> ## Documentation Index > Fetch the complete documentation index at: https://docs.vortexiq.ai/llms.txt > Use this file to discover all available pages before exploring further. # Customer Churn Risk, Adobe Commerce > Customer Churn Risk for Adobe Commerce stores. Tracked live in Vortex IQ Nerve Centre. How to read it, why it matters, and how to act on it. **Metrics type:** [Supporting Metrics](/nerve-centre/overview#metrics-types-explained) • **Category:** [Ecommerce Platform](/nerve-centre/connectors#connectors-by-type) ## At a glance > Predicted percentage of the active customer base that will not place another order in the next 90 days. Combines recency, frequency, and monetary signals into a single score per customer, then aggregates the at-risk cohort. On Adobe Commerce, the card runs on both consumer and B2B accounts but treats them with different cadence assumptions (B2B accounts have established PO cadences; consumer customers are more variable). | | | | ------------------------------------------------------ | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **What it counts** | For each active customer (defined as: at least one order in the trailing 12 months): compute their personal RFM score and infer churn probability based on (a) days-since-last-order vs their established cadence, (b) order-frequency trend over the trailing 6 months, (c) monetary trend (rising AOV is retention-positive, declining AOV is churn-leading). Aggregate as % of base predicted to churn. | | **API field** | `customer_id`, `created_at`, `grand_total`, `customer_group_id` from `GET /rest/V1/orders`. Customer entity attributes from `GET /rest/V1/customers`. B2B Company attribution via `extension_attributes.company_attributes.company_id`. | | **VAT / tax treatment** | Monetary scoring uses `grand_total` (tax-inclusive on B2C, often tax-exempt on B2B). The asymmetric tax treatment is acceptable because RFM scoring is normalised within each customer's history; the score reflects relative trend, not absolute spend. | | **Shipping inclusion** | Included via `grand_total`. | | **Discounts** | Deducted (post-promotion). A customer who switched to discount-only purchasing has lower monetary trend, raising churn probability. | | **Credit Memo refund treatment** | NOT subtracted. A customer with a high refund rate may rank as low-churn-risk based on order frequency alone but should be re-scored with refund-adjusted LTV; this is a known limitation. | | **`state` machine inclusion** | All states except `canceled`. `pending_payment` is included for B2B (net-30 PO pipeline). | | **`pending_payment` quirk** | Included for B2B; an account whose orders all sit in `pending_payment` (broken AP) appears as actively ordering even though no money is captured. Cross-check [B2B Accounts Gone Quiet](/nerve-centre/kpi-cards/adobe-commerce/b2b-accounts-gone-quiet) for stuck-pending edge cases. | | **Multi-currency `grand_total` vs `base_grand_total`** | Uses `base_grand_total` for cross-currency normalisation in the RFM monetary dimension. | | **Store View scope (`store_id`)** | All Store Views by default. Per-Store-View variants useful for distinct shopping behaviours (UK and US consumers have different frequency norms). | | **Time window** | `90D` (the prediction horizon). Inputs are computed over trailing 6 to 12 months of customer history. | | **Alert trigger** | `>35% predicted churn`. The alert intent is "the at-risk cohort has grown beyond manageable rescue capacity". | | **Roles** | owner, marketing | ## Calculation Calculated automatically from your Adobe Commerce data. See the At a glance summary above for what the metric tracks and the worked example below for a typical reading. ## Worked example A homewares brand on Adobe Commerce 2.4.6 with a B2B portal Store View and consumer Store Views. Snapshot Monday 4 May 26. **Active customers (at least one order in last 12 months):** | Segment | Active count | Predicted to churn | Churn % | | ------------- | ------------ | ------------------ | --------- | | B2B Companies | 180 | 22 | 12.2% | | B2C consumer | 9,640 | 3,180 | 33.0% | | **Blended** | **9,820** | **3,202** | **32.6%** | **Cohort breakdown of B2C predicted churners by trailing-12-month LTV:** | LTV cohort | Count | Predicted churn | | ------------------------ | ------------ | ------------------- | | Top 10% (>\$420 LTV) | 964 active | 78 (8.1% churn) | | Mid 60% ($100 to $420) | 5,784 active | 1,840 (31.8% churn) | | Bottom 30% (\<\$100 LTV) | 2,892 active | 1,262 (43.6% churn) | **B2B predicted churners (22 of 180):** | LTV bucket | Count | Predicted churn | | ----------------------- | --------- | ---------------- | | Top 20 (>\$50k 12m LTV) | 36 active | 2 (5.6% churn) | | Middle 60 ($5k to $50k) | 108 | 14 (13.0% churn) | | Bottom 20 (\<\$5k) | 36 | 6 (16.7% churn) | What this is telling marketing and Sales: 1. **Blended churn risk is 32.6%**, just under the 35% alert threshold. Rising trend over the last 4 weeks (was 28% in early April). 2. **B2B churn is materially lower than consumer** (12% vs 33%). Expected pattern; B2B accounts are stickier because of contracts, integration costs, and procurement workflow. 