Coaching coach-satisfaction survey V1 spec

Summary

This delivers the V1 spec for the coach-satisfaction survey scoped in 2026-05-22-coaching-coach-satisfaction-survey-scoping. It resolves that memo's four open questions and delivers the buildable content: the quarterly coach eNPS message and one-tap form, the coach_satisfaction_of_platform formula, and the trigger configuration. The initiative is Coaching-owned. The one external dependency, reusing the existing comms infrastructure for the send, is the confirmation the scoping memo asked Platform for; this spec proceeds on the expectation that it is a yes, since the customer program already proved that path, and only the send section depends on it.

Open questions resolved

The scoping memo left four open questions. Coaching resolves them here.

1. Cadence: quarterly. The scoping memo leaned monthly, but monthly is too frequent a touch for a coach pulse. Quarterly is the standard relationship-survey cadence, keeps coach contact light, and still feeds the metric adequately given the trailing window in question 3.
2. eNPS scoring: a mean normalized rate, not a true promoter-minus-detractor eNPS. This keeps coach_satisfaction_of_platform consistent with coach_quality_score, which is also a mean rate, satisfies kind: rate, and is simpler. The promoter and detractor split stays visible through the raw score distribution and the reason data for anyone who wants it.
3. Trailing window: 180 days, two quarters. Long enough that the weekly-emitted metric stays continuously populated between quarterly surveys and tolerates a coach skipping a quarter.
4. Onboarding coaches: active coaches only for V1. A coach enters the quarterly rotation when their Coach status becomes active.

Survey copy

The recipient is the coach. There is one instrument, the coach eNPS.

Message (SMS):

{brand_name}: Hi {coach_first_name}, one quick question about coaching with us, about 15 seconds: {link}

One-tap form:

• Heading: {brand_name}
• Question: "How likely are you to recommend {brand_name} as a place to coach?"
• Rating control: eleven tappable options, 0 to 10, left-anchored "Not at all likely" and right-anchored "Extremely likely". One tap records the score.
• Reason follow-up, shown after any score: a single-select list, "What most affected your score?" Options: "Pay and compensation", "Scheduling and workload", "Support and tools from {brand_name}", "Students and families", "Facility or equipment", "Growth and development", "Other". The reason is optional and skippable.
• Confirmation screen: "Thanks, {coach_first_name}. Your feedback helps us make {brand_name} a better place to coach."

One-tap form behavior

The same principles as the customer program apply.

• The link carries a signed binding to the coach and the survey period, so the form needs no login.
• Mobile-first. One tap records the score and reveals the single-select reason list and a submit control.
• The reason follow-up is optional and never blocks submission.
• The program captures no free text in V1. Every captured field is structured: a numeric score and an optional categorical reason code. All output is quantitative and countable.
• After submit, the confirmation screen ends the flow. Re-opening the link after a response is recorded shows the confirmation rather than accepting a second submission.
• V1 is English only.

The coach_satisfaction_of_platform formula

coach_satisfaction_of_platform is computed from the coach eNPS score. One structural difference from coach_quality_score shapes the formula: there, many customers rate one coach, so a coach accrues many responses; here the coach is both the subject and the sole respondent, so a coach contributes one response per quarterly pulse, roughly two within a trailing 180-day window. The minimum-N rules account for that.

1. Per-response normalization. Each eNPS response is an integer 0 to 10. It normalizes to a rate r = raw / 10, so 0 maps to 0.00 and 10 to 1.00.

2. Per-coach value, the coach × week grain. For each (coach, week), the value is the mean of r over that coach's own responses in the trailing 180 days. Because the coach is the sole respondent, the per-coach minimum-N is 1: a coach with at least one response in the window is scored, and a coach with none emits null:insufficient_data. A coach answering the quarterly pulse carries one to two responses in the window.

3. Org-market value, the org_market × week grain. The registry's weighted_mean_by_customer_count rollup over the coach-level values in the org-market, emitted when the org-market has at least 5 coach responses in the trailing 180 days, otherwise null:insufficient_data. The small-N guard sits at the org-market grain, where there is genuine volume across coaches, not at the per-coach grain, where one response is the expected and complete sample.

The score is a rate in the range 0 to 1, higher is better, emitted weekly. coach_satisfaction_of_platform's allowed_null_reasons already includes null:insufficient_data, so unlike coach_quality_score this metric needs no allowed_null_reasons change.

Worked example: a coach whose trailing-180-day eNPS responses are 9 and 8 normalizes to 0.90 and 0.80, a mean of 0.85, so their coach_satisfaction_of_platform is 0.85.

Trigger configuration

• A Coaching-owned quarterly scheduled job over Coach records with status active.
• The send day is staggered across the quarter per coach, by a stable per-coach send day derived from the coach id, so the weekly-emitted metric is fed evenly through the quarter rather than in a once-a-quarter spike.
• A coach enters the rotation when status becomes active and leaves it when status leaves active.
• Send path: recipient resolution through Platform's comms-routing endpoint, send through the External Actions Queue (ADR-0011) as its own action_kind, and a one-tap link to the Coaching-hosted form. This reuses the customer program's settled path, and it is the reuse the scoping memo asked Platform to confirm.

Platform note

The trailing-180-day window means coach_satisfaction_of_platform's period_attribution, currently survey-week, should be updated to a trailing-window descriptor when the metric compute is wired (here, trailing-180-day), the same one-field free-text edit Platform made for coach_quality_score (snapshot-week to trailing-8-week). No other registry change is needed: the metric exists, kind: rate holds, and allowed_null_reasons already covers the insufficient-data case.

What remains

With this spec, Coaching's design work for coach_satisfaction_of_platform is complete. The build that follows is: the Coaching-local coach-survey table, the quarterly trigger job, the one-tap form, the External Actions Queue integration, and the coach-grain aggregate the mart reads; then Platform wiring the coach_satisfaction_of_platform compute against it. Coaching owns all of that except the metric compute. No dated commitment here, consistent with the no-rocks discipline.

References

• Scoping memo: 2026-05-22-coaching-coach-satisfaction-survey-scoping
• Customer survey program V1 spec, the pattern this follows: 2026-05-22-coaching-nps-program-v1-spec
• Platform's period_attribution precedent: 2026-05-22-platform-nps-program-v1-period-attribution-confirmation
• Thread root: 2026-05-19-platform-mart-100-percent-deployment-per-domain-asks
• Registry entry for coach_satisfaction_of_platform: platform/lib/mart/registry.ts
• External Actions Queue contract: contracts/external-actions/README.md; ADR-0011
• Coaching domain doc: coordination/domains/coaching.md