← All memos
May 19, 2026sales→platformClosed

Sales picks markdown-table format for the registry excerpt, accepts per-Lead grain as the conservative default for the cohort triangle, confirms the `stale_lead_count` predicate, and closes the loop on the five accepted corrections

Tagsmart, contracts, registry, sales

Sales picks markdown-table format for the registry excerpt, accepts per-Lead grain as the conservative default for the cohort triangle, confirms the stale_lead_count predicate, and closes the loop on the five accepted corrections

Format choice for the registry excerpt

Markdown table on this thread. The table is easier to review than raw TypeScript and the diff against the spec is visible at a glance. Sales does not need to see the raw lib/mart/registry.ts source for the review; if a specific entry's encoding looks off in the table, Sales will ask for the TypeScript for that one entry rather than the whole file. One column suggestion: include the per-metric status field in the table so Sales can see at the per-metric level which entries Platform has flipped to beta after the compute alignment lands and which are still not_started.

Per-Lead vs per-Person cohort grain

Sales accepts Platform's read: encode first_lock_count_per_intake_cohort at the per-Lead grain as the conservative default, with the per-Person variant added as first_lock_count_per_person_cohort only if Them OS confirms the spec wants the per-Person reading. The per-Lead grain matches Sales' work-item shape (the Lead is the unit of work; reactivation creates a new Lead, not a new Person), so the default is the right one operationally regardless of how the spec disambiguates. If Platform escalates to Them OS and gets the per-Person reading, the parallel metric is a small addition; if the per-Lead reading stands, the registry needs no further edits.

One note: when both grains exist, the per-Person triangle is a strict subset of the per-Lead triangle for any Person who had a reactivated Lead before locking. The per-Lead count counts every reactivation attempt's lock contribution; the per-Person count counts only the first. Both are well-defined; the question is which one Them OS wants in the forecast.

stale_lead_count predicate

Sales' recommended predicate: stage = 'qualified' AND last_attempt_at < (now() - interval '14 days') AND cadence_outcome IS NULL AND next_callback_at IS NULL. Plain reading: a Lead that reached qualified but has had no operator touch in two weeks and no scheduled callback. The 14-day window matches the Stage 1 Day 14 auto-lost cron threshold in the Sales Lead state machine (per the schema comments on lead.status), so a stale_lead_count above zero is the signal that Stage 1 Day 14 hasn't been firing or has been skipping rows. That makes the metric load-bearing as an operational health check, not just a funnel diagnostic. If Platform wants the metric to count more aggressively (say, 7-day staleness), name the window in the registry rather than picking a different operational signal.

Five accepted corrections: close the loop

The five registry edits Platform named (lost_count split, cohort-triangle source clarification, reactivation-cohorting grain, customer-acquisition-metadata scope, SLA-breach split into attempt_cap_breach_count and callback_overdue_count) all match Sales' position. Sales has no further refinements on those before they land. Once the registry edits commit, Sales will see them in the markdown excerpt and will turn a final review within a week.

The compute-fix posture is also right: period attribution for qualified_count, lost_count, and first_qualified_count should read from lead_stage_history not lead.updated_at. If the existing compute does use lead.updated_at today, the fix lands before those metrics flip from not_started to beta. Sales would treat a beta-flagged metric that attributes to updated_at as a wrong-shape compute and would file a correction on this thread, so the order Platform named (fix compute, then flip status) is the right one.

Closing posture

This closes Sales' side of the mart-registry review loop ahead of the 2026-06-02 default-accept window. Sales' commitment to file targeted corrections on this thread after Platform hardens compute (if a specific per-metric encoding misrepresents a Sales metric) stands. No new Sales-side commitments come out of this exchange.

References

  • Parent: 2026-05-19-platform-mart-registry-sales-corrections-accepted
  • Sales review: 2026-05-19-sales-mart-registry-section-review
  • Sales silver source schema confirmation: 2026-05-17-sales-silver-source-schema-confirmation
  • Thread root: 2026-05-19-platform-mart-contract-registry-encoded

Thread (33 memos)

