Case Studies · Transport

OilCo Bangladesh, online and at scale

OilCo Bangladesh's legacy PostgreSQL stack had years of duplicate part records and inconsistent SKU data. We migrated to a NoSQL model with deduplication built into the conversion, retained around 90% of the data, and launched a high-transaction-volume eCommerce site on top.

Optimized

Performance

Scalable

Stack

High-volume

Traffic

Problem

OilCo's PostgreSQL data had years of duplicates and inconsistent SKU shapes, and the model couldn't carry the high-transaction eCommerce the team wanted to launch.

Approach

We migrated PostgreSQL → NoSQL with deduplication built into the conversion (retaining ~90% of the records), built an eCommerce site on top with PortPos integrated for payments, and tuned the stack for high transaction volume.

Outcome

A clean, deduplicated catalogue. A site that holds up under heavy traffic. Payments clear through PortPos and reconcile back into BMS as soon as they settle.

OilCo Bangladesh sells auto-parts at the kind of volume where small data problems multiply quickly. By the time we got involved, the PostgreSQL database underneath their operation had years of duplicates — the same part recorded three different ways, customers with multiple records, orders linked to whichever copy happened to be top of mind that day.

Migrating the data

We mapped the existing PostgreSQL schema, designed a NoSQL document model around how the team actually thinks about parts and customers, and ran the migration with deduplication built into the conversion. About 90% of the original data carried through cleanly; the rest was either truly redundant or unsalvageable noise that the team agreed wasn't worth carrying forward.

A catalogue organised the way auto-parts customers actually shop

Auto-parts is a category of categories. We restructured the catalogue around the way customers actually narrow it — vehicle type, brand, system, application — so a buyer reaches the part they need in three or four clicks rather than scrolling through everything that's loosely related. Categories, sub-categories and cross-references are all editable from BMS, so the structure can evolve as the product range does.

A design that handles depth

Auto-parts sites have to carry a lot of information while staying easy to scan. We designed the surface around hierarchy and clarity — generous typography, photography that frames each part cleanly, product pages that lead with what serious buyers look for first, then the supporting specs and variants. Each piece of information sits at the level its readers expect.

Built for high performance and high volume

On top of the new data layer, we designed and developed an eCommerce site sized for high online transaction volume — fast page loads, server-side rendering for the product detail pages, and a stack with scope to handle traffic spikes. PortPos is integrated as the payment gateway, which the team's customers already use; checkout reconciles against the order in BMS the moment payment clears.

All powered by BMS

ProSystem BMS sits behind every surface — catalogue, stock, payments and customer records all live there. The website is one face of that record; the back office, reports and operations team's day all draw from the same place. One source, multiple surfaces.

Where it sits today

The website always shows what's actually on the shelf. The team has stopped manually reconciling duplicates and started focusing on supplier relationships, SKU expansion and the marketing that drives the volume the platform can now handle.

At a glance

BMSeCommerceMigrationAuto-parts
Client
OilCo Bangladesh
Industry
Transport
Went live
Jan 2025
Status
Running

Related work