kartel

order assignment and delivery tracking for warehouse operation

This case study reflects a real product I designed professionally. To respect client confidentiality, the name, branding, and interfaces shown here have been recreated independently and do not represent the actual shipped product.

YEAR

2025

TIMEFRAME

8 Months

CATEGORY

Product Design

TEAM

1 product manager, 4 developers,

PROBLEM

When an order can be fulfilled from more than one warehouse, and drivers can only carry a few deliveries at a time, someone has to make dozens of small, time-sensitive decisions every hour - which warehouse has stock, which driver is nearby, which delivery is actually urgent. Doing that manually doesn't scale past a handful of orders a day.

RESEARCH & DISCOVERY

This was a client project, so discovery didn't start from a blank brief. It started from a long list of requirements, and the job was turning that list into a clear picture of how an order actually moves through the business, and where it breaks. How I approached it Broke the requirements down by role. Four people touch every order: the customer who places it, the warehouse or store manager who fulfils it, the driver who delivers it, and the admin who oversees it. I mapped what each of them needs to know at every step, and where one person's action is another person's blocker. Mapped the order lifecycle for every delivery type. Express, next-day, scheduled, and store pick-up each route differently, with different warehouses, time limits, and fallbacks. Laying them side by side showed they share one backbone (placed → packing → assigned → in transit → delivered) with different rules on top. That's what made a single dashboard possible, instead of four separate ones. Treated the failure cases as the real brief. The requirements spelled out what should happen when things go wrong: no stock, no driver nearby, a driver who doesn't start, a late express order, a route that fails to load, a delivery with no OTP. The happy path was never where managers lost time, so I designed around these first. Looked at how other dispatch tools solve the same problem. Food delivery and logistics dashboards already handle "who's closest and free" at scale. The question was what to borrow, and what didn't fit an operation that also has to manage stock across locations. Mapping the dispatch flow end to end made the problem concrete. Assignment wasn't hard in the abstract. It was hard because every single assignment depended on three things nobody could see at the same time: stock, driver location, and driver workload. What that surfaced: Proximity and workload were guessed, not known. Managers assigned drivers based on who was probably nearby and probably free. Stock problems surfaced too late. If the main warehouse couldn't fill an order, someone had to chase other stores by hand, one order at a time. Urgency wasn't visible. A 2-hour express order and a next-day order looked the same in the queue until someone read the details. Problems only showed up after they happened. A driver who never started an order, drifted off route, or missed the 2-hour window became a customer complaint instead of an alert.

DESIGN PROCESS

Instead of assigning orders one by one, a manager bundles orders heading the same way into a job and hands the whole job to one driver, who can accept or reject it. The assignment screen only shows drivers belonging to that warehouse, with their live location and how many orders they're already carrying. Anyone at capacity simply can't be selected. At the same time, stock-hunting came off the manager's plate entirely. If the main warehouse can't fill an order, the system checks the other stores and reroutes it on its own. Express items with no stock are hidden from the storefront up front, instead of being accepted and failing later. The shift: from "assignment takes memory, a map, and a few phone calls" to "assignment is one decision, made with the real picture in front of you."

FINAL SOLUTION

01

Operations dashboard

Order board that moves every active order from pending to delivered, with live KPIs for express timing, driver wait and capacity.

02

Operations dashboard

Order list for filtering by stage, payment, driver and due time, with bulk driver assignment and export across the day’s orders.

03

Operations dashboard

Order detail view that brings items, customer notes, assigned driver, ETA, cash to collect and delivery OTP status into one place.

04

Operations dashboard

Driver assignment workspace that queues waiting orders and recommends the best-fit driver by distance, open slots and vehicle.

05

Operations dashboard

Returns board for tracking refund and exchange requests from pickup through inspection to restock, with quick return creation.

06

Operations dashboard

Refund review queue for checking amount, reason and original payment before approving, with supervisor sign-off on larger refunds.

07

Operations dashboard

Warehouse rules screen for setting express windows, driver wait limits, on-time targets and per-driver capacity that drive alerts.

08

Operations dashboard

Manager profile for personal details, default warehouse, role-based access and sign-in security, including two-step verification.

DESIGN DECISIONS WORTH CALLING OUT

I disabled over-capacity drivers in the list instead of showing a workload number, because a limit a manager has to remember and enforce themselves gets ignored under time pressure. The rule needed to live at the point of decision, not beside it. I kept both a Kanban and a list view instead of picking one, because triaging urgent orders and looking up a specific order by customer or date are different jobs, done at different moments in a shift. Forcing one view would have made one of them slow. I designed for exceptions instead of expecting managers to watch the map, because the failures that actually hurt - an express order past its 2-hour window, an order not started half an hour after assignment, a driver far off route - are rare but expensive. The system raises them as alerts the moment they happen, and the manager's next step is reassigning, not investigating.

OUTCOMES & IMPACT

The dispatch model - route-based jobs, a capacity-aware assignment view, automatic rerouting of out-of-stock orders, and exception alerts - made up the core scope. Driver-facing extras like per-delivery earnings, commission tracking, and salary views were deliberately held back for a later phase so the dispatch core could come first.

KEY TAKEAWAYS

The interesting problem wasn't the dashboard, it was deciding how much to automate. Some things the system should just do: find stock elsewhere, block an overloaded driver, raise an alarm when an express order is late. But picking the right driver, not just an eligible one, stayed with the manager, because drivers are people with context a system can't see. I'd want to say where I think that line landed well, and where I'd move it now.

CHALLENGES

CHALLENGES

CHALLENGES

Designing for things going wrong: no stock at the main warehouse, a driver who never starts, a missed express window, or a delivery that can’t be confirmed by OTP.

FUTURE OPPURTUNITIES

FUTURE OPPURTUNITIES

FUTURE OPPURTUNITIES

  • Driver-facing views for earnings per delivery, commission and targets.

  • Smarter assignment that suggests jobs from route, load and due time.

SAY HELLO

i'm open for freelance projects, feel free to email me to see how can we collaborate

SAY HELLO

i'm open for freelance projects, feel free to email me to see how can we collaborate

Made with ❤️ by Vaisakh

Made with ❤️ by Vaisakh

Create a free website with Framer, the website builder loved by startups, designers and agencies.