Powering flexible and global retail pricing

THE CHALLENGE

Overcoming technical debt at a massive scale

Decathlon is universally known as a sports retailer, but it has also grown into a major technology operator, with nearly 4,000 people working at its digital hub in Lille, France, where it develops and maintains tools that power the organization’s growth. 

One of these platforms is an internal pricing builder designed to centralize retail prices and maintain a single source of truth for global commerce. It’s a system that needs to track and account for standard prices, discounts, multiple currencies, and often strict legal regulations.

“The advantage of centralized pricing is that we can also centralize compliance with pricing laws, of which there are many, depending on the country,” explained Julien Lampin, Solutions Architect at Decathlon. “In France, for example, it is forbidden to sell at a loss except during sales. We are not pricing decision-makers, but if someone sets a price too low, we are obligated to raise it to be compliant.”

It’s an important business process, but Decathlon’s eight-year-old legacy system, Masterprice, which was built on a PostgreSQL database, was buckling under the weight of severe technical debt, burdensome data stores, and a series of critical paradigm shifts. These include local ‘subsidiarity’, which allows over 600 stores in France to adjust their own prices, as well as the launches of AI-driven dynamic pricing and a new third-party marketplace. As a result, the business’s daily data volume skyrocketed, reaching up to 1 million dynamic prices and 4 million marketplace prices. 

“Mechanically, it generated a huge load increase,” added Lampin. “In France alone, we have around 1 million references. Multiply that by 600 stores, and that potentially creates 600 million timelines.”

Masterprice’s PostgreSQL database also suffered from unpurged, compressed timeline data and heavily duplicated tables, which slowed momentum and feature delivery. Decathlon needed a new platform that could scale seamlessly to support future models without creating infrastructural bottlenecks.

“Masterprice was holding up, but we didn’t know how long for,” Lampin noted. “We had a vast amount of unpurged data, and scalability wasn't perfect. We needed a clear strategy for handling massive loads over the coming years.”

OUR SOLUTION

A flexible, NoSQL document architecture

To handle this explosive growth and modernize its infrastructure, Decathlon decided to rebuild its pricing system from scratch. The new system was Oneprice, replacing PostgreSQL with MongoDB Atlas.

“One Price is essentially the V2 rewrite of Masterprice,” said Lampin. “The initial objective was clear: centralize prices and build a single source of truth.”

MongoDB was chosen because its document-based structure natively handles the complexity of nested pricing data. Rather than forcing data into rigid relational tables that require heavy joins, Decathlon now uses MongoDB Atlas to denormalize the data, storing business models, channels, entities, taxes, and purchase prices together within unified documents.

“The advantages of MongoDB Atlas include denormalization, as all our data can be grouped under the relevant business model, channel, and entity,” explained Lampin. “We can retrieve all related data—taxes, purchase prices, and so on—at the same time to calculate timelines, so being able to store them together is a big plus for us.”

To safely manage data streams and ensure high resilience, Decathlon integrated Apache Kafka queues alongside the MongoDB Atlas database. MongoDB’s vertical scalability also proves to be highly practical for accommodating the extreme fluctuations that typify retail.

“Pricing timelines are large, and there are nested documents that we need to manipulate, making document storage totally appropriate,” said Lampin. “And MongoDB Atlas’s vertical scalability is extremely useful for both our seasonal and nightly traffic spikes.”

Decathlon also chose MongoDB Atlas Online Archive to address bloated data storage. Whenever a price timeline is updated, older timelines are deactivated and automatically moved from the active cluster. This ensures that older data no longer weighs down on indices or processing costs, while remaining accessible for historical queries.

“We generate bulky timelines that overwrite old ones, but occasionally we still need the old data,” said Lampin. “MongoDB Atlas Online Archive means we don’t clog up the database, as it automatically moves deactivated timelines that are older than a certain date into cold storage.”

OUTCOME

Future-ready agility and seamless migrations

The implementation of MongoDB Atlas enabled Decathlon to orchestrate a seamless, risk-free migration. By targeting the marketplace data first, it ran both the legacy PostgreSQL system and the new MongoDB Atlas-backed system concurrently to test for accuracy without impacting end users.

“We chose the marketplace as our initial migration path because it generates the highest volume,” Lampin explained. “It allowed us to test the load with relatively few complex legal rules. We could lay out the full infrastructure—monitoring, databases, queues, APIs, and gateways—without overcomplicating initial permissions.”

Oneprice currently handles over 20 Kafka consumers and around 60 API consumers. None experienced any disruption during the migration; API users continued to call the same URLs, and Kafka consumers continued reading from the same topics.

To date, Decathlon has successfully migrated 100% of its marketplace partners and 4 million marketplace processes. It has also balanced infrastructure costs across its systems, and by abandoning rigid relational design, developers can now focus purely on solving business problems rather than wrestling with a data model. 

“The tooling is really cool,” said Lampin. “Manipulating objects and documents is very natural and flexible for developers, and tools like MongoDB Compass, Atlas UI, and Charts are fantastic.”

Importantly, Decathlon has also wiped out its technical debt. And with Oneprice, it has a system ready to handle millions of data points and their related legal complexities each day across Decathlon’s 85 countries of operation.

MongoDB Atlas also ensures that Decathlon is fully prepared for future integrations, such as B2B business models and advanced AI transparency tools.

“By designing on MongoDB Atlas, we can think more clearly about our data and how we use it,” Lampin concluded. “Ultimately, we're here to solve business problems, and tools like MongoDB play a key role in informing our choices.”