CRM Monthly Business Review

Triple Whale (Linear All 28-day) is the revenue source for cross-channel attribution. Shopify net sales is the % of DTC denominator. Cost rows include all CRM platform subscriptions; revenue and cost rows share the same column layout so they read top-to-bottom.

Definitions: Sent ≥1 email = unique profiles who had at least one Received Email event this month (delivery-attempt proxy; bounced-only profiles are <0.5% of volume and excluded). Open / Click / Order stage rates = unique at this stage / unique at previous stage. Stage 7 "Sent to order" = unique placed-order / unique sent (overall conversion). Emails per user = total delivery events ÷ unique recipients (mean exposure). Opt-out rate = unique unsubscribers ÷ unique recipients.
Parent vs sub-row methodology: For Sent / Opened / Clicked, parent and Campaign / Flow sub-rows are Klaviyo unique-profile counts from the same metric-aggregates feed, queried at interval=month with a single-month window per call (multi-month spans return truncated values; daily summed inflates uniques by counting the same profile across each day). Sub-rows split events by Klaviyo's $flow attribute: empty = campaign-sourced, set = flow-sourced. Campaign + Flow can exceed Total because a single subscriber may receive both a campaign and a flow message in the same month (unique within each bucket, but only once in the Total). Typical overlap is 15–25%.
Stage 5 (Placed an order): True email-driven conversion. Parent = Campaign + Flow, both computed from Triple Whale's per-campaign / per-flow attribution (orders TW links to a Klaviyo email touch — open or click within the 28-day window). This is the click-then-purchased funnel: only orders Klaviyo's emails actually drove. Klaviyo's raw placed_order_unique metric (73K for May 2026) counts ALL profiles regardless of email engagement and would inflate stage-to-stage conversion to ~70%; the §2.2 K-customer count (53K) is all subscribers who bought (regardless of email touch) and still inflates because most of those purchases would have happened anyway. The TW-attributed parent (~9–10K for May) is the figure that answers "did the email drive the buy?" Campaign + Flow sum to the parent by construction (no "Other" sub-row needed).
Column structure: Each metric shows the current month value, then for both MoM and YoY a "prior value" + "Δ" pair. Δ is shown as % change for count metrics (e.g. ▲ 13.4%) and as percentage points for rate metrics (e.g. ▼ 0.4pp). The unit follows the metric's natural unit, and the prior value next to each Δ lets you do the math directly.
Stage 1 trend (Total subscribed): Sourced directly from Klaviyo's Profiles → Subscriber Growth dashboard export (data/klaviyo_subscriber_growth_us.csv and _eu.csv). The dashboard's weekly "Email subscribers" series is Klaviyo's own population count; we use the last weekly snapshot whose Start Date falls in each calendar month as the end-of-month value. Refresh those CSVs from the dashboard before each monthly report build to keep history accurate.

July 2026: K cohort = customers with consent=SUBSCRIBED in Klaviyo who placed an order this month; non-K = everyone else who ordered. New = first-ever Shopify order in this month (lifetime numberOfOrders == orders this window). Revenue uses Shopify net sales (currentSubtotalPriceSet — post-discount, post-refund, pre-tax/shipping).

US Champions, daily. Shaded = vendor bug window; dashed line = the re-score. The level recovers in a single day, which is why the July 2026 decline cannot be read as churn — real recency decay does not reverse overnight.

Spans 31 May→11 Aug (~2.5 months, including the corrupted window) — directional only, and not a July 2026 figure. Champions is roughly flat against the last clean reading, so the "halving" was never real. Recent's fall is partly genuine decay and partly the re-score moving mis-parked customers out of it; the two cannot be separated without per-customer history, which the frozen export denies us.

Rebuilding this needs per-customer scores for dates Reveal no longer serves: their RFM export job has not regenerated since 13 Jul (HTTP Last-Modified confirms it), so no fresh snapshot exists to compare against. It returns once that export is fixed and two clean snapshots exist — an open item with the vendor. Aggregate segment sizes are unaffected and shown in §2.4.

Revenue uses Triple Whale Linear All 28-day attribution joined to platform campaign_id. Heat banding: top quartile green, bottom quartile red, applied to Revenue + Rev/rec (+ ROI for Attentive).

Klaviyo flow revenue from TW Linear All 28-day, joined to flow_id. Attentive journeys use Attentive-native revenue (TW exposes Attentive at channel level only). Engagement metrics are platform-native.

NPS = (promoters 9-10) − (detractors 0-6) as % of respondents. Calendar month grouping uses Reveal's Responded At timestamp. Reply rate = responses ÷ NPS emails Klaviyo delivered (flow Omniconvert - Post Delivery NPS), not OmniConvert's invitation count: OC's figure is sent+consumed+expired, and expired is an age-based state that absorbs never-sent invitations, so it counts every invitation ever created and inflates the denominator by ~2× — increasingly so the older the month, which manufactured a false rising trend. This will not match OmniConvert's dashboard, which uses that same definition. OC's count is shown alongside for reconciliation. Themes are raw word-frequency over the free-text comment fields (after stopword strip); verbatims are quoted unmodified. Industry benchmarks from npspack 2025 study of 1,000+ e-commerce brands.