Logistics · Design engineering / 2026

Swigo

Make a delivery easy to follow for the person sending it and the rider carrying it.

My role
Design Engineer · design direction and front-end code across four apps
Scope
Design direction and front-end code for the customer site, rider app, pharmacy portal and operations
Year
2026
Product
Visit website
Team
Swigo is part of Swiftmedix. I authored 64 of the repository’s 75 pull requests, as matt-marly on GitHub.
64 of 75
repo pull requests I authored, all merged
4
apps under one design direction
237
hard-coded colours found in my audit
11 days
first to last merged PR, Nov 2026

Counts from the Swigo GitHub repository, reviewed 22 November 2026. Swigo is live and still in progress. The repository is private, so the sources below link to the live site.

Swigo product screens
The live Swigo homepage in November 2026. The parcel service now sits beside pharmacy orders and emergency help.
01 / Overview

One order, three people looking at it

Let me start with what Swigo is and how I got involved. Swigo Logistics is a delivery service in Lagos, run by its parent company Swiftmedix. It is live at swigologistics.com and still in progress. A customer books a pickup, a rider collects the parcel and an operations team coordinates the trip in between. All three look at the same order from a different screen.

I joined in November 2026 as design engineer, so I designed the screens and wrote the front-end code for them. Over eleven days I authored 64 of the repository’s 75 pull requests, all merged. They cover the customer site, the rider app, the pharmacy portal that came later and the operations workspace. I also made the film that plays on the Swigo homepage, shown below.

64 of 75
pull requests in the repo, authored by me and merged
11 days
8 to 18 November 2026
The homepage leads with the two things most people come to do: send a parcel or track one.
The homepage leads with the two things most people come to do: send a parcel or track one.
The Swigo homepage film, which I made.
02 / Problem

Everything looked the same, so nothing was clear

This section is about what I found when I audited the apps. Each one had grown its own style. I counted 53 corner radii, 126 shadows, 65 font sizes and 237 hard-coded colours, with two separate form systems. A button, a filter, a tab and an address field were all the same white pill, so people could not tell what was tappable.

Phones made it worse. Header links wrapped onto a second row, and on a 390 pixel screen the tracking map escaped its box and stretched to 435 by 1928 pixels. For a rider reading a screen in the sun, or a customer checking where their parcel is, that kind of mess costs real time.

237
hard-coded colours across four apps
53
different corner radii
65
font sizes
The phone views after the mobile audit. A menu drawer replaced the wrapping links, and the tracking map now stays inside its own card.
The phone views after the mobile audit. A menu drawer replaced the wrapping links, and the tracking map now stays inside its own card.
03 / Approach

Give every colour and shape one job

Here is how I decided what to change before touching individual screens. I wrote a design direction that the whole team, and every later change, had to follow, then put it into the code as shared styles and components. Plum means do this, and appears once per screen. Indigo means go there. Mint means live or done. Buttons are pills, tabs are never pills, and fields are the only things with an edge.

The rule I cared about most was words first, colour second. Every state shows a plain word, like You’re online or Waiting 12 min, and colour only backs it up. Riders often read on cracked phones in bright light, and a colour on its own is easy to miss.

Tracking on desktop. The status reads as words, the steps share one timeline component, and the single plum button is the action to take.
Tracking on desktop. The status reads as words, the steps share one timeline component, and the single plum button is the action to take.
04 / Challenges

The screen can only promise what the system knows

The hardest part of this project was honesty, and this section covers why. A customer returning from checkout has not necessarily paid; the backend has to verify it first. A rider who is offered a job has two minutes to accept, and the delivery is not theirs until they do. Live GPS pauses when the rider leaves the page.

Each of those moments could easily show a reassuring screen that turns out to be wrong. So I designed the in-between states as carefully as the happy path. When the booking service cannot answer, the page says so, offers a retry and confirms that nothing was charged.

2 min
for a rider to accept a job offer before it moves on
A real failure state from the live site. It names the problem, says nothing was charged and gives one clear way forward.
A real failure state from the live site. It names the problem, says nothing was charged and gives one clear way forward.
05 / How we solved it

Shorter first steps and a visible next step

This part walks through the changes riders and customers actually see. For riders, I restructured the application so the first step only asks for city and vehicle, and heavier details like the plate number come later. A status tracker then shows where a sent application stands, from documents checked to ready to ride. The application has since grown to seven saved steps on the live site.

Inside the rider app, a short tutorial called How a shift works opens on the first idle shift and stays one tap away. On the customer side, the phone home became a grid of real services, so booking, tracking and orders are reachable without scrolling past a hero image.

The rider application on the live site. Eligibility questions come before the account, so nobody fills in a long form only to find they cannot ride.
The rider application on the live site. Eligibility questions come before the account, so nobody fills in a long form only to find they cannot ride.
06 / Lessons

A reassuring screen has to be an accurate one

To close, what I am taking from Swigo. The audit numbers taught me that a design system is mostly about saying no. Each new colour felt harmless when someone added it, and together they made the product hard to read.

The bigger lesson is about language. An application sent is not an application approved, and a rider offered a job is not a rider assigned. As Swigo grows into pharmacy orders and emergency help, getting those words right matters even more. Swigo is live and still in progress, so this case covers the design and the code rather than delivery numbers. The next thing I would do is watch riders in Lagos use these screens on their own phones.

The emergency page tells people to call 112 first, before anything Swigo can do. It shows the same words-first rule on a higher-stakes screen.
The emergency page tells people to call 112 first, before anything Swigo can do. It shows the same words-first rule on a higher-stakes screen.

Based on my merged pull requests in the Swigo repository and the live Swigo site. Reviewed November 2026. Screens show the version documented in this case study.

NEXT PROJECT

Stratum V2

Stratum V2 preview