Late Shipments, DPD

Metrics type: Key Metrics • Category: Shipping & Courier

At a glance

Absolute count of DPD parcels that arrived after the promised delivery date in the last 7 days. The actionable companion to OTD Rate, OTD says “what percent missed”, this card says “how many tickets are about to land in CS”. Late means past the calendar date; out-of-slot is a separate measurement on Predict Slot Accuracy.


What it counts	`COUNT(parcels WHERE actual_delivery_at > estimated_delivery_at)` over the trailing 7 days. Each late parcel scores 1; multi-piece consignments count one per parcel, not one per consignment.
Delivery success criterion	DPD `parcelStatus = DELIVERED` with a POD scan timestamp. In-transit parcels with no `actual_delivery_at` yet are excluded from this count even if their estimated date has already passed; track those on the in-flight live count.
On-time threshold	DPD’s `estimatedDeliveryDate` per parcel, no grace window. A parcel delivered at 23:55 on the promised date counts on-time; 00:05 the next day counts late.
Service-level scope	All DPD services pooled (Next Day, Next Day by 12:00, Next Day by 10:30, Saturday, Sunday, EU Classic, EU Express). Predict-slot parcels score against their date; the in-slot score is on its own card.
Tracking event semantics	DPD pushes a `delivered` webhook on POD scan. The card uses the first delivered event; subsequent corrections (carded then redelivered next day) leave the original timestamp in place. PUDO-shop drop-offs scan delivered when the parcel reaches the shop, so a late parcel that PUDO-recovered late counts late on this card.
Returns / RTO	Excluded. Refused, undelivered-after-3-attempts, and RTO parcels appear on Returned to Sender, not here.
Geographic scope	UK domestic + DPD UK outbound to EU and rest-of-world. EU origins served by DPD Group sister carriers report through their own connectors.
Peak-period degradation	Q4 typically lifts the count by 2 to 4x vs the trailing 30-day baseline; this is normal for any UK courier. The alert threshold (`>5% of total`) is a rate, not a flat count, so it scales with volume.
Lost vs late	This card includes late-but-arrived parcels only. Truly lost parcels (no scan in 14 days) appear on Open Claims; permanently undelivered surface as Failed Deliveries.
Time window	`7D` (rolling 7 days; the 30D rate is OTD Rate)
Alert trigger	`>5% of total` (late count exceeds 5% of total parcels delivered in the same 7-day window)
Roles	owner, operations

Calculation

Calculated automatically from your DPD 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 UK premium-DTC consumer-electronics brand (AOV £240, ~1,400 orders / week) ships every order via DPD Next Day with predict-slot enabled. Reading taken at 09:00 GMT on 12 Mar 26 for the trailing 7 days (05 Mar 26 to 11 Mar 26).

Service tier	Parcels delivered	Late count	Late rate
Next Day (standard)	1,920	58	3.0%
Next Day with Predict	740	19	2.6%
Saturday	180	11	6.1%
EU Classic	145	17	11.7%
All DPD (this card)	2,985	105	3.5%

The card reads 105 with a 3.5% rate; the >5% of total alert is not tripped at the aggregate. Five things to notice:

The 105 is roughly the CS-ticket volume to expect this week. Industry rule of thumb: ~80% of late-delivery customers contact support (where-is-my-parcel tickets), of which ~25% escalate to refund-the-shipping-fee or refund-in-full. Plan ~85 WISMO tickets and ~20 refund requests against this number.
EU Classic 11.7% is the live problem. Even though the absolute count is small (17), the rate is over 2x the alert threshold for that lane. Likely a customs-clearance issue at the destination depot; pair with UK to EU Cross-Border Exception Rate to confirm.
Saturday 6.1% trips on rate but only 11 parcels. Saturday delivery is paid premium (~£3 to £5 uplift) and customers expect higher reliability. 11 missed Saturdays = 11 weekend-ruined customers; the cost in goodwill exceeds the £1k revenue at risk.
Cost of the 105 misses. At AOV £240 with industry-average 4% post-late refund propensity, exposure is ~£1,000 in refunds + ~£400 in CS handling time + unmeasured churn. Compare against Late-Delivery Revenue at Risk on the DPDLocal connector for the proper finance-CFO calculation.
Compared against same week last year (05 Mar 25 to 11 Mar 25), this brand had 84 late on 2,610 parcels = 3.2%. Volume up 14%, late rate flat. DPD network performance is stable year-on-year for this account; the alert is volume-driven, not service-decay-driven.

