The difference between a survey metric and a KPI, the four families of survey metric, and the evidence rules that make a reported metric comparable, traceable, and worth acting on.
Survey metrics — also called survey measures — are the numbers a survey produces, from response rate and completion rate to satisfaction, NPS, and outcome measures, describing participation, data quality, experience, and change. Survey KPIs are the small subset of those metrics tied to a decision. A metric describes what happened; a KPI is a metric you would act on. Most teams track many metrics and have almost no KPIs.
The problem is not too few metrics; it is metrics reported as lone numbers, cut loose from the responses that produced them, so no one can say why a number moved or what to do about it. This guide separates survey metrics from KPIs, sets out the four families of survey metric, pairs quantitative measures with the qualitative reasons behind them, and gives the evidence rules that make a reported metric defensible.
Key takeaways
The difference between a survey metric and a survey KPI is not the number — it is the role. A metric describes what happened; a KPI is a metric attached to a decision, so a change in it changes what you do. A team tracking forty metrics and calling all of them KPIs has no KPIs, because none of them is tied to an action.
The discipline is subtraction: from the metrics you can measure, pick the few that would actually change a decision, and demote the rest to monitoring. Read the table. Choosing the few that matter is walked through in define the metrics funders want, and where they surface for a decision on actionable insights.
| A survey metric | A survey KPI | |
|---|---|---|
| Role | Describes what happened | Drives a specific decision |
| Number of them | Many — you track dozens | Few — the handful you act on |
| Test | Is it accurate? | Would a change in it change what you do? |
Survey metrics fall into four families: participation (response and completion rates), quality (data cleanliness and item non-response), experience (satisfaction, NPS, effort), and outcome (change in the respondent over time). Each family answers a different reliability question, and only the outcome family answers whether the program worked.
The trap is stopping at experience metrics — a high satisfaction score or NPS feels like success but says nothing about whether anyone's situation changed. The outcome family is the one that requires the same respondent measured over time, which is why it is the family most survey tools cannot support. The instrument design behind these is on survey design, and the outcome family in depth on impact survey questions.
| Family | What it measures | The reliability question it raises |
|---|---|---|
| Participation | Response and completion rates | Is the sample who actually answered representative? |
| Quality | Data cleanliness, item non-response | Can the metric be trusted before it is read? |
| Experience | Satisfaction, NPS, effort | Does a good score mean anything changed? |
| Outcome | Change in the respondent over time | Did the program move the thing it exists to move? |
A survey metric reported as a lone number is fragile: 72% satisfaction, a 4.1 average, an NPS of 30. It describes, but it cannot explain why, for whom, or what to change, because the responses that produced it were left behind. A reviewer or a board that asks the second question gets silence.
Sopact calls the standard The Evidence-Linked Metric: every reported metric links to the responses behind it, carries the open-ended reason alongside the number, is comparable over time, and traces to its source. The difference is a data-model one — a metric cut loose in a deck versus one that drills from the number to the respondents and their words. The quantitative and qualitative halves belong on one record, which is covered on survey analysis and how to analyze survey data. The stage below shows one metric reported both ways.
A survey metric reported once a quarter describes the last quarter; a metric read on arrival, with its reason attached, drives a decision this one. That is the premise of the Loop, Sopact's method for continuous impact intelligence: collect clean at the source, analyze the moment data arrives, improve while you can still act.
The Loop is also what makes a metric defensible. Every number traces to the responses behind it, and the same definition returns the same figure twice, so a KPI resolves to its evidence. That standard has its own chapter in reliability and reproducibility. Where survey metrics feed the wider practice is on impact measurement.
One method, three moves that never stop
Then the cycle runs again, a little sharper each wave. Read the method: the Loop methodology →
The fastest way to fix a metrics deck is to test each number against a decision and attach its evidence. Each prompt below pastes into Sopact Sense's Assistant, or reasons through with your team; the arrow above each links the Academy walkthrough that shows the expected output and the tips.
Academy walkthrough → Separate metrics from KPIs
From this list of survey metrics we track, separate the KPIs from the metrics: [PASTE METRICS]. For each, decide whether a change in it would change a decision we make; if yes, it is a KPI and name the decision; if no, mark it monitor-only. Return a table: Metric / Family / KPI or monitor / Decision it drives.
