State of Agent Shipping 2026: What 19,000+ Tracked Parcels Show

Key findings

Median door-to-door delivery from China agents to the US is ~16 days (n=196 parcels; rep.tools dataset, July 2026).

The UK is the fastest high-volume route we track: median ~9 days, typical window 7-16 days (n=35).

Canada runs 10-20 days typical with a 14.5-day median (n=138, all agents); the 69 parcels from our own KakoBuy account inside that sample show the same 14.5-day median (fastest 7.1 days, slowest 73.5).

Germany (n=264, our largest route sample) runs 12-24 days typical, median ~16, with DHL dominating the final leg.

South Africa is the slowest measured route: median ~48 days, typical window 13-60 days (n=16).

Direct single-carrier lines (EMS-family tier) run roughly 40% faster than consolidated budget lines: US direct-line parcels delivered in a ~10-day median (n=815, full account export) versus the 12-19 day consolidated-inclusive window; Canada direct ~9 days (n=272) versus 14.5 measured on consolidated lines. Paying for the faster line buys you about a week.

The consolidated China leg shows no tracking scans for the first 5-10 days on most routes — the single most common cause of false 'lost parcel' alarms.

Chinese New Year submissions can stall warehouse dispatch for weeks: our worst first-party case took 26 days from payment to first scan (median is 1 day).

Why the measured column shows hundreds of parcels, not 19,000

Transparency about the funnel, from the full account export (July 2, 2026): 19,000+ unique trackings created all-time → 7,455 unique confirmed deliveries → 4,151 with a known destination country → and the percentile table below is built on the subset whose timelines survive honest measurement (a known destination AND an untruncated first-scan-to-delivered timeline; consolidated budget lines swap carriers mid-route, which truncates most timelines). One column note: the deliveries count uses explicitly recorded destinations from the raw export, while the measured column also counts parcels attributed via their carrier (DHL Germany numbers are German parcels regardless of what the destination field says) — which is why Germany's measured sample can exceed its recorded-destination count. We'd rather publish hundreds of honest parcels per route than thousands of polluted ones — and the funnel is exactly why: our own 69 first-party consolidated-line parcels to Canada measured a 14.5-day median door-to-door, while the untruncated (mostly direct-line) subset alone would claim ~9 — so the windows below keep the consolidated-inclusive calibration that matches measured reality.

Methodology

Parcels are tracked through the rep.tools parcel tracker (public, free) and our own KakoBuy account. The delivery window measures first carrier scan after agent handover to the delivered scan; where a single tracking number truncates at a carrier swap, the door-to-door window is reconstructed by stitching the China leg and destination leg from the same parcels. Warehouse/consolidation time before ship-out is excluded — it's buyer-controlled. Routes appear here only with n≥10; smaller samples stay internal until they mature. Percentiles refresh as new parcels complete (last rebuild: 2026-06-18; page updated July 2026).

Known limitations: the tracker over-represents KakoBuy and Canada (our own account), under-represents express courier lines (users track worried parcels — couriers rarely worry anyone), and single-number tracking truncation means some windows are modeled from leg-stitching rather than observed end-to-end. Sample sizes per country also reflect carrier data quality rather than true traffic mix: Germany leads (n=264) because DHL's feed reliably exposes destination and full timelines, while many US-bound lines don't report a destination at all — so US parcels fall out of the clean-attribution filter disproportionately, and the US sample (n=196) understates the US's actual share of parcels.

Use this data

The full route table is downloadable as CSV for journalists, researchers and community spreadsheet maintainers: /static/agent_shipping_2026.csv. Attribution: "rep.tools parcel dataset" with a link to this page. If you need a cut we don't publish (per-line, per-month), contact us via /contact — if the sample supports it, we'll share it.

Disclosure: the KakoBuy link is an affiliate link — it costs you nothing and supports rep.tools. Every fact above is stated the same way regardless.

Frequently asked questions