How a B2B Marketplace Built Real-Time Pricing Intelligence Across Five Markets

A B2B e-commerce company headquartered in Singapore connects global brands with distributors, resellers and independent sellers across five markets. As it expanded, its pricing data stayed scattered across disconnected systems, so pricing decisions were made on numbers that were already out of date. OnPoint Insights built a real-time analytics foundation on AWS, consolidating pricing data into a Redshift warehouse and putting product, distributor and reseller performance into interactive QuickSight dashboards.

Pricing data in one place

Pricing and operational data from disconnected systems consolidated into structured objects built specifically for pricing analysis.

Real-time, not overnight

AWS Kinesis and Data Migration Service move data continuously, so pricing decisions use current distributor and market activity.

Product, distributor and reseller level

Performance can be explored down to individual products, distributors and resellers, which the previous reporting could not do.

The problem

Pricing moved faster than the data did

In a marketplace, price is the product. It is set against competitors, distributor terms and demand that all shift week to week. A pricing team working from last week’s consolidated extract is answering a question that has already changed.

Our client was expanding across five markets and a growing network of distributors, resellers and independent sellers. The pricing data existed, but it sat in separate systems with no centralised warehouse behind it, so nobody could see performance across the network consistently.

What was getting in the way

Pricing and operational data were spread across disconnected systems, which limited visibility and made consistent insight difficult.
Delayed processing slowed the response to changing market trends, distributor behaviour and reseller performance.
With no centralised data warehouse, teams struggled to access, manage and analyse pricing data efficiently.
There were no interactive dashboards or deep-dive reporting for product, distributor and reseller level performance.

Our approach

Centralise it first, then make it move

What we found

Real-time delivery is worth nothing if what arrives is inconsistent. Pricing and operational data had to be organised into objects built for pricing analysis before any pipeline was worth building, so centralising and structuring came first and the streaming came second.

What we did

Data organised around pricing questions. Fragmented datasets were consolidated and dedicated data objects created for pricing analytics, rather than reusing structures built for day-to-day operations.

A real-time pipeline on AWS. AWS Kinesis, Data Migration Service, Glue and Relational Database Service move and transform data continuously, so analysis runs on current activity rather than a scheduled extract.

A warehouse sized for pricing history. AWS Redshift stores and manages large volumes of processed pricing data, which is what makes historical comparison and forward planning possible.

Dashboards that go down to the row. Filter-enabled QuickSight dashboards with tabular reporting and deep-dive views, so product, distributor and reseller performance can be examined individually.

The choice we made

We engineered for streaming rather than scheduled batch loads. It is more to build and more to run. For a marketplace where competitors reprice during the day, a nightly refresh would have meant making today’s pricing decisions on yesterday’s market, which is the problem the client came to us with.

BUILT ON

MySQL Workbench
Amazon RDS
AWS Database Migration Service (DMS)
Amazon Kinesis
Amazon Redshift
Amazon QuickSight

The results

What changed for the business

Pricing stopped being a monthly review exercise and became a decision the team could make with current numbers.

Pricing moves with the market

Stakeholders see pricing performance as it happens, so a competitive response is a same-day decision rather than something raised at the next review.

Opportunities that aggregate reporting hid

Deep-dive views at product, distributor and reseller level surfaced pricing opportunities that summary reporting had averaged away. The client reports an effect on both sales and margin.

Partners work from the same numbers

Unified data views improved alignment between the client, global brands and distribution partners, so a pricing conversation starts from one version of performance instead of three.

Let's Connect