Academy walkthrough → Make every metric evidence-linked
For each reported survey metric, build its evidence link: [PASTE METRICS]. Give the exact definition, the responses behind it, the calculation, the segment cut, and whether it is comparable to last period. Flag any metric whose definition changed or whose source you cannot reconstruct. Return a table: Metric / Definition / Responses / Comparable? / Traceable?
Academy walkthrough → Cut every KPI by segment
Disaggregate our survey KPIs by the segments we collect: [PASTE KPIS + SUBGROUP FIELDS]. Report each KPI per segment, flag gaps larger than chance, and mark cells too small to interpret. Do not report a gap the cell size does not support. Return a table: KPI / Segment / Value / Gap / Interpretable?
Academy walkthrough → Attach the reason to the number
For each survey metric that moved this period, attach the reason from the open-ended responses: [PASTE METRICS + OPEN RESPONSES + CODEBOOK]. Give the dominant theme behind each movement and two verbatims. If the responses do not explain the movement, say so rather than inventing a cause. Return a table: Metric / Movement / Reason / Verbatims.
Each walkthrough is a hands-on companion written to run on your own data: what to do, the prompt to run, the output to expect, and the tips that keep it reliable.
Watch: turning a deck of lone survey metrics into a few evidence-linked KPIs that drill from the number to the responses behind it.
Survey metrics are the numbers a survey produces that describe participation, data quality, respondent experience, and outcomes — response rate, completion rate, satisfaction, NPS, and change measures. They describe what happened. Sopact's framing is the Evidence-Linked Metric: a metric matters only when it links to the responses behind it, is comparable over time, and drives a decision, rather than sitting as a lone number in a report.
A survey metric describes what happened; a survey KPI is a metric attached to a decision, so a change in it changes what you do. You track many metrics and have few KPIs. The test is simple: would a change in this number change what you do? If yes, it is a KPI; if no, it is a metric to monitor. Sopact's guidance is to demote most metrics to monitoring and keep the handful that drive action as KPIs.
Survey KPIs are the small set of survey metrics tied to a specific decision — the ones a team acts on rather than merely tracks. A KPI is defined by its link to an action, not by being a particular metric; the same number can be a KPI for one team and background monitoring for another. Sopact treats a KPI as an Evidence-Linked Metric: defined once, traceable to its responses, and connected to the decision it informs.
The metrics that matter are the ones in the outcome family — change in the respondent over time — plus the quality metrics that tell you whether the outcome data can be trusted. Participation and experience metrics matter for operations but say nothing about whether the program worked. A high satisfaction score with no outcome measure is a comfortable number, not a meaningful one. Sopact prioritizes outcome metrics because they answer the question a funder actually asks.
Participation (response and completion rates), quality (data cleanliness, item non-response), experience (satisfaction, NPS, effort), and outcome (change in the respondent over time). Each answers a different reliability question, and only the outcome family answers whether the program moved anything. Sopact's point is that most dashboards stop at experience metrics, which feel like success but measure sentiment, not change.
Quantitative survey metrics are the numeric measures — rates, averages, scores, and change values — as opposed to the qualitative themes from open-ended responses. Both belong together: a quantitative metric describes the movement and the qualitative reason explains it. Sopact keeps the number and the theme on one record so a quantitative KPI drills to the open-ended responses behind it rather than standing alone.
Write one definition for the metric — the exact numerator, denominator, and filter — and keep it fixed across periods, the way you would treat an accounting definition. A metric whose definition drifts between quarters is not comparable, and a trend built on it is an artifact. Sopact locks each metric's definition so the same figure means the same thing every wave, which is what makes a trend real rather than a moving target.
Attach evidence to every number: the definition, the responses behind it, the segment cut, the comparison to last period, and the open-ended reason for any movement. A board trusts a metric it can drill into, not a lone figure on a slide. Sopact generates the report from the responses so every metric in it resolves to its source, which is the difference between a defensible report and a decorative one.
Next: read the metrics on survey analysis, or design the outcome family on impact survey questions.
