, denominators and caveats to results. Keep an evidence ledger for headlines. Validate cohort accounting, thresholds, missingness, sensitivity results and suppression across every output, then inspect offline desktop and mobile layouts.
Log executed checks and unavailable analyses separately. If data are insufficient, produce a clearly labelled readiness and component report instead of a success-shaped burnout assessment.

<!-- analysis-policy:start -->
Use product-specific licence evidence. For documented native licensed-only M365 Copilot metrics, null means unlicensed and zero activity means licensed but inactive. Prefer Copilot Enabled User or Total Copilot enabled days. Do not extend these semantics to unlicensed Copilot Chat or Consumption credits.
Investigate conflicting licence evidence. Keep affected eligibility or activity unresolved, and withhold affected rates or estimates until resolved.
Within a supplied complete query's configured scope, absent activity rows contribute zero. Keep failed joins, unavailable sources, out-of-scope periods and known coverage failures unresolved with reasons.
Use configured grouping, week start and metric windows; date gaps do not establish cadence. State grain, units and estimand. Count distinct people; do not sum stocks or overlapping windows.
Use max(organisation requirement, workflow floor): 5 distinct people for general analysis and 10 for joined-query, dashboard, causal or process workflows. Protect complementary cells and embedded data; never publish individual rankings.
Use only HR attributes present in the export. If none suit a requested breakdown, omit it, explain why, and continue with supported overall results.
Treat low-activity flags as review candidates, not verified holidays, employment status or job classifications. Value scenarios require explicit conversion and realisation assumptions.
Default to association. Make causal claims only with a supported design and compatible estimators. Record unavailable checks and label synthetic data.
Check installed API versions, signatures and return types. Separate executed results from expectations and do not silently fork upstream bugs.
<!-- analysis-policy:end -->

Adaptation notes

• Supply an agreed absolute workload threshold if organisational policy provides one, keeping it distinct from the relative percentile rule.
• Use survey measures to analyse reported wellbeing only with suitable governance and compatible observation windows.
• For daily exports, verify metric additivity and complete-week coverage before deriving weekly hours.
• Broaden the reference window if fewer than four complete observed weeks are available. Do not relax the rule silently.
• A causal question requires a separate study design. This card provides descriptive signals and exploratory associations.

Common failure modes

• Telemetry is labelled burnout. Activity cannot establish a clinical condition, personal circumstances or wellbeing.
• Overlapping hours are added. The four measures capture overlapping behaviours and must remain separate.
• Incomplete observations become low risk. Missingness must be shown outside the flag denominator, not converted to zero.
• Percentiles are treated as health thresholds. They are population-relative descriptors. Report cut-offs and sensitivity.
• Driver analysis rediscovers the flag’s ingredients. Exclude defining measures and their transformations from claims about independent associations.
• Small high-exposure groups are exposed. Protect flagged and complementary cells in every representation, including downloaded data.

🔒 Responsible use & data privacy

• Suppress small groups. Apply the stricter of your organisation's minimum group size and the selected workflow's privacy floor to every breakdown, segment and chart.
• Never expose individuals. Do not print raw UPNs or email addresses. Hash, truncate or report only aggregates. Person Query PersonId is already de-identified, but Purview UserId is not.
• Report group patterns. Use these outputs to understand cohorts and trends. Do not evaluate, rank or monitor named employees.
• Keep data in approved environments. Follow your organisation's policies when choosing an agent and deciding which data it may access. Validate every agent output before sharing.

← Previous: Copilot Causal Toolkit Next: External Collaboration →