Sibling cards merchants should reference together

Late count is the operational trigger metric. Pair it with these to triage the root cause and the downstream business impact.

Card	Why pair it with Late Count	What the combination tells you
On-Time Delivery Rate	The rate counterpart of this absolute count.	Rate = late count divided by total parcels. If volume is climbing while late rate holds, the absolute count rises mechanically; do not panic.
Predict Slot Accuracy	Out-of-slot is a different miss. Most out-of-slot parcels are on-time-by-date and do not show here.	High late count + high slot accuracy = network-wide date misses; high in-slot misses + low late count = predict promise breaking but parcels arriving same day.
Exception Rate	Catches parcels that never reached the customer.	Late + rising exceptions = a slice of “late” is actually “lost or refused”; verify against claims.
OTD by Route	Splits the misses by destination.	One-region late spike = local-depot or driver issue; spread across all routes = peak-period or weather.
Failed Deliveries	The structural-failure counterpart.	Late = arrived after promise. Failed = never arrived (3 attempts then RTO). The two are sequential; today’s failed cohort started as last-week’s late cohort.
Open Claims	Downstream financial impact.	Late count predicts claim volume at 5 to 10 days lag (customer waits a few days, then files).
Cross-connector: `shopify.unfulfilled_orders`	Upstream cause.	Climbing unfulfilled count predicts a late-count spike 1 to 2 days later (warehouse couldn’t pick in time, parcel handed to DPD too late for SLA).
Cross-connector: `shopify.refund_rate`	Downstream impact.	Late count rise of 2 to 3x precedes a 0.5 to 1.5 pp refund-rate rise at 7 to 14 days lag.

Reconciling against the vendor’s own dashboard

Where to look in DPD’s own dashboard: MyDPD Business portal -> Track -> Filter “Late deliveries” -> Last 7 days. The portal shows per-parcel rows with the gap between estimated and actual delivery; export as CSV for the audit trail. The aggregate count is on Reports -> Service Performance -> Late Parcels with the same date filter. Why our number may legitimately differ from MyDPD’s portal:

Reason	Direction	Why
Time zone	Boundary days off	MyDPD defaults to Europe/London (BST or GMT). The card uses UTC. The count for the most-recent boundary day can shift by a few parcels.
Tracking-feed lag	Ours lower for last 4 to 6 hours	DPD pushes status webhooks within 5 to 30 seconds of each scan, but rural-route POD scans may not upload until the driver returns to depot. Most-recent-day numbers fully reconcile by T+1.
Peak-period throttling	Ours lower during BFCM	Webhook queue can throttle during BFCM peak; deliveries may post 2 to 6 hours late. T-2 days fully reconcile.
Predict-slot vs date semantics	MyDPD higher (some views)	MyDPD’s “Late” filter includes both date-misses and slot-misses by default. The card here counts date-misses only. Toggle the MyDPD filter to “Date late only” to align.
Re-attempt timestamp	Ours uses first delivery	If a parcel was carded then delivered next day, the card uses the first delivered scan (next day). MyDPD’s “delivered date” column shows the same. Some merchants run their own warehouse-side reports against the carded date which inflates the count.

Cross-connector reconciliation against commerce platform unfulfilled:

Card	Expected relationship	What causes legitimate divergence
`shopify.unfulfilled_orders`	Upstream cause. Late picks become late shipments.	Persistent unfulfilled count rising while DPD late count holds = warehouse problem not yet hit DPD; expect the spike in 24 to 48 hours.
`shopify.refund_rate`	Downstream impact.	2 to 3x rise in late count -> 0.5 to 1.5 pp rise in refund rate at 7 to 14 days lag.
Cross-3PL: `shipbob.sb_otd_rate`	Peer 3PL outcome metric.	Different parcel populations entirely. Do not arithmetically reconcile.

