Automated Box Sizing & Carrier Routing for a Coffee Subscription by Vladislav PidberezhnikAutomated Box Sizing & Carrier Routing for a Coffee Subscription by Vladislav Pidberezhnik

Automated Box Sizing & Carrier Routing for a Coffee Subscription

Vladislav Pidberezhnik

Vladislav Pidberezhnik

Verified

PROJECT OVERVIEW
A specialty coffee roaster running a subscription programme needed the correct shipping box and carrier service selected automatically for every order.
Their fulfilment team was reading box sizes off a printed ten-row chart and choosing services by hand. Meanwhile, every order in the account — regardless of weight or destination — was being shipped on USPS Priority Mail, the most expensive domestic option for their package profile.
The work spanned four connected systems: Shopify held the product data, Loop Subscriptions identified subscribers, BOGOS.io added a free bonus bag to first-time orders, and ShipStation had to read all of it to pick a box and a carrier.
THE CHALLENGE
The stated brief looked simple. A ten-row chart mapping order weight to box size, with a second column for first-time subscription orders that included a free "green sample bag." Plus: route subscription orders to USPS Ground, and anything over 8lb to UPS Ground.
MY ROLE
Sole consultant, end to end:
Discovery across four systems and interviews via the operations lead Audit and documentation of the pre-existing automation estate Solution architecture and tolerance analysis Implementation in ShipStation Client communication, scoping, and estimation Written handover documentation
RESEARCH AND KEY INSIGHTS
The ten-row chart was really a five-box ladder The chart listed ten scenarios. Reading it as a structure rather than a list, it collapsed: five box sizes in ascending order, where "with bonus bag" simply meant starting one rung higher, and the largest box was a ceiling.
That reframing cut the rule count roughly in half and meant a future size tier would be one addition rather than two.
The bonus bag didn't need a rule at all Everyone — client included — assumed the automation needed to detect the bonus bag and "bump up a size." But because BOGOS adds it as a real product SKU with a real weight, the order weight already accounts for it.
At 6oz, the bag pushes every order past its weight boundary into the next box on its own. The client's own summary — "any order with the additional green bag will go up a size with the exception of 5lb orders" — turned out to be an emergent property of the arithmetic, not something that needed implementing. The 5lb exception fell out too, since 5lb was already in the largest box.
ShipStation has a native USPS Zone condition I initially designed around the assumption that it didn't, planning to translate zones into state lists — a tedious, imprecise approach requiring a USPS zone chart lookup and ongoing maintenance.
It was wrong. ShipStation exposes USPS Zone directly as a rule condition. Correcting that removed 2–3 hours of estimated work, eliminated a maintenance burden, and made the routing exact rather than approximate. I revised the estimate down and told the client why.
The subscription tags came from a custom field Loop's subscription identifiers weren't arriving as ShipStation tags. Three existing rules were reading Custom Field 3 and generating the tags.
That meant matching the source field directly instead of the tags: one condition instead of two, catching both first-time and recurring subscriptions, and eliminating a rule-ordering dependency that would otherwise have been invisible and fragile.
The existing rules were a data source Their thresholds encoded knowledge nobody had written down. A rule cutting at "≤17oz" for 1lb orders implied the real product weight was about 16.5oz, not the nominal 16. That corroborated my weight bands against production reality rather than assumption.
WHAT I LEARNED
Read the existing system before designing for it. It changed the design three times and prevented a silent collision in production.
Design tolerance, then state it. When the bonus bag weight changed mid-build, I could confirm in seconds that nothing needed editing.
The most valuable findings were the ones nobody asked for — a broken rule, a coverage gap, an address error, and a quantified loss on their most common order.
Like this project

What the client had to say

Vlad is amazing! We needed work with intricate detail completed on a tight deadline and he really came through!

Bethany Ren, Graffeo

Aug 4, 2026, Client

Posted Aug 5, 2026

Built ShipStation automation that picks the right box and carrier for every subscription order automatically, replacing a manual chart lookup.

Likes

1

Views

0

Timeline

Aug 3, 2026 - Aug 4, 2026

Clients

Graffeo