Measure Shopify email revenue honestly — not rented last-click credit

Flow-specific attribution windows — conservative defaults

Default ESP windows over-credit long consideration cycles. Use shorter windows for high-intent flows, longer only where justified. Cart recovery: 24–48 hour click window — customer was mid-purchase. Welcome: 7 days for first purchase attribution; exclude if customer purchased from paid social same session. Post-purchase cross-sell: 14 days but tag orders where recommended SKU was already in browse history — directional only. Winback: 7 days; exclude subscription renewals and wholesale reorders by tag. Replenishment: 3 days — timing-sensitive consumable logic.

Document windows in ops wiki alongside flow logic. When finance asks why Q3 winback attributed revenue dropped after window change, you have an answer that is methodology, not performance collapse. A home fragrance brand shortened cart window from 7 days to 48 hours — attributed cart revenue fell 31% but incremental holdout lift unchanged. They were reporting timing credit, not causation.

Shopify order tag reconciliation

ESP attributed revenue should reconcile to Shopify orders with matching UTM or ESP order tags. Weekly export: ESP attributed orders, Shopify orders with utm_source matching ESP, intersection count, delta explanation. Common deltas: guest checkout attribution lag, phone orders without UTM, subscription renewals tagged incorrectly, wholesale orders in DTC flows, partial refunds not netted in ESP dashboard.

Tag every flow link consistently — Klaviyo auto-UTM plus manual verification on button blocks; Sequenzy campaign slugs mirrored in Shopify analytics. Finance spreadsheet columns: order ID, gross revenue, discount amount, COGS if available, ESP attributed Y/N, channel (email/SMS), flow name. Monthly review with retention lead and finance — 30 minutes prevents board-level attribution arguments.

Multi-touch deduplication rules

When email and SMS both run cart recovery, define primary credit: first click within window wins; if SMS and email click same session, credit channel of first open. Postscript and Klaviyo do not deduplicate automatically — spreadsheet rules or BI layer required. Same for Yotpo loyalty email plus Klaviyo promotional — loyalty redemption orders should not credit winback flow because customer clicked winback email then used points.

Subscription renewals are the silent attribution inflator. Exclude Recharge/ Skio active subscribers from winback and replenishment attribution exports. Tag renewal orders in Shopify; filter ESP reports before monthly leadership review. One skincare brand discovered 19% of winback attributed revenue was auto-renewals — board presentation corrected overnight.

Profit per send — the metric finance actually wants

Revenue per send is directionally useful; profit per send is decision-grade. Formula: (attributed revenue − discounts − COGS − allocated ESP/SMS cost) ÷ sends. Cart recovery with 22% attributed revenue and 18% average discount on $45 AOV may be negative profit per send at scale. Welcome with no discount and higher first-purchase rate may beat cart on profit despite lower gross attribution.

Rank flows by profit per send quarterly; reallocate creative and ops hours to positive-contribution flows. Kill or redesign flows negative three consecutive months unless strategic (e.g., post-purchase education for CX). Sequenzy pay-per-email makes send cost visible; Klaviyo requires plan cost allocation by send volume estimate.

Holdout test playbook — 30-day minimum

Pick one flow with stable volume — cart recovery ideal. Create random 8% holdout segment in Klaviyo or Sequenzy. Exclude from flow sends 30 days. Track orders within 7-day window both cohorts. Incremental lift = sent conversion minus holdout conversion. Run quarterly rotating across welcome, cart, winback. Document results in ops wiki; adjust flow investment by true lift not gross attribution.

Cross-read discount discipline for margin-adjusted reporting and analytics use case for workflow-specific metrics.