Predicted chargeback $ over the next 30 days based on EFWs and any commerce-sibling shipment delays that historically convert to product_not_received disputes.
At a glance
A 30-day forward projection of chargeback dollars the merchant is likely to see, based on (a) Stripe Early Fraud Warnings (EFWs) already received, multiplied by their historical conversion-to-chargeback rate, plus (b) commerce-sibling shipment delays that historically convert to product_not_received disputes. Predicts dollars out the door before the chargebacks land.
| What it counts | (EFW_count_30d × historical_efw_to_chargeback_conversion_pct × avg_dispute_amount) + (commerce_sibling.shipment_delayed_orders × product_not_received_dispute_rate × avg_dispute_amount). Two components: fraud-driven (Stripe EFWs) and operations-driven (shipment delays from the commerce sibling). |
| API resources & fields | Stripe /v1/radar/early_fraud_warnings for the EFW count; /v1/disputes for the historical conversion rate and avg_dispute_amount; commerce sibling for shipment_delayed_orders count. |
| Currency | The commerce sibling’s primary store currency. Multi-currency disputes are FX-converted at the day’s mid-market rate. |
| Fees / processing cost | NOT included in the forecast. Stripe charges 15 × forecast_count for fee-inclusive view. |
| Refunds | NOT a factor on inputs. EFWs and shipment delays predict future disputes; refunds happen on succeeded charges that weren’t disputed. |
| Disputes / chargebacks | This is the metric output. The card forecasts the dollar value of disputes that have not yet landed. |
| Failed / declined payments | NOT a factor. Disputes only happen on succeeded charges. |
| EFW conversion rate | The merchant’s last 90 days of EFW-to-dispute conversion (typically 60, 80%). New merchants without EFW history use a 70% default. |
| Shipment-delay conversion | The commerce-sibling’s historical delayed_order → product_not_received_dispute rate (typically 2, 8% of delayed orders escalate). |
| Page cap relevance | EFWs typically run < 100 / refresh, well under 1,000-page caps. Disputes also rarely hit caps. |
| Time window | Forward 30 days; uses 90 days of history for the conversion-rate inputs. |
| Alert trigger | > $2k forecast, level at which dispute response work (evidence submission, customer outreach, refund-instead-of-dispute negotiation) typically pays for itself |
| Roles | owner, finance, operations |
Calculation
Calculated automatically from your Stripe 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 US DTC electronics merchant on Shopify + Stripe. Computed on 12 Apr 26 for the next 30 days (12 Apr 26 to 12 May 26). Component A: Stripe EFW-driven forecast| Input | Value | Source |
|---|---|---|
| Early Fraud Warnings received last 30 days | 22 | stripe.radar/early_fraud_warnings |
| Historical EFW → chargeback conversion (90D) | 73% | stripe.disputes joined to historical EFWs |
| Average dispute amount (90D) | $148 | stripe.disputes.amount |
| Input | Value | Source |
|---|---|---|
| Orders with shipping delay >7 days last 30 days | 86 | shopify.orders.fulfillment_status + tracking.estimated_delivery |
| Historical delay → product_not_received dispute rate | 4.8% | stripe.disputes WHERE reason='product_not_received' joined to delayed orders |
| Average dispute amount | $148 | (same as above) |
- EFWs are the dominant predictor. 2,987 sits in the fraud bucket. Stripe sends an EFW typically 7, 14 days before the chargeback lands; the merchant has a window to refund the customer, which prevents the dispute (and the $15 chargeback fee). A merchant who refunds 100% of EFW-flagged charges proactively can recover almost all of this number; a merchant who ignores them sees roughly 73% land as actual disputes.
- Shipment delays add a small but real second component. $611 from delayed orders. Often invisible on Stripe-only views because the disputes haven’t landed yet. The fix is on the commerce side, expediting delayed orders, proactive comms to affected customers, and pre-emptive partial refunds for inconvenience all reduce the conversion rate.
- The forecast is additive to refunds and existing disputes. This card doesn’t include disputes already filed (those are in Dispute Value) or refunds already issued (those are in Refund Value). It’s the new forecast over the next 30 days. Treat the three cards as three separate buckets.
Sibling cards merchants should reference together
| Card | Why merchants reach for it next to Chargeback Forecast |
|---|---|
| Early Fraud Warnings | The EFW count input. The single most actionable card; refund proactively to prevent EFWs from converting to disputes. |
| Dispute Value | The actual disputes already filed (vs forecast). Pair to read “filed + forecast = next-quarter total”. |
| Dispute Reasons | The decomposition of what kind of disputes have landed. Tells you which forecast component (fraud vs operations) is most predictive for your merchant. |
| Dispute Rate | The current rate. Forecast pushing the merchant above the 0.9% Visa threshold is a leading-indicator alert. |
| Dispute Win Rate | The merchant’s effectiveness at fighting disputes. Low win rate means even disputes that do land cost the merchant the full amount; the forecast doesn’t apply a “we’ll win some” discount. |
| Refund Rate | The cheap escape hatch. Refunding an EFW-flagged charge before the chargeback lands costs the merchant the full charge minus a small refund-fee, but avoids the $15 dispute fee and the dispute-rate damage. |
| Shopify Fulfillment Time / BC Order Status Distribution | The shipment-delay input on the commerce side. Reducing fulfilment delays directly reduces the operations-driven forecast component. |
| Payment Health Score | High forecast usually drags health score down (via the dispute-rate amplifier ×50). Acting on the forecast lifts both the score and the actual dispute rate. |
Reconciling against the vendor’s own dashboard
Where to look in Stripe Dashboard: Stripe Dashboard does not expose a forward-looking chargeback forecast. The closest Stripe-native screens for cross-checking the inputs:- Radar → Early fraud warnings for the EFW count input.
