A blockchain solution for international Amazon deliveries

Justin Goro
Mar 17, 2018 · 7 min read

Like most people with an internet connection, I marvel at the wide array of products and services available online, but unlike people living in developed countries, I live in a developing country and often have to marvel at all the online goodies through a fence of text that reads “this item does not ship to your country.”

In the case of all the Amazon products often denied to me, it is the unfortunate flip-side of Amazon’s excellent customer service: sellers are often pressured to refund customers in the face of late shipping, even if the sellers have no control over the logistics involved. If a country is known for stalling international post at customs, sellers might choose to block that country entirely, simply to remain in Amazon’s good graces. I recently had an experience where I simply inquired with an Amazon retailer if they knew why my package was taking so long. They immediately refunded me (even though I didn’t request it) and a week later I received the package (don’t worry, I sent the money back to the retailer).

In the face of such uncertainty, many retailers have taken my country off their list after bad experiences, even though I and many others here would willingly put up with the delays and risks. This can be immensely frustrating for people outside of the group of favoured countries, who feel as though they’ve been excluded from a market they can technically afford to participate in.

I’m going to propose a blockchain based solution to this problem that not only allows people in developing countries to consume anything they see for sale online but at a much lower cost than the existing solutions to this problem. But first, let’s review the economics of the existing forwarding services. Once we understand the economics, it will become clear how a decentralized blockchain based approach would be more efficient than the prevailing solutions.

Disclaimer: I should mention for clarity before going ahead that items that are fulfilled by Amazon are always promptly delivered at low cost. The types of retailers I’m using in this example are independent retailers who use Amazon, Etsy, Ebay etc to create an online presence but who organize shipping themselves. This is certainly not a hit piece on Amazon. On the contrary, I’m a very big fan of theirs. Praise Bezos. I’m focusing on Amazon because the independent Amazon retailers are held to a high standard which means the more risk averse ones are choosing not to ship internationally.

An old fashioned, centralized solution

In response to lost business and cut off consumers, numerous services have popped up which offer a U.S. based address for you to ship to. They then forward the mail and assume the burden of international shipping — at great expense to the end consumer. Using one online calculator (I don’t want to mention names), it cost about $60 to get an item the size of an A4 book from the UK to my country.

Multiple Hops: a risk reduction strategy

The solution above breaks the trip from Amazon to my door into 2 hops: one from Amazon to the forwarding service provider (FSP, for short) and another from the FSP to me. The FSP faces great uncertainty that the Amazon retailer was unwilling to tolerate. In return, they factor the risk of unsuccessfully delivering the package into the cost of forwarding which accounts for the steep price. It may not seem intuitive but we can show using probability theory that breaking the trip into multiple hops can actually reduce the total shipping cost, saving the consumer money on shipping.

Suppose we’re having an item shipped from the California to Kenya. The costs include the flight to D.C., the flight to Kenya and the local courier delivery from the airport to the address.

In each of these, there is a risk of failure from an unforeseen event. For instance, Kenyan Customs might seize the product or damage it during inspection, the U.S. authorities might block the shipping of the product, the transport itself might fail along the way etc. Let’s assign fictional probabilities of success to each step:

Pr(Amazon seller to FSP in California) = 0.95
Pr(FSP to D.C.) = 0.9
Pr(D.C. to Nairobi) = 0.89
Pr(Clearing Customs in Nairobi) = 0.65
Pr(Nairobi Airport to final address) = 0.7

Because these events are independent of one another, the final probability of successfully shipping from Amazon to Kenya is the product of all of the above: 0.34 or 34%. The chance of failure is therefore 66%.

Let’s suppose that the business model of the FSP is that in the face of a failed delivery, the FSP will refund the consumer the value of the purchase. They will have to factor into the final price the cost of refunding the user 2 out of 3 times in this fictional example.

To calculate the risk premium the FSP charges:

Risk adjustment to cost =  0.66 * Value

This means that on average every 3 times, 2 packages will be lost, costing the FSP 2 packages of refunds. Because they’d been charging 2/3 times the value of the package, every 3 attempts, they’d have (2/3)*3 = twice the value of package which matches up to the 2 packages lost and allows them to offset the costs perfectly.