Known limitations / merchant FAQs

DPD vs DPDLocal, how do I decide which to use for the order I am about to ship? DPD is the higher-tier service (full network, paid uplifts, contracted SLAs around 98%, predict-slot, Saturday and Sunday). DPDLocal is the same network but a lower-cost product line for smaller merchants and lower-volume shippers. For premium DTC over £80 AOV use DPD; for AOV £20 to £60 where every £1 to £2 per parcel matters and same-day promise is not the differentiator, DPDLocal is usually the right answer. The same Late Shipments card on each connector will show whether the cheaper tier compromises service or whether network parity holds. Why does my late count climb during peak even though DPD’s contracted SLA is 98%? Three structural reasons. (1) Carrier networks are saturated in late November and December. Even DPD, which is purpose-built for parcels, sees 30 to 60% volume uplift; depot sortation and final-mile fleet run at capacity. (2) Weather, snow, fog, freezing rain disrupts trucking; 2024 saw two named storms in mid-December that lifted late counts ~30%. (3) Customer-availability friction, recipients are out (work parties, away, travelling) more in December; the first-attempt rate drops, and a re-attempt next day automatically counts late. Plan for a 2 to 4x lift in absolute count vs trailing 30-day baseline; this is structural, not a process fault. How is in-slot delivery measured, and why is a parcel “late” here but “in slot” elsewhere? This card measures the date. The paired Predict Slot Accuracy card measures the customer-chosen 1-hour or 30-minute window. A parcel can arrive on the right date but outside the chosen slot (counts on-time here, out-of-slot there); or arrive in-slot but next day (counts late here, in-slot there if measured against the redelivery slot). Read the two together. For premium DTC where customers paid £4.95 for predict, both numbers matter; for standard Next Day without slot upgrade, only this card matters. The customer says “DPD never delivered” but the card shows on-time. What is happening? Three usual edge cases driven by DPD’s “Your DPD” mobile app. (1) Left-with-neighbour, the recipient gave in-app permission for the driver to leave with a neighbour; parcel scans DELIVERED at the neighbour. The customer collects later or did not realise. (2) PUDO drop, recipient was out and a Pickup shop is enrolled; DPD drops there and scans DELIVERED at the shop. Customer has 7 days to collect. (3) Wrong-address, driver mis-scanned at a similar door (e.g. flat 12 vs 12A); rare but happens around 0.05% of parcels. Pull the POD photo from MyDPD; if the door does not match, file a claim. None of these change the count, but they all generate CS tickets, which is why this card pairs with Open Claims. Why does today’s count look high but yesterday looks normal? Tracking-feed lag. DPD posts most scans within 30 seconds, but rural-route POD scans may not upload until the driver returns to depot at end of shift. The most recent 4 to 6 hours can show inflated late counts because some on-time deliveries have not yet posted. Wait until 06:00 next-day for the count to fully reconcile. Should I optimise to reduce late count or to reduce slot-miss count? Depends on the customer base. Premium-DTC customers who paid for predict slots care more about the slot than the date; tickets are about “you said 10 to 11 am, parcel arrived at 13:45” not “parcel a day late”. Mass-market accounts with standard Next Day care more about the date. Look at the Open Claims card and bucket reasons: if “wrong slot” exceeds “didn’t arrive”, focus on slot accuracy; if “didn’t arrive” dominates, focus on this card. What about returns flow? DPD returns ride either DPD’s “Inbound Consumer Returns” service (label sent to customer, customer drops at PUDO or hands to driver on next collection) or a manual courier return. Returned parcels do not appear here. They appear on Returned to Sender. A late outbound that the customer then returned still counts late on this card. Multi-shipper strategy, when do I add a second carrier alongside DPD? Most premium-DTC brands above ~5,000 parcels per week run two carriers: DPD or DPDLocal as the fast lane (AOV >£60 or expedited), Royal Mail Tracked 48 or Evri as the slow lane (AOV <£40 or non-urgent). Adding the second carrier becomes worth it once the volume on the cheap lane exceeds ~25% of total; below that, the negotiated-rate volume discount on DPD/DPDLocal alone outweighs the dual-API operational overhead.

Tracked live in Vortex IQ Nerve Centre

Late Shipments is one of hundreds of KPI pulses Vortex IQ tracks across DPD 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 or book a demo to see this metric running on your own data.

Get Started

The AI OS

At a glance

Calculation

Worked example

Sibling cards merchants should reference together

Reconciling against the vendor’s own dashboard

Known limitations / merchant FAQs

Tracked live in Vortex IQ Nerve Centre

​At a glance

​Calculation

​Worked example

​Sibling cards merchants should reference together

​Reconciling against the vendor’s own dashboard

​Known limitations / merchant FAQs

​Tracked live in Vortex IQ Nerve Centre

At a glance

Calculation

Worked example

Sibling cards merchants should reference together

Reconciling against the vendor’s own dashboard

Known limitations / merchant FAQs

Tracked live in Vortex IQ Nerve Centre