- Disputes for the historical conversion rate and average dispute amount inputs.
- Disputes → Reasons breakdown for the historical fraud-vs-operations split.
- Reports → Dispute report for the historical conversion-rate trend.
- Stripe’s “Pending disputes” tile. Counts disputes already filed but not yet resolved; that’s a backward-looking measurement. Our card is forward-looking.
- Stripe Sigma “EFW conversion” custom queries. Closest equivalent for the fraud-driven component, but rarely include the commerce-sibling shipment-delay component.
- Chargebacks911 / Verifi forecasts (third-party fraud tools). External chargeback-prediction services that overlap with our signal but use different data (cardholder-side history vs merchant-side EFWs).
| Reason | Direction | Why |
|---|---|---|
| Operations component | Ours higher if shipping is unstable | Stripe doesn’t see commerce-sibling shipment delays. A merchant with deteriorating fulfilment will see our forecast climb while Stripe-only views look stable. |
| Conversion-rate input | Variable | We use 90-day historical conversion; merchants whose fraud landscape has shifted (new product, new region, new acquisition channel) may see actual conversion diverge from the historical input. |
| Page cap (1,000 charges) | Marginal | EFW and dispute volumes rarely hit the cap. Affects only very-high-volume merchants. |
| FX conversion | Estimate-grade | Multi-currency disputes converted at the day’s mid-market rate. |
| Time horizon | Strict 30-day forward | Some merchants want a 14-day or 60-day view; the card is fixed at 30 days to match Visa / Mastercard chargeback monitoring windows. |
| Refresh lag | Ours 5, 30 min behind | EFW data syncs every refresh; the latest hour of EFWs may be slightly stale. |
| Comparison | Expected relationship | When divergence is legitimate |
|---|---|---|
stripe_xc_chargeback_forecast ↔ stripe.stripe_disputes_open (next month) | Forecast should approximate the open-disputes count × avg amount that materialise in the following month. Validate by waiting 30 days then comparing. | Calibration error of ±25% is normal; >50% suggests either a fraud-landscape shift or commerce-side fulfilment improvement / deterioration. |
stripe_xc_chargeback_forecast ↔ stripe.stripe_dispute_rate (forward) | Forecast / monthly_revenue ≈ next-month dispute rate. If forecast / revenue > 0.9% the merchant is heading toward Visa monitoring. | Acting on EFWs proactively (refunding before they convert) breaks this relationship in the merchant’s favour. |
stripe_xc_chargeback_forecast (operations component) ↔ commerce-sibling Fulfilment Time / Order Status Distribution | Reducing the delayed-order share directly reduces this card’s operations component. | A merchant who fixes shipping should see this card’s operations component fall within 30, 60 days. |
Known limitations / merchant FAQs
Reconciliation questions are answered in the Reconciling against the vendor’s own dashboard section above.“My forecast feels too high; will all those disputes actually land?” No. The forecast applies historical conversion rates (typically 73% for EFWs and 4, 8% for shipment delays). Some EFWs will be resolved by the customer (returned product, fraud team caught it) and some delayed orders will arrive without dispute. The forecast is a propensity-weighted prediction, not a deterministic count. If you proactively refund EFWs and expedite delayed orders, the actual landed dispute total can drop to ~10, 20% of the forecast. “My forecast feels too low; how can it predict so few disputes?” Two common cases. (1) Your EFW volume is low because Stripe Radar isn’t sending many warnings, which can mean either healthy fraud rates or a Radar configuration issue (rules suppressing EFWs). (2) Your commerce sibling reports few shipment delays, which can mean healthy fulfilment or that the commerce-platform connector isn’t capturing all shipment events. Validate by checking Early Fraud Warnings volume against Stripe Dashboard directly, and the commerce-sibling fulfilment cards. “I’m a US merchant with no SCA, why is my forecast so high?” US merchants don’t have SCA protections, which means more friendly-fraud (
fraudulent reason where the cardholder claims they didn’t authorise) lands as a chargeback. The fraud-driven component will run higher than for UK / EU merchants on the same revenue base. The fix is the same: act on EFWs proactively.
“What’s the relationship between EFWs and chargebacks?”
Stripe Radar sends an EFW typically 7, 14 days before the actual chargeback lands. The merchant has a window to refund the customer or contact them; if you refund, the dispute almost always doesn’t materialise (the customer no longer needs to call their bank). If you ignore the EFW, ~70, 80% become chargebacks within 30 days. Acting on EFWs is the single highest-ROI work surfaced by this card.
**“Are chargeback fees (15 × forecast_count for the fee-inclusive view. For the example above (390 in fees.
“Multi-currency, what currency is the forecast in?”
The commerce sibling’s primary store currency. Multi-currency EFWs and disputes are FX-converted at the day’s mid-market rate. Estimate-grade.
“My subscription store, do recurring-payment disputes feed this card?”
Yes. Stripe issues EFWs for subscription charges the same way as one-off charges. The disputes that follow are typically subscription_canceled (customer claims they cancelled before the renewal) or fraudulent (customer doesn’t recognise the charge). The forecast captures both. Subscription stores often see disproportionately high EFW counts because customers forget about renewals; proactive renewal-reminder emails reduce this materially.
“How do I act on a high forecast?”
Three usual moves, ranked by impact:
- Triage EFWs daily. The Radar → EFW screen lists each warning. Refund or contact the customer within 24 hours. Each EFW you neutralise removes ~150 of forecast.
- Expedite delayed orders. The commerce-sibling fulfilment delay component is harder to fix instantly but compounds: every order that ships on time is one fewer
product_not_receiveddispute waiting to happen. - Tighten Radar rules if the EFW conversion rate is climbing. A rising conversion rate means Radar is missing fraud the issuer catches; the rule set may need updating.