Short answer
Useful employee engagement KPIs have written definitions that another analyst can reproduce. Each entry should state its numerator, denominator, time window, population, data source and interpretation limit. Keep three groups separate: engagement measures, collection and follow-through measures, and workforce outcomes. Participation, activity and retention can add context, but none of them is a direct substitute for a defined engagement measure or proof of cause.
Use a KPI specification card
Before a number reaches a dashboard, complete this record:
| Field | What to write |
|---|---|
| KPI name | A label that does not claim more than the calculation |
| Purpose | The decision or review the KPI informs |
| Numerator | Events, people or scores included |
| Denominator | The eligible population or opportunities included |
| Time window | Point in time, month, quarter, campaign or rolling period |
| Population | Employment status, location, tenure or other scope rules |
| Source | Survey, HR record, action log or approved qualitative material |
| Exclusions | Missing items, leave, transfers, joiners or other defined cases |
| Breakdown rules | Permitted groups and minimum reporting conditions |
| Interpretation | What the KPI describes and what it cannot establish |
| Owner | Person responsible for data quality and review |
Keep the card versioned. Changing a denominator or completion rule can move the number even when the underlying experience did not change.
Engagement measures
Engagement index
Definition: A composite of specified survey items intended to represent a stated engagement construct.
Formula: Apply the documented item scoring and weighting, then aggregate only under the stated completeness rule.
An index is not a universal measure. The UK Civil Service People Survey uses five questions and converts its five-point scale to 0, 25, 50, 75 and 100 before averaging. The US Office of Personnel Management's 2024 FEVS technical report uses subindices representing Leaders Lead, Supervisors and Intrinsic Work Experience. These official systems illustrate why a dashboard must publish its own construct and calculation.
Interpretation limit: A group average describes responses under that instrument. It does not diagnose a cause, represent nonrespondents automatically or score an individual's future behaviour.
Percent favourable
Definition: The share of usable answers in favourable categories for a specified item or item set.
Formula: favourable answers / usable answers × 100.
If 51 of 75 usable answers are “agree” or “strongly agree,” percent favourable is:
51 / 75 × 100 = 68.0%
State whether “favourable” includes one or several response categories. Show the full distribution nearby. The Civil Service quality and methodology guidance demonstrates that two groups with the same percent-positive result can have different whole-scale index scores.
Interpretation limit: Percent favourable loses information about neutral and unfavourable responses. It should not be compared with an index calculated from the entire scale as though the two figures were equivalent.
Item-level change
Definition: The difference between the same item measure in two stated periods.
Formula: current-period result minus prior-period result, using unrounded values before final rounding.
If clarity is 61.7 in one quarter and 64.1 in the next, the observed change is:
64.1 - 61.7 = 2.4 points
Interpretation limit: A point change is not a percent change. It may reflect real differences, sampling variation, response mix, timing, question order or other conditions. Report uncertainty where the design supports it and avoid “improved” until comparability has been checked.
Collection and follow-through measures
Complete-response rate
Definition: The share of the eligible invited population meeting a pre-defined completion rule.
Formula: complete responses / eligible people invited × 100.
If 160 eligible employees were invited and 112 completed the required items:
112 / 160 × 100 = 70.0%
The AAPOR Standard Definitions provides formal survey outcome-rate conventions and explains why response rate alone cannot determine whether nonresponse error exists. For an internal census-style campaign, record eligible, delivered, started, partial and complete counts so the calculation can be checked.
Interpretation limit: This is participation in a collection process. It is not an engagement level, a trust measure or evidence that respondents represent the workforce.
Item response rate
Definition: The share of respondents presented with an applicable item who provide a usable answer.
Formula: usable answers / respondents presented with the item × 100.
Suppose the campaign completion rule permits a respondent to omit a supplementary workload item. If all 112 campaign completers were shown that applicable item but 96 provide a usable answer:
96 / 112 × 100 = 85.7%
Interpretation limit: A skipped item is not a negative answer. Review whether wording, routing, sensitivity or technical access affected missingness.
Action closure rate
Definition: The share of promised actions due in the period that have a recorded decision and an employee-facing update under the organisation's closure rule.
Formula: actions closed under the rule / actions due × 100.
If seven actions are due and four have both a decision and an update:
4 / 7 × 100 = 57.1%
Keep deferred, cancelled, overdue and completed actions separate. Do not close an action merely because a task was marked complete.
Interpretation limit: This measures follow-through activity. It does not reveal whether the action helped or whether engagement changed because of it.
Conversation coverage
Definition: The share of an eligible population that completed a specified conversation during a defined window.
Formula: people completing the conversation / eligible people offered it × 100.
Use one person once in each denominator unless the KPI explicitly measures conversations rather than people. If 22 people complete 31 conversations, people coverage is based on 22, not 31.
Interpretation limit: Coverage describes access and participation. Conversation length, number of words and number of follow-ups are activity measures, not engagement scores.
Workforce outcomes that require context
Voluntary quit rate
Definition: Voluntary quits during a specified period relative to the chosen employment denominator.
