Field Note: Third-Party Delivery Fraud

If someone can take advantage of a system and make money, then they will. I mean that in the way that Chef Gusteau meant “anyone can cook”: not everyone will try to defraud an app, but if there’s an opening, someone will.

It shouldn’t be surprising then that delivery fleets powered by gig drivers constantly fight abuse; they pay out money, have a low barrier to entry, and at scale it can take a lot of resources to proactively catch and stop scams.

This is a dive into one specific type of scam, one that occasionally breaks through driver app guardrails, but never gets past Getcho’s orchestration safeguards.

Driver marks a delivery as complete, never having started it

People overseeing high delivery volume might be familiar with this.

A “delivered” notification doesn’t prove that an order got to where it needed to go.

Three things can result in an invalid “delivered” notification:

1. (mostly benign) caused by a technical glitch; the job is about to be complete and marked early, or not about to be complete but the miscommunication is soon rectified
2. (harmful but not malicious) caused by confusion or incompetence; a driver delivers the wrong thing to the wrong address, for example.
3. (malicious) driver finds ways to take credit for a delivery never completed, usually by spoofing safeguards – coordinates, delivery photos and signatures

To stop the confusion caused by (1) Getcho waits for at least two delivery signals before considering a delivery complete. As for (2), Getcho’s dispatch team uses delivery reliability tech to flag wrong addresses instantly and escalate or send drivers back – Getcho boasts a 0.05% customer-reported issue rate, well below the typical parcel customer-reported issue rate (0.4%) and much lower than the 3% on-demand fleet users experience *.

* Derived from Getcho apples-to-apples datasets; we collect baselines on partner accounts prior to onboarding.

Hawai’i 5.0: high fraud incidents on island courier deliveries

The malicious type of “incorrectly marked delivered” is of course severely damaging for brands and retailers:

• They’ve paid for a delivery that is not complete, which is either a problem for the claims team or plainly a money leak
• They need to quickly assess and redispatch, which is a problem for ops teams.
• End customers associate the fraud with the brand rather than the delivery provider, which is a problem for marketing and CX teams.

Consider a recurring scam in Hawai’i * where drivers of one particular fleet abused its driver app constantly. They assign themselves to several orders, mark them all picked up with fake proof-of-pickup photos and then immediately mark them as delivered with fake delivery photos.

We can trace the outline of the entire cat-and-mouse game: Bad actors start with clearly phony photos, coordinates and timing. The timing gets more sensible after the engineers set up detection rules. And then finally, drivers start submitting real images of the address from Google Streetview as their PODs.

* Hawai’i itself isn’t responsible for this localization – one bad actor in an arbitrary geography exploits the system and shares the exploit with acquaintances and neighbors.

Getcho: the antivirus for logistics and delivery

Hawai’i customers using Getcho didn’t notice anything amiss. Each time the scam emerged, Getcho flagged the deliveries for review and then automatically redispatched them with new couriers after insta-claiming the delivery fees and reporting the original drivers.

Concurrently in this geo, Getcho’s dispatch logic began favoring fleets not plagued by such shenanigans.

If not for Getcho, customers would keep receiving delivered notifications, go look for a phantom package, and then file reports with the business rather than the carrier. Moreover, our data shows that the average delivery failure is reported by the customer too long after the fact for the business to have a straightforward recourse. Finding out 4 hours later that a courier never delivered the package leaves little room to rectify the issue.

Why Getcho

DNA

Getcho has always been the delivery platform for high-value goods. Our first customers shipped expensive wine, clothing and even jewelry. By necessity we stay on top of each package. We’ve scaled up in the agentic era where we can supplement laborious human reviews with artificial intelligence.

Incentives

While individual couriers are usually dedicated and honest, fleets suffer from a perverse incentive that rewards deliveries being marked as “completed” as quickly as possible, regardless of whether the order actually got to the right place. For an orchestrator like Getcho, on the other hand, a higher quantity of falsely-reported “delivered” orders does not help our bottom line: it just means more time spent processing claims and refunds. That’s why we work hard to proactively prevent and correct delivery issues, protecting brands and saving everyone a lot of headache.