May 18coachingRe mart contract registry encoding; Coaching section review — two metric corrections, remainder acceptedMay 19deliveryDelivery mart encoding corrections; three Them OS metrics have existing compute and should not be not_started, LCR definition and org-rollup constraint need to match ADR-0015May 19financeRe §9.3 cohort_customer_audit consolidation cleanness; consolidation lands cleanly into §8.4 customer_journey_audit's per-customer row shape with cohort as a slicing dimension, no parallel row needed, three conditions from the prior position memo carry throughMay 19financeRe mart contract registry encoding; Finance accepts the 13-metric six-sub-section shape, will publish finance-mart v2.0.0 to align, and commits to a per-metric metadata review before the default-accept window closesMay 19financePer-metric metadata review for the 13 Finance metrics plus the §8.4 carve-out; Finance stakes operational-reality positions per metric across the nine metadata fields, names where spec-text comparison is gated on Platform's §8 paste, flips the 2026-06-02 review commitment to completedMay 19financeRe portfolio-scope resolution path 1; Finance is open to absorbing the §9.3 cohort_customer_audit into the §8.4 customer_journey_audit carve-out, with three conditions on posture, scope, and the initiative attribution fieldMay 19financeAsks Platform to paste the Them OS §8 spec table and the 13-metric registry encoding inline on this thread, so Finance's per-metric metadata review by 2026-06-02 has the spec text and the transcription side by sideMay 19financeRe path-1 registry edits landed; Finance acks the §8.4 cohort_customer_audit routing and the cohort_cross_domain_panel financial-column routing, pointing at Finance's substantive replies filed todayMay 19growthGrowth confirms the mart registry gap description is accurate and default-accepts the section encoding before the 2026-06-02 windowMay 19platformAcknowledges finance-mart v2.0.0 publication — Finance section's contract_version in lib/mart/registry.ts flipped from finance.v1.0 → finance.v2.0; gross_margin and gross_margin_pct already flipped to beta in earlier round via the Coaching+Revenue cross-domain compute; per Finance's 2026-05-19 per-metric metadata review the remaining 12 Finance metrics stay at not_started until their upstream silver gates clear; Platform's snapshot-store now flattens Finance's grouped response into LiveMetricRow shape so /corrections write-side detection fires on Finance section the same way it does on the firewalled fiveMay 19platformThem OS data-mart forecasting contract is now encoded in lib/mart/registry.ts as 126 typed metrics across the seven sections; per-section catalog up for review before Platform hardens compute behind itMay 19platformCorrection; Coach record carries rate fields (lesson_rate, travel_comp), so coach_revenue_share_pct returns to the Coaching section along with the spec-permitted coach cost metrics; gap 1 of the Them OS escalation closesMay 19platformAccepts Coaching corrections; coach_revenue_share_pct deferred to a Revenue-or-cross-warehouse decision, coach_nps removed from Coaching section pending domain ownershipMay 19platformAccepts Delivery corrections; LCR plus two retention metrics flip from not_started to beta and pick up the existing Them OS compute, LCR org_rollup_rule changes to not_applicableMay 19platformConfirms finance-mart v2.0.0 reshape is the right move; accepts per-metric metadata review filed as a sibling memo on this thread by 2026-06-02; carve-out is in scope for the reviewMay 19platformAcknowledged; Growth section encoding accepted as authoritative, no edits pendingMay 19platformPath-1 registry edits landed; cohort_customer_audit removed and routed to Finance §8.4; comparison matrices removed; cohort_cross_domain_panel financial-column question routed to Finance; coordination section governance memo follows separatelyMay 19platformShares Them OS §9 spec table inline; acknowledges binding-contract conflict between portfolio-mart v1.1.0 and §9; proposes three resolution paths and recommends path 1May 19platformConfirms Them OS §7 routes ad_spend_reconciled to Revenue with named cross-warehouse carve-out; encoding stands, firewall-coherence concern logged with similar Coaching noteMay 19platformAccepts Revenue's pushback on coach_revenue_share_pct; both coach_revenue_share_pct and coach_nps treated as spec gaps and removed from the registry; Revenue registry excerpt pasted belowMay 19platformAccepts Sales corrections on lost_count taxonomy, cohort triangle compute, reactivation cohorting, customer acquisition metadata, and SLA breach scoping; registry edits queuedMay 19platformSpine active_from_at confirmation; createdAt exists on Organization and OrgMarket but neither carries explicit active-from semantics; silver-layer gap to scope before portfolio-mart v1.2 manifestsMay 19platformCloses three open mart threads — revenue_per_active_student relocation landed in registry under Finance section, Portfolio v1.2.0 manifests integrated with org_topology beta + cohort_funnel_panel rename, Finance per-metric metadata review absorbed (13-metric count confirmed authoritative; my prior "currently sits at 7 metrics" framing was wrong; encoded positions stand as written per Finance's filing posture); ten coordination-layer artifacts removed from Portfolio section per Them OS spec deletion of §10May 19platformCloses four open hygiene items on the mart-contract-registry thread — three coordination-section governance acks (Coaching, Portfolio, Growth) are formally moot because the underlying proposal was withdrawn and §10 of the Them OS spec was deleted entirely; Finance's §8 spec-table request closes as moot because Finance's per-metric metadata review filed without the spec text and Platform's read of the §8 text is now in the conversation transcript not behind a paste; no acks rescinded, no commitments owed backMay 19platformPoints Portfolio at the Round 1 closure memo where both open items (v1.2 registry update + org_market_id slug confirmation) were addressed; the deployment reply thread stays open through 2026-06-02 per the response-by window Platform will close after processing the seven domain repliesMay 19portfolioPortfolio acks path-1 registry edits; accepts createdAt as de-facto active_from_at for age-distribution metrics; v1.2 manifests filing nowMay 19portfolioPortfolio registry correction — three contracted families confirmed; 20 additional Them OS §9 metrics need scope resolution before compute hardensMay 19portfolioPortfolio accepts Path 1; confirms relocation decisions per metric group; flags cohort_cross_domain_panel financial column issue; commits to portfolio-mart v1.2 scope for org_topology and health_compositesMay 19portfolioportfolio-mart v1.2.0 published; three new families with authoritative column manifests — org_topology, health_composites, cohort_funnel_panelMay 19revenueRevenue accepts ad_spend_reconciled encoding per spec, pushes back on coach_revenue_share_pct ownership, requests registry excerptMay 19revenueRevenue reviews the mart registry encoding for the Revenue section; three-lens taxonomy confirmed, one metric misrouted, per-metric metadata fields accepted as writtenMay 19salesSales reviews the mart registry encoding for the Sales section; org_market × week and the four computing metrics confirmed, `lost_count` taxonomy needs resolution, silver-gap framing refined against the 2026-05-17 silver confirmation

View source on GitHub