CSAT & CES Measurement
Read for operational touchpoint satisfaction/effort surveys. Voice defines the instrument; Field owns one-off research design, Pulse owns dashboard governance, and Echo produces synthetic hypotheses—not measured customer responses.
Instrument and scale contract
Freeze question wording, labels, direction, channel and instrument version before collecting responses. Never pool different scales or wording variants without a documented comparability analysis.
| Instrument | Question | Scale | Calculation |
|---|---|---|---|
| CSAT | How satisfied were you with [specific interaction]? | 1 Very dissatisfied; 2 Dissatisfied; 3 Neutral; 4 Satisfied; 5 Very satisfied | 100 × count(score in {4,5}) / valid answered ratings. Label T2B; a T1B result is not interchangeable. |
| CES agreement | [Company] made it easy for me to [task]. | 1 Strongly disagree through 4 Neutral to 7 Strongly agree | Mean of valid 1–7 ratings; higher means easier. |
| CES ease variant | How easy was it to complete [task]? | 1 Very difficult through 4 Neutral to 7 Very easy | Same numeric direction, but record a distinct instrument version; do not silently mix with agreement wording. |
Report valid n, invited/eligible denominator, response rate, missingness, time window, cohort/channel, distribution, T2B or mean, and the uncertainty method/interval. With n=0, report not estimable. Keep CES 4 neutral in raw data; do not recode it as a high-effort answer to strengthen a risk claim.
Touchpoint and survey rules
| Touchpoint | Instrument / trigger |
|---|---|
| Support resolution | Select CSAT or CES for the decision; within 2h of close. |
| Self-service / settings / task completion | CES; article use within 30min, other tasks after completion. |
| Onboarding / first successful feature use / checkout | CSAT after completion, or CES when task effort is the explicit question. |
| Cancellation intent | Exit survey, not a satisfaction proxy. |
| Relationship / advocacy | NPS, not a touchpoint CSAT substitute. |
One rating + free-text why + at most one optional segment question; no bundled CSAT/CES/NPS survey. Ask neutrally, not from a relationship-biased sender. Apply the existing 30-day cross-survey suppression rule. Preserve consent, minimization and access restrictions; use a protected stable subject identifier where follow-up is authorized.
Follow up on CSAT 1–2 (cause of dissatisfaction) or CES 1–3 (highest-friction step) with the recovery owner within 24h. For positive scores, ask what to preserve only when that qualitative context is needed. A missing why is missing evidence, not a guessed reason.
Operational decision defaults
These are Voice defaults to confirm against the engagement, not universal industry benchmarks or independent authorization to stop a release.
| Metric | Default interpretation / next action |
|---|---|
| CSAT T2B ≥85%; 75–84%; 65–74%; <65% | Maintain / iterate on bottom-box themes / targeted improvement / escalate to the accountable owner. |
| CSAT bottom-box (1–2) | Target ≤5%; >10% triggers service-recovery escalation even with high T2B. |
| CES mean ≥6; 5–<6; 4–<5; <4 | Maintain / improve difficult tasks / investigate effort sources / escalate critical effort. |
| CES dashboard target | Mean ≥5.5, high-effort share (1–3) <20%, low-effort share (5–7) >60%; name this target separately from the interpretation bands. |
For a launch/block decision, include the sample limits, measured effect, owner and the approved release gate; Voice does not invent deployment authority. Compare external benchmarks only after matching source year, population, question, scale, scoring and channel. Otherwise report non-comparable rather than using a cached industry table.
Triangulation
Join CSAT/CES/NPS only on comparable cohorts/windows; interpretations below are investigation hypotheses, not measured causes.
| CSAT / CES / NPS | Investigate / action |
|---|---|
| High / High / High | Protect the observed experience. |
| High / High / Low | Brand/value or advocacy gap. |
| High / Low / High | Loved product with task friction; investigate and reduce effort. |
| High / Low / Low | Possible hidden dissatisfaction/churn risk; inspect verbatim evidence. |
| Low / High / High | Specific dissatisfaction despite loyalty; recovery follow-up. |
| Low / Low / Any | Service-recovery priority; close the loop within 24h. |
| Mixed cohorts | Report the divergence; do not let the aggregate conceal the affected cohort. |
Minimal data contracts
interface CSATResponse {
score: 1 | 2 | 3 | 4 | 5;
touchpoint: string;
feedback?: string;
}
interface CESResponse {
score: 1 | 2 | 3 | 4 | 5 | 6 | 7;
touchpoint: string;
feedback?: string;
userId: string; // protected identifier; never export identity in public reports
timestamp: string;
}Store instrument version, collection window, eligibility/response denominators and cohort metadata alongside the response collection. Keep consent/access metadata under the project's existing privacy contract; do not infer consent from a score.
Report and handoff
Output: metric/value/target/status; touchpoint/score/n/trend; high-effort issue/count/verbatim evidence/uncertain cause; priority/current→target/action/owner; sampling limits and follow-up disposition. Do not fill root-cause cells without evidence.
Route why text to voice thematic, relationship triangulation to voice nps, stable metrics to Pulse, recovery cohorts to Growth, redesign briefs to Spark, effort hypotheses to Echo, reliability-correlated drops to Beacon, and strategic cohort trade-offs to Magi. The handoff carries instrument version, cohort/window and evidence—not an unsourced industry rank.