Fake Delivery Attempts in India: How to Catch Them

Some portion of every NDR report is fake — the parcel was never actually attempted. Industry data puts it at 12–15% of unsuccessful deliveries. Here's how to spot it, measure it, and use the numbers to negotiate.

The mechanic is simple. A delivery agent has 70 parcels for the day and time for 50. Twenty parcels need to be returned to the hub un-attempted. The agent picks the easiest 20 to mark "customer unavailable" — usually parcels in zones the agent doesn't know well, or to addresses that look incomplete. The brand sees a clean NDR. The customer never gets a call. The parcel goes back to origin.

The systemic cause

This isn't deliberate fraud at the individual level — it's a structural failure of the gig-economy last-mile model:

Agents are paid per attempt logged, not per successful delivery.
Distribution hubs run on inflow/outflow pressure — un-attempted parcels need to clear space for incoming ones.
Route planning at most couriers is poor. Some parcels are always at the margin of what an agent can cover.
COD adds friction — agents must collect cash, so when running behind, COD parcels get dropped first.

The four signals to look for

Signal 1: Same-zone, cross-courier delta. Compare "customer unavailable" rates across two or more couriers serving the same pin code. A 10+ point gap is courier behavior, not customer behavior.

Signal 2: Post-delivery survey "no parcel received" responses. Send a WhatsApp survey to addresses your courier marked "delivered" or "attempted". A 70% "no" rate on attempted parcels is the cleanest fake-attempt indicator that exists.

Signal 3: NDR clustering by agent or run-time. If your courier shares agent-level data, fake attempts cluster — same agent, same hour-of-day, similar zones. Real "customer unavailable" doesn't cluster like that.

Signal 4: NDR vs repeat-buyer cross-reference. Pin codes with high "customer unavailable" rates AND high repeat-buyer percentage are mathematically suspicious. The customers there are clearly contactable. The courier isn't reaching them.

How to use what you find

Three things you can do once you have the numbers:

Re-route at-risk parcels to a different courier. If Courier A is faking 15% of attempts in Pune-411038, route everything to Courier B in that pin code. Track the delta over 30 days.

Switch to photo-POD or geofence-required couriers. Some couriers now require a GPS-tagged photo at the delivery address. Fake-attempt rates on these accounts drop close to zero overnight.

Use the data in your QBR. Most courier account managers will negotiate hard if you walk in with "your fake-attempt rate in our data is 18% and our cost of that is ₹X lakh per quarter." Suddenly the relationship is data-driven, not relationship-driven.

The structural fix

None of the above eliminates the failure mode — it just measures and routes around it. The only durable fix is removing the gig-economy last-mile from the equation entirely: either own riders on dedicated routes, or pickup lockers that don't require an agent to find a customer.

This is, not coincidentally, the architecture Podrones NanoHub runs on. Hyperlocal dispatch from a city hub, with PodBank lockers as the fallback. Fake-attempt rate isn't 5% lower — it's structurally zero, because the failure modes that produce fake attempts don't exist. Read how it works →