3. **The top-10% LTV cohort has only 8% predicted churn**, meaning the most valuable consumers are sticky. Good. The bottom-30% LTV (low-value customers, often one-purchase) churn at 44%, also expected; these are typically gift-buyers or single-purpose shoppers who never re-engaged. 4. **The mid-60% LTV cohort is the rescue priority**: 1,840 customers at risk, at $100 to $420 LTV each. A targeted retention campaign converting 15% of them (276 customers) at average LTV of $250 = ~$69k of saved revenue if save-rate matches industry norms (typically 12 to 20% on a well-executed win-back email). 5. **B2B churn priorities**: 2 top-20 Companies are flagged as at-risk. Each rep $50k-plus annual revenue. A Sales call this week is worth ~$100k of preserved annual revenue if both are saved. The middle-60 cohort (14 flagged) is best routed through a Customer Success follow-up rather than direct Sales engagement. 6. **Cross-check with [B2B Accounts Gone Quiet](/nerve-centre/kpi-cards/adobe-commerce/b2b-accounts-gone-quiet)**: 14 of the 22 B2B churn-risk accounts also appear in the silence list. The two cards agree on the priority Companies. The remaining 8 are not silent but their order-frequency trend is contracting, an earlier signal. 7. **Action plan**: marketing runs win-back email cadence to the 1,840 mid-LTV consumer churners; Sales calls the 22 flagged B2B Companies. Re-measure churn risk in 30 days. The point: aggregate churn is a strategic dial; per-cohort decomposition is the action layer. A 33% churn rate at the bottom of the LTV pyramid does not need rescue; a 13% churn rate at the top of the B2B pyramid absolutely does. ## Sibling cards merchants should reference together | Card | Why pair it with Customer Churn Risk | | ----------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------- | | [B2B Accounts Gone Quiet](/nerve-centre/kpi-cards/adobe-commerce/b2b-accounts-gone-quiet) | The B2B-specific silence detection. Subset overlap with churn risk for B2B; the silence card is more action-oriented for individual accounts. | | [Customer Count](/nerve-centre/kpi-cards/adobe-commerce/unique-customers) | The active customer base (denominator). | | [Customer Trend](/nerve-centre/kpi-cards/adobe-commerce/customer-acquisition-trend) | New vs returning trend. A churn rate that's stable but customer count growing means new acquisition is masking churn. | | [Repeat Customer Rate](/nerve-centre/kpi-cards/adobe-commerce/repeat-customer-rate) | Inverse: high repeat = low churn. Useful sanity check. | | [Customer Segments](/nerve-centre/kpi-cards/adobe-commerce/customer-spend-segments) | LTV cohort breakdown that this card uses for prioritisation. | | [AOV](/nerve-centre/kpi-cards/adobe-commerce/average-order-value) | Monetary trend dimension of RFM scoring. AOV trend per customer is one of the inputs. | | [`klaviyo.email_engagement`](/nerve-centre/klaviyo/email_engagement) | Email engagement is a leading indicator of churn; declining open rates precede order silence. | | [`shopify.churn_risk`](/nerve-centre/kpi-cards/shopify/customer-churn-risk) | Cross-platform peer for agencies. | ## Reconciling against the vendor's own dashboard **Where to look in Adobe Commerce Admin:** Adobe Commerce does not have a native churn-risk score. The closest views: > **Reports > Customers > Customers by Number of Orders** lists customers ranked by lifetime order count. A customer with 1 order in the last 12 months is a likely churner; one with 12 is unlikely. > **Reports > Customers > Customers by Orders Total** lists by lifetime spend. Combined with the order-count report you can manually approximate RFM scoring. For B2B (Adobe Commerce paid edition): > **Customers > Companies** shows per-Company order history; you can manually scan for cadence-vs-silence, but at scale this is only practical via export. For email engagement (which this card uses as a churn-leading signal indirectly): > **Marketing > Newsletter Subscribers** shows subscribed status. A customer who has unsubscribed is a strong churn signal but not a one-to-one mapping. Other Adobe Commerce Admin views that look relevant but are not: * **Customers > Now Online**: real-time login state, not behavioural pattern. * **Reports > Customers > Wishlist**: wishlist activity, weak signal. * **Reports > Sales > Coupons**: coupon usage, mid-strength signal. **Why our number may legitimately differ from any manual computation:** | Reason | Direction of divergence | | ------------------------------------------------------------------------------------------------------------------------------ | ----------------------------------------------------------------- | | **Definition of "active"**. Card uses 12-month-active. Admin's customer list is all-time. | Card population smaller | | **Time-zone**. Admin in Store View timezone; card in UTC. Border-day customers may shift. | Negligible at 90-day horizon | | **B2B vs consumer treatment**. Card uses different cadence assumptions for each segment. Manual RFM doesn't typically segment. | Material if