Measurement plan
Moral Trade needs enough instrumentation to learn whether visitors understand the pilot, reach a safe first action, and trust the review workflow. It does not need analytics that rank moral views, expose exact wishes, or optimize for engagement.
The current event taxonomy follows the same privacy limits as the product: public route paths, coarse stage labels, consent state, and bucketed performance signals. Query strings, hashes, raw search text, source notes, receipts, and private messages stay out of funnel events.
page_viewWhich public routes do visitors reach before choosing a first pilot path?
Allowed metadata: path, referrerPath, navigationType
Used for: Find confusing entry points and improve navigation copy.
hero_primary_cta_clickedDo visitors choose the worked-example route before signup pressure?
Allowed metadata: hrefPath, label, routeFamily
Used for: Tune the homepage and primer toward concrete, low-risk examples.
worked_example_openedWhich examples make the mechanism legible enough to inspect details?
Allowed metadata: exampleId, template, routeFamily
Used for: Prioritize clearer examples and remove misleading ones.
signup_startWhere do visitors begin account creation after seeing the pilot boundaries?
Allowed metadata: path, partnerSlug, primaryGoal
Used for: Detect whether the site asks for accounts too early or from the wrong surfaces.
signup_completeCan serious participants finish a low-friction account setup?
Allowed metadata: partnerSlug, primaryGoal
Used for: Identify account friction before widening the cohort.
onboarding_completeDo new users choose a role, cause area, and first action?
Allowed metadata: participantKind, primaryGoal, firstAction
Used for: Route new users toward one reviewable action instead of a blank marketplace.
create_trade_startedDo users draft terms after inspecting examples and safety rules?
Allowed metadata: mode, template, generatedBy
Used for: Improve draft scaffolding when users stall before creating terms.
clone_example_actionWhich worked examples become editable first actions?
Allowed metadata: exampleId, template
Used for: Invest in example formats that produce reviewable drafts.
wish_profile_startedDo users start broad wish previews before requesting private introductions?
Allowed metadata: causeAreaCount, fieldCount
Used for: Improve private matching onboarding while keeping exact wishes undisclosed.
cohort_interest_startedWhich cohort paths attract people willing to invite one serious counterparty?
Allowed metadata: partnerSlug, primaryGoal
Used for: Prioritize cohort pages that create reviewable, consent-gated activity.
privacy_grant_changedAre people able to understand and change field-level sharing grants?
Allowed metadata: accessLevel, fieldCount, decision
Used for: Improve consent controls and privacy copy.
detail_request_submittedDo broad previews lead to explicit, consent-gated detail requests?
Allowed metadata: targetKind, candidateBucket, bothConsented
Used for: Measure whether staged disclosure is working without exposing exact wishes.
detail_request_resolvedCan detail requests resolve without pressure or hidden disclosure?
Allowed metadata: targetKind, decision, stage
Used for: Improve resolution states and operator follow-up.
safety_report_submittedWhere do visitors flag coercion, fraud, pressure, or externality concerns?
Allowed metadata: targetKind, stage
Used for: Prioritize safety review and public rulebook clarifications.
match_consent_recordedDo proposed introductions record both-party consent before identity-specific disclosure?
Allowed metadata: bothConsented, stage, hasMatch
Used for: Audit whether the matching workflow remains consent-gated.
background_scan_runCan operators review broad matching signals without raw private text?
Allowed metadata: matchesCreated, candidateBucket, taskCount
Used for: Monitor operator queue quality, not user surveillance.
performance_metric_recordedDo core routes stay fast enough for mobile and first-time visitors?
Allowed metadata: metricName, metricRating, metricValueBucket
Used for: Fix LCP, INP, and CLS regressions on the heaviest public routes.
donation_route_clickedDo visitors reach external donation routes from clear non-custodial pages?
Allowed metadata: targetKind, hasTargetUrl, causeAreas
Used for: Improve route clarity without claiming escrow, custody, or tax handling.
donation_loggedCan users record external-payment evidence for review?
Allowed metadata: causeAreas, stage, hasSession
Used for: Improve evidence workflows while avoiding payment custody claims.
evidence_submission_startedWhere do users begin proof submission for commitments or public-good actions?
Allowed metadata: targetKind, stage
Used for: Improve proof scaffolding and reviewer triage.
public_good_action_loggedAre shared public-good actions becoming reviewable records?
Allowed metadata: causeAreas, stage, taskCount
Used for: Track public-good pilot participation in aggregate.
Protocol-quality audits
The evaluation contract measures whether assistance improves clarity without replacing human review, leaking private content, hiding false matches, or under-surfacing thresholded groups.
Cadence: monthly public aggregate. Privacy boundaries: aggregate only by default, no raw private wish text, no contact details, no source note leakage, small cell suppression, deviation review log redacted, free text quote requires consent.
Tracks whether structured drafts become valid faster while reviewers still overrule weak suggestions.
Keeps blocking, privacy, and false-match checks visible before expanding any assisted workflow.
Looks for material preview gaps across privacy-thresholded slices before broader rollout.
Reports whether evidence, appeals, duplicate-proof issues, and unresolved disputes are improving.
Surfacing parity sample: pass across 20 eligible previews, with 2 reviewed deviation log(s) and 0 unreviewed.
UX readiness sample: pass; blockers: 0.
Open evaluation JSONThe first performance baseline should cover the routes people use before trusting the pilot. Web Vitals may be recorded as LCP, INP, and CLS buckets; Lighthouse or PageSpeed-style runs should stay aggregate and route-level.
Run MORALTRADE_BASE_URL=http://127.0.0.1:3000 npm run measure:routes against a production build to capture local mobile and desktop route-health evidence before optimizing. The output path defaults to /tmp/moraltrade-public-route-baseline.json.
Devices: Mobile baseline 390x844; Desktop baseline 1440x1000.
Budgets: DOM content loaded under 10000ms, load under 15000ms, at least 400 body-text characters, and no more than 80 script tags.
/Mobile and desktop route-level baseline.
/what-is-moral-tradeMobile and desktop route-level baseline.
/offersMobile and desktop route-level baseline.
/worked-examplesMobile and desktop route-level baseline.
/cohortMobile and desktop route-level baseline.
/signupMobile and desktop route-level baseline.
/loginMobile and desktop route-level baseline.
/background-networkingMobile and desktop route-level baseline.
/wish-registryMobile and desktop route-level baseline.
/mpgfMobile and desktop route-level baseline.
/measurementMobile and desktop route-level baseline.
active
Record only the approved funnel events and sanitized metadata listed on this page.
active
Run the public route-baseline command on the home, offers, cohort, signup, login, Moral Trade, private matching, and public-goods routes on mobile and desktop before optimizing.
active
Expose an opt-out control that clears the attribution cookie and suppresses optional funnel-event inserts for the current browser.
planned
Use Search Console only for aggregate query, indexing, and crawl diagnostics; do not merge search-console data into private wish or offer records.
planned
Publish cohort-scale counts for route clarity, review queues, safety reports, performance health, and disclosure resolution without exposing individual records.
Measurement is useful only if visitors can inspect the surrounding promises: privacy, pilot status, trust boundaries, validation, and public updates.