Order-count divided by quote/cart-create count, fires when a checkout regression silently kills conversion.
At a glance
Real-time canary that compares completed-order count against quote-create count to derive a checkout abandonment rate. Fires when the rate exceeds 2 standard deviations above the 30-day baseline. Designed to catch silent checkout regressions: a broken JavaScript event, a payment-method dropdown that no longer renders, a tax-calculation service timing out, an Adobe Commerce 2.4.x patch that changed the address-validation flow.
Calculation
Worked example
A specialty kitchenware retailer on Adobe Commerce 2.4.6 with a US Store View, a UK Store View, and a B2B portal Store View. Tuesday 19 Apr 26, 11:00 GMT. The alert just fired on the UK Store View. 30-day baseline (Tuesdays 11:00 GMT, UK Store View):
Current hour (this Tuesday 11:00):
The alert fired because UK abandonment jumped from a baseline 67.9% to 80.0%. Quote volume looks healthy (446 vs 412 baseline) so traffic isn’t the issue; it’s that 357 of those 446 quote-creators didn’t reach order-complete.
What the investigation reveals:
- The US store is fine. The regression is UK-specific.
- B2B is fine (small sample, Company users with saved cards rarely abandon). This rules out a backend Adobe Commerce platform issue (the UK and B2B Store Views run on the same Magento installation).
- Pulling the most recent 50 abandoned UK quotes: 41 of 50 reached the shipping-method step but no order followed. That points at the shipping or payment phase.
- Stripe Dashboard for the same hour: Stripe shows no payment attempts that failed, just no attempts at all. So the customers aren’t reaching the payment step.
- Adobe Commerce admin > Stores > Configuration > Sales > Shipping Methods (UK Store View scope): a new free-shipping rule was deployed at 09:30 with a syntax error in the conditions, the
Royal Mail Tracked 24method silently disappeared from the storefront for any cart under £50. - The fix: roll back the shipping rule. Within 30 minutes of the rollback, abandonment returns to ~68%.
- Lost revenue estimate for the 1-hour incident window: 357 abandoned quotes × baseline conversion rate (32%) × baseline UK AOV (~£68) = ~£7,750 of revenue at risk if the rule had stayed live for the full day.
Sibling cards merchants should reference together
This alert is most useful when triangulated with funnel-stage cards and payment-side cards.Reconciling against the vendor’s own dashboard
Where to look in Adobe Commerce Admin: Adobe Commerce has a native abandoned-cart view, though it is shaped for outreach rather than spike detection:Reports > Sales > Abandoned Carts lists quotes with at least one item that did not convert, in the Adobe Commerce sense (no associated order). Filter by date range to see the hour-by-hour count. The Adobe view shows the absolute count; this card derives a ratio against quote volume which the Admin view does not natively compute.For Adobe Commerce B2B (paid edition), the Companies view also surfaces unfinished quotes per Company, but again does not compute the ratio. For the order side of the ratio:
Sales > Orders filtered to today, with a counter row at the top, gives the order denominator.Other Adobe Commerce Admin views that look relevant but are not:
- Reports > Sales > Tax: tax aggregations, no funnel signal.
- Reports > Sales > Coupons: coupon usage, lagging indicator at best.
- Customers > Now Online: who is currently logged in, no quote signal.
- Marketing > Cart Price Rules: rules engine, not realtime cart state.
Cross-connector reconciliation (when these connectors are connected for this merchant):
The cross-connector contrast is itself a diagnostic for this card.
Known limitations / merchant FAQs
The alert fired but my Google Analytics checkout funnel looks fine, why? Two common causes. First, GA4 misses 10 to 25% of customer activity due to ad blockers and consent rejection; the card uses Adobe-side server data which is closer to ground truth. Second, the abandonment may be happening at the order-confirm stage (after the customer clicks “Place Order”) which GA4 attributes as a successful purchase event but Adobe never receives the gateway callback to markstate = processing. Cross-check Order State Distribution for pending_payment accumulation.
What’s the difference between the Adobe Commerce Abandoned Cart report and this card?
The Adobe Admin report counts abandoned carts in absolute terms over a date range, optimised for the marketing team to send recovery emails. This card computes a ratio against quote volume and detects deviation from baseline, optimised for engineering and operations to detect regressions. The Admin report is a list; this card is a smoke alarm.
Adobe Commerce vs Magento Open Source: does this card work on the free edition?
Yes. The quote and order tables exist in both editions. The B2B Companies filter (used in some peer cards) requires Adobe Commerce; this alert does not. The only relevant difference is that some Adobe Commerce edition features (Page Builder, customer segments) can change checkout flow and create their own regression sources, those are absent on Open Source.
Why does the alert use 2-sigma rather than a fixed threshold like 75% abandonment?
Baseline abandonment varies hugely between merchants. A B2B-heavy Adobe Commerce store may have 35% baseline abandonment (logged-in repeat buyers); a fashion DTC store may baseline at 78%. A fixed 75% threshold would constantly fire on the fashion store and never fire on the B2B store. The 2-sigma comparator normalises against each merchant’s own baseline and against same-hour-of-week patterns.
My multi-store Adobe Commerce, can I get a per-Store-View alert?
Yes, configure per-Store-View variants in the manifest, filtering both quote and order counts by store_id. This is strongly recommended for any merchant with regional checkout differentiation: different gateways per region, different shipping providers, different tax services. A region-locked regression on the combined view often dilutes below the 2-sigma threshold.
Why does abandonment fluctuate hour-by-hour even without a regression?
Smaller volume buckets are noisier. A merchant doing 50 orders per hour at baseline can swing the ratio by 5 to 10 points just on Poisson variance. The 2-sigma comparator absorbs natural variance; only sustained or extreme deviations trigger. For very low-volume stores (under 100 orders/day) consider widening the bucket to 6-hour rolling.
B2B accounts barely abandon, why include them in the count at all?
You’re right that B2B abandonment baselines are far lower (often single digits). A B2B portal Store View with 100 quotes and 95 orders is typical. By default the card aggregates all Store Views, which can wash out a consumer-side regression. Filter to consumer Store Views only, or set up per-Store-View variants. The B2B portal having its own variant catches a different problem (the supplier relationship breakdown rather than checkout regression).
Quote-create count seems suspiciously high, are bots being counted?
Possibly. Adobe Commerce creates a quote on first add-to-cart, including from headless storefronts and from automated price-checker bots. The card excludes the most obvious pattern (single-item, guest, last-updated within 5 minutes of create) but sophisticated bots can evade. If your quote-create count is more than 5x your active-session count from GA4, install bot management (Cloudflare Bot Management, Akamai BotManager) and treat the elevated baseline as a known data quality issue.
Why doesn’t the alert fire when revenue is down?
Two reasons. First, this is a leading indicator on the funnel; revenue drops are often lagging effects 2 to 6 hours later. Second, a revenue drop caused by upstream traffic loss (an SEO penalty, a paid-ads budget exhausted) shows up as fewer quotes AND fewer orders proportionally, leaving the abandonment ratio steady. Use Revenue Drop Alert for cash-side monitoring and this card for funnel-side monitoring; they catch different failure modes.