your manual analysis treats both segments identically | | **Refund-adjusted LTV**. Card does not subtract refunds from LTV in the monetary score (known limitation). | Card overstates LTV for high-refund customers | **Cross-connector reconciliation (when these connectors are connected for this merchant):** | Pair | Expected relationship | What divergence tells you | | ------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------- | | [`klaviyo.email_engagement`](/nerve-centre/klaviyo/email_engagement) | Declining email engagement precedes churn risk increase by 2-4 weeks | Strong leading indicator. | | [`google_analytics.ga_returning_users`](/nerve-centre/kpi-cards/google-analytics/returning-users) | Returning-user count should correlate inversely with churn risk | If GA4 shows steady returning users but churn risk is rising, customers are visiting but not buying (a different problem: site or product issue, not retention). | | ESP unsubscribe rate | Unsubscribe spike often precedes churn-risk spike | Adobe Commerce's mailing-list status is a coarser version of the same signal. | ***
Same-metric documentation cross-reference (for agencies running multiple platforms) * [`shopify.churn_risk`](/nerve-centre/kpi-cards/shopify/customer-churn-risk) * [`bigcommerce.churn_risk`](/nerve-centre/kpi-cards/bigcommerce/customer-churn-risk)
## Known limitations / merchant FAQs **The card says my churn risk is 33% but my repeat rate is 60%, isn't that contradictory?** No. Repeat rate is past-looking (what % of historical customers returned); churn risk is forward-looking (what % of currently-active customers will not return in next 90 days). Active customers churn at higher rates than the lifetime base because the lifetime base includes customers who already churned long ago. The two numbers measure different things. **Adobe Commerce vs Magento Open Source: any difference?** The underlying customer and order data is identical between editions. The card runs the same scoring on both. Adobe Commerce paid edition's B2B Companies module gives explicit Company-level segmentation, which makes the B2B churn view cleaner. On Open Source, B2B detection relies on Customer Groups. **Why does B2B churn look so much lower than consumer?** Genuine pattern. B2B accounts have switching costs (procurement workflow, integration, contract terms) that make them stickier. A B2B account "churning" usually means they've gone through a multi-month process of evaluating an alternative supplier; the churn signal often appears 60 to 90 days before the final break, which is why the [B2B Accounts Gone Quiet](/nerve-centre/kpi-cards/adobe-commerce/b2b-accounts-gone-quiet) card uses cadence-deviation as the leading indicator. **My churn risk just spiked 5 points, what should I do?** First, decompose. Is the spike concentrated in one segment (consumer LTV cohort, B2B portal)? In one Store View? In one customer-acquisition channel (recent paid-traffic cohorts often show high early churn)? The decomposition shapes the response. A spike concentrated in newly-acquired low-LTV consumers is normal and acceptable; a spike in mid-tier B2B is urgent. **My multi-store Adobe Commerce, can I get per-Store-View churn?** Yes, configure per-Store-View variants. Customer behavioural patterns differ materially by region; UK consumer cadence runs faster than US consumer cadence in most categories. A blended view masks regional patterns. **Why doesn't the card refund-adjust LTV?** Known limitation. A customer who orders £500/month and refunds £300/month has the same monetary trend in this card as one who orders £500/month and keeps everything. Refund-adjustment is on the roadmap; for now, cross-check [Refund Rate](/nerve-centre/kpi-cards/adobe-commerce/refund-rate) per customer manually for high-LTV churners. **My ESP shows different churn signals (unsubscribes, inactivity), should I trust this card or the ESP?** Both. The ESP shows email-engagement churn; this card shows order-behaviour churn. They overlap by 50 to 70%; the divergence is informative. A customer who unsubscribed from email but continues ordering is not at order-churn risk; a customer with high email engagement but declining order frequency is at order-churn risk. Use both signals together for the action prioritisation. **The model is a black box, can I see why a specific customer was flagged?** Yes, the Vortex IQ workspace has a per-customer detail panel showing the RFM score components: days since last order, order frequency trend, monetary trend, plus the inferred churn probability. Useful for explaining to Sales why a specific Company is on the rescue list. *** ### Tracked live in Vortex IQ Nerve Centre *Customer Churn Risk* is one of hundreds of KPI pulses Vortex IQ tracks across Adobe Commerce and 70+ other ecommerce connectors. Nerve Centre runs the detection layer; Vortex Mind investigates the cause when something moves; Ask Viq lets you interrogate any number in plain English. [Start for free](https://app.vortexiq.ai/login) or [book a demo](https://www.vortexiq.ai/contact-us) to see this metric running on your own data.