How the 61-day log was kept and what counts as a delivered parcel
The log holds one row per parcel and four timestamps: the moment we released it, the moment the warehouse marked it outbound, the first trunk scan, and the delivery confirmation. A parcel counts as delivered only when the final timestamp exists, and 21 of the 24 rows reached that state inside the window (Recorded).
Recording opened on 2026-08-01 and closed on 2026-09-30, which is 61 days. Every row came from an order we placed ourselves, which keeps the collection method identical across the set and makes the comparison between rows meaningful.
1. Timestamp the release. 2. Timestamp the outbound mark. 3. Timestamp the first trunk scan. 4. Timestamp the delivery and subtract the first figure from the last.
Three of the 24 rows were still moving when the window closed, so their day counts are censored rather than complete. We report those rows separately and keep them out of the median, which is why the headline figure rests on 21 delivered parcels.
Timestamps arrived from three places: our own notes at release, the order screen for the outbound mark, and the line tracking for the scans. Mixing three sources is the main weakness of the log, because a note typed in the evening can date an event to the wrong day, and a single misdated row moves a median once the set is this small.
The 24 parcels in the record, grouped by line and by weight
By weight the set splits into 14 parcels under 2 kg chargeable and 10 at or above it. The lighter group held a median of 11 days and the heavier group 13, a two-day gap that 24 rows cannot prove is a real effect (Recorded).
By line the set spreads over four UK services. Two of those carried 17 of the 24 parcels, and the remaining two carried three and four, which is far too thin to publish a median for without a warning attached.
Weight also drives splitting. Above 4.5 kg chargeable the ladder we keep adds £4.20 for each extra kilogram (Recorded, 2026-09-25 review), so a heavy batch is more likely to travel as two parcels, and a split parcel re-enters the warehouse queue before it leaves.
Naming the lines would not explain much, because weight is the pattern we found rather than line performance. Parcels above 4.5 kg were split far more often, and a split adds a second opening rung on the ladder plus a second journey through the same packing bench, which is where two extra days usually come from.
One further split of the set is worth stating plainly: the six parcels released on a Thursday reached delivery a median of one day later than the parcels released on a Monday. Twenty-four rows pointing at a weekend is weak evidence, and it is the kind of weak evidence we prefer to publish rather than round away (Recorded, 24 parcels).
Parcel count matters as much as weight. Nine of the 24 rows were second parcels from an order that had already shipped once, and those rows carried a median of one extra day, because a split parcel passes the bench twice and pays a second opening rung on the ladder we record (Recorded, 2026-09-25 review).
Cross-check the two weight groups against the split count before drawing anything from them. All four parcels above 4.5 kg were second parcels from an order that had shipped once already, so the extra day in the heavier group belongs to the split rather than to the weight, and the honest reading is that we cannot separate the two effects with 24 rows.
Where the days go: warehouse, trunk and last mile
Warehouse: median 2 days, counted from release to the outbound mark. That single figure covers shelving, QC photography and the packing queue, and it is the one stretch where a shopper can still change a size or a colour without cost.
Trunk: median 9 days from the outbound mark to the first UK scan. This is the longest stretch in the whole journey and the one we can influence least, because it depends on line capacity and on customs processing rather than on anything we do.
Last mile: one to three days from the UK scan to delivery, with one day typical for a weekday address and three when a first attempt fails and a redelivery is booked.
1. Split the total into warehouse, trunk and last mile. 2. Compare each stretch with the medians of 2, 9 and 1 to 3 days. 3. Chase the stretch that is running over instead of chasing the whole parcel.
The warehouse figure deserves its own breakdown, because it hides three queues. Shelving is usually done within a day, QC photography waits for a slot, and the packing bench is the part that stretches when volume rises. Only the order screen timestamps let us separate those three, and we have not yet logged them separately.
Customs processing sits inside the trunk leg rather than beside it. A parcel can wait two days in a clearance queue without producing a single tracking event, which reads exactly like a lost parcel on a phone screen and is almost never one.
Bias in our own log: what a tracking page cannot show us
A tracking page shows the last event, not the work in progress. A parcel can sit in a packing queue for two days with no new event, and 2 of our 24 rows showed a three-day quiet gap that turned out to be ordinary handling (Recorded).
Nothing before release is visible in our timestamps. Domestic freight to the warehouse, seller handling time and any relisting delay all sit outside the log, which means the 12-day median measures the part of the journey we can see and not the whole of it.
Survivorship applies to any set of parcels that arrived. Logging what turned up pulls a median towards the parcels that behaved, and the 27-day case in the same window is the reminder that the tail runs longer than the middle.
Season matters more than any of the other biases in this log. Our window runs from August into September, which sits outside the two queue windows we describe elsewhere on this site, so the medians here describe an ordinary season rather than a busy one, and a busy season would move them.
Manual timestamps carry a bias we cannot remove. We typed them in the evening, so an event that happened at 09:00 on Monday and an event that happened at 17:00 on Monday are recorded as the same day, which flattens short gaps and makes the four-day cases look slightly longer than they were.
Which figures you can reuse and which are out of scope
Reuse the medians as planning figures: 12 days in total, 2 in the warehouse, 9 on the trunk leg and 1 to 3 for the last mile. They come from 24 parcels over a 61-day window and they describe one shipper, one destination country and one season.
Do not reuse them as a promise. A 12-day median says that half the parcels arrived sooner, nothing about the parcel you are about to release, and the 27-day case sat inside the very same set of rows.
Out of scope: other destination countries, freight-only services, and any month outside 2026-08-01 to 2026-09-30. For a peak-week figure we hold no measurement at all, so treat every peak claim you read elsewhere as Not verified.
1. Quote the median together with its sample size. 2. Quote the 7 to 27 range beside it. 3. State the window. 4. Refuse to extend any of the three figures beyond that window.
Treat our figures as a sanity check on any timeline you are offered. A promise of five days from release to a UK door sits below the fastest parcel in this record at seven days, which does not make the promise false, only unusual enough to ask a second question about.
Quote the count alongside the days. Twenty-four parcels with 21 delivered inside the window is a small set, and a reader who knows that number will read a 12-day median differently from one who is shown the median alone.
A day count you can defend to someone else
A defensible day count carries three parts: the number, the sample and the window. Ours reads 12 days, from 24 parcels recorded between 2026-08-01 and 2026-09-30.
Give the spread as well. Anyone who hears a median without a range treats it as a deadline, and the honest answer is 7 to 27 days with a median of 12.
Name the parts you did not measure. We cannot see the domestic leg, the seller handling time or the days a parcel spends waiting for a payment confirmation, so 12 days is a lower bound on the whole journey from listing to doorstep.
1. State the median. 2. State the range. 3. State the window and the count. 4. Name what the log leaves out.
Ask a seller the same three questions this log answers: how many parcels, over what window, and measured from which event to which. A seller who answers all three is describing a record; a seller who answers none is describing an impression, and the difference shows up the first time a date matters.
Compare a quote with our median in the same units. Ask whether the five days or the ten days being promised covers release to delivery or only the trunk leg, because our 12-day median splits into 2 warehouse days, 9 trunk days and 1 to 3 last-mile days, and a quote that names only the middle figure is describing less than half of the journey.
The data point behind this note
Outbound-to-delivery time across 24 recorded UK parcels in the 2026-08-01 to 2026-09-30 window ran from 7 to 27 days with a median of 12 (Recorded).
Rates last checked 2026-10-02. Where a figure is community-reported we say so; where we could not verify it, we write Not verified instead of estimating.