For internal reporting, use the organisation's reconciled payroll definition and state it. The US Bureau of Labor Statistics JOLTS technical note defines quits as voluntary separations and calculates a monthly quits rate from quits divided by employment. That public statistic has its own coverage and estimation method; an employer's dashboard should not borrow the label while using an unstated denominator.
If a fictional employer records six voluntary quits in a month and uses 300 employees as its documented employment denominator:
6 / 300 × 100 = 2.0%
Interpretation limit: Quits are a workforce outcome. They may relate to many personal, organisational and labour-market conditions. They do not directly measure engagement and do not identify why a person left.
Cohort retention
Definition: The share of a fixed starting cohort still employed at the end of a stated period, after applying documented exclusions.
Formula: starting cohort members retained at period end / eligible starting cohort × 100.
If 180 eligible employees were in the starting cohort and 169 remained at period end:
169 / 180 × 100 = 93.9%
Do not add joiners to the numerator of a starting-cohort measure. State how transfers, retirement, fixed-term endings and leave are handled.
Interpretation limit: Retention is not the inverse of an engagement score. A person may stay for many reasons, and a person may leave while engaged with the work.
Internal mobility rate
Definition: Employees making a qualifying internal move during a period as a share of the population eligible for such moves.
Formula: employees with a qualifying move / eligible employees × 100.
Define whether promotions, lateral moves, temporary assignments and location transfers count. A vacancy shortage can constrain the numerator, while policy changes can expand it.
Interpretation limit: Mobility reflects opportunity, eligibility and organisational structure as well as employee choice. It is not a direct measure of motivation or commitment.
A fictional quarterly KPI sheet
The following figures describe one fictional UK service operation. They are designed to show the calculations, not provide benchmarks. The engagement index uses the same four items, scale, scoring and completeness rule in both quarters. All 112 Quarter 1 campaign completers and all 122 Quarter 2 completers answered those four index items. The separate workload item is supplementary and may be omitted without making the campaign response incomplete.
| KPI | Quarter 1 | Quarter 2 | Comparable? | Interpretation |
|---|---|---|---|---|
| Engagement index | 66.0 | 68.4 | Questions stable; respondent mix changed | Observed rise of 2.4 points, cause unknown |
| Percent favourable on priority clarity | 68.0% | 69.5% | Same item and scale | Distribution still needs review |
| Complete-response rate | 70.0% (112/160) | 76.3% (122/160) | Same eligible count; added kiosk access | Collection coverage rose; mode may matter |
| Workload item response | 85.7% (96/112) | 91.0% (111/122) | Same routing | Fewer missing answers; reason to investigate |
| Action closure | 57.1% (4/7) | 75.0% (6/8) | Different actions due | Follow-through activity, not engagement |
| Retention at 90 days | 93.9% (169/180) | 94.4% (170/180) | Two distinct starter cohorts of 180; same eligibility and 90-day rule | Workforce outcome with multiple explanations |
The Quarter 2 engagement index and response rate both increased. That co-movement does not show that kiosk access improved engagement. It may have changed who responded, and the two quarters may contain different work events. Review the respondent mix, score distribution and employee accounts before proposing an explanation.
Rules for comparable reporting
Use the same checks before every trend claim:
- Keep the question wording, order, scale and scoring rule stable.
- Keep eligibility and exclusion rules stable or show the change.
- Compare equivalent calendar or operational periods where seasonality matters.
- Show complete and item-level denominators beside percentages.
- Examine which groups are missing, newly represented or too small to report safely.
- Separate full-population, respondent-only and fixed-cohort views.
- Calculate with unrounded values and round only the final result.
- Label revised historical data and explain the revision.
- Preserve qualitative counterexamples and unresolved evidence.
- Keep observed change separate from attribution.
For the overall design, including construct and cohort choices, use the employee engagement measurement guide. For open comments and interview material, use the qualitative engagement data guide.
Do not turn a dashboard into a causal claim
Monitoring can show that an index, process metric or outcome changed. It cannot by itself show that a manager programme, listening campaign or policy caused the change. The UK government's Magenta Book explains the additional design needed to assess attribution, including a suitable comparison or counterfactual where appropriate.
Record alternative explanations before reviewing results. If the intended decision requires a causal answer, involve an analyst who can design a proportionate evaluation. If it does not, describe the result as an observation and use it to choose the next question.
Add employee context without creating a score
Lontra can help a team explore a bounded work question through employee conversations and prepare a manager brief. Managers do not receive raw employee responses. HR aggregate views start at five respondents, but that threshold does not guarantee anonymity. The product does not turn conversation content into an engagement score, predict individual outcomes or establish that one action caused a metric to move.
The Lontra product overview shows how conversations, review and manager preparation fit together. Keep any conversation themes beside the relevant KPI as reviewable context, with source and period, rather than blending them into a new composite number.
Frequently asked questions
- What are employee engagement KPIs?
- Employee engagement KPIs are defined measures used to track a stated engagement construct, the quality of its collection process, or related workforce outcomes. Each KPI needs a formula, denominator, time window, population and interpretation rule.
- Is survey participation an employee engagement KPI?
- Participation is a collection-process measure, not an engagement score. It helps assess coverage and missing groups, but a high or low response rate does not reveal why people did or did not respond.