Suppose instead that we broke up the trip such that a different FSP would assume responsibility at each step so that there would be a relay of handovers (or hops) from Amazon to FSP1 and then from FSP1 to FSP2 in D.C. and then from FSP2 to FSP3 in Kenya and then from FSP3 to the final address. The new business model is that each FSP would place a deposit equal to the value of the product divided by the number of hops. In other words, if there are 4 FSPs on the route and the package is worth $80 then each re-shipper would place a deposit of $20. If the package doesn’t reach its destination, all deposits are sent to the consumer.

Comparing costs

Let’s assume that we’re shipping a $100 package. Under the old model, the risk charge that the FSP would have to add is 0.66*$100 = $66

In the new model, there are 3 FSPs which means that each has to place a deposit of $33. To calculate the total adjustment that we must make to the shipping costs to account for the sum of the individual risks, we simply work out the risk adjustment for each FSP and sum it all up:

Cost(Amazon seller to FSP1) = (1 - 0.95) *33 = 1.65
Cost(FSP1 to re-shipper2 in D.C.) = (1-0.9) * 33 = 3.3
Cost(FSP2 to Kenya) = (1-0.89) *33 = 3.63
Cost(FSP2 for Clearing Customs in Kenya) = (1-0.65) * 33 = 11.55
Cost(Nairobi Airport via FSP3 to final address) = (1-0.7)*33 = 9.9summing the above totals yields $30.03

Using this new model, the total supply chain of FSPs need only add $30.03 as the risk adjusted cost for shipping as compared with $66 previously offered by the solitary FSP. The model that makes all FSPs pay for failure, regardless of where it occurred creates a skin in the game incentive for each FSP to choose the next link in the chain carefully when handing over the package.

Speaking of chains…

Now that it is clear that a decentralized international shipping network is more efficient than a traditional 1 or 2 hop courier trip, we can implement this solution using existing blockchain technology. My choice would be Ethereum because all the technology is in place to implement a decentralized multi-hop courier service. Let’s outline a minimum viable dApp:

Ethereum Forwarding Service

Couriers would register themselves on-chain as a forwarding service provider (FSP) with 4 statistics: maximum radius of service in km, price per volumetric weight(a density measurement used by couriers to calculate shipping cost), maximum value per shipment, and a blockchain friendly location value that corresponds to specific geographical co-ordinates. Another smart contract could be used to monitor successful handovers and record these scores for each shipper, creating an on-chain, verified reputation system. Each product shipped can have its unique reference number stored on-chain as a hash, creating a lookup table for auditing purposes.

Finally, accompanying software would calculate the shortest, cheapest route from seller to buyer using graph weighting algorithms similar to what the Lightning Network uses in its onion routing strategy and it would initiate a multi-hop transaction on a smart contract that waits for each courier to agree to the route by placing their deposits into a smart contract. Once all deposits are placed, the consumer is instructed to place the total cost of the delivery into a smart contract and have the parcel delivered to the first hop on the route. From there, the hopping continues without direction from the consumer. Once the package is delivered, the deposits of every courier is released by the smart contract as well as the payment from the consumer for shipping.

Optional: There of course may be disputes in the process but I didn’t want to complicate the MVP with a dispute layer just yet. However, the solution above provides enough of an audit trail that a layer of what Max Borders refers to as hypermediation can be used to resolve disputes. For instance, a judiciary DAO that uses international arbitration best practices borrowed from the private sector to resolve disputes and encourage honest behaviour (example inspired by Tom W Bell’s excellent book on decentralized governance, “Your next Government?”).


There are so many opportunities to improve the world with decentralizing technology but I am only one person and so I’ve chosen to spend my time on one project (at a time). For this reason, I don’t have the time to create this FSP dApp but my hope is that someone reading this finds it inspiring enough to try their hand at creating it (even if just a proof of concept for attracting funding or attention). If you are reading this and would like to create this service, I’ll be sure to give you as much free publicity as my medium audience can tolerate.