A PIVOTREE CASE STUDY
3 years of work done in 1 day with Sonepar
Leading global distributor Sonepar Group is an independent family- owned company with market leadership in the business-to-business (B2B) distribution of electrical, industrial, safety products, and related solutions. As the digital revolution advances and product technologies continue to evolve, there’s a widening range of applications for electrical equipment and a growing need for new services.
Sonepar Group is established in 40 countries, including the United States and Canada, and has embarked on an ambitious transformation agenda to become the first global B2B electrical distributor to provide a fully digitalized and synchronized omnichannel experience to all customers.
Drawing on the skill and passion of its 45,000 global associates, Sonepar had sales of €33.3 billion in 2023, €11b of which was from online channels. Their vision is clear: "Sonepar makes our customers’ lives easier, over the counter, visiting customers, by phone or online – however we are needed.”
Sonepar Canada represents 9 electrical and industrial distribution brands with 1800+ employees across 90 locations in 8 provinces. Pivotree worked with Sonepar Canada to automate their data enrichment processes to streamline manual effort, improve time-to-market, and grow revenues for this growing market leader.
Achieving scale at this level has allowed us to bring our products to market much faster while making customer experience seamless.
Data enrichment was the number one problem
Sonepar Group set out to digitize the customer experience. In Canada, that ambition ran into a catalog growing by about 50,000 SKUs a year.
The growth came from acquisitions. Every company that joined brought its own vendors, its own attribute names, and its own definition of a complete product record.
Sonepar Canada worked with Pivotree to automate parts of the enrichment pipeline, starting with governance: a common definition of a product, held across the whole organization, before anything got automated on top of it. That was the dependency underneath everything else the business wanted from the project — revenue targets, speed to market, and an operation not spending its week reconciling spreadsheets.
A rising volume of data from an increasing number of sources can lead to data inconsistencies and ultimately a poor customer experience, meaning that instituting data governance is key.
Expertise + AI/ML Innovation: A Powerful Combination
Sonepar Canada needed a partner who had already built PIM and MDM systems for global distributors at this size, rather than one learning the category on the job. Pivotree had done that work before, which is why we got the call.
Volume was the other half of the problem, and manual effort alone can't scale to 50,000 SKUs a year. Pivotree DIVE, an AI/ML enrichment platform, connected directly to the MDM application Sonepar Canada already runs, so the automation went on top of the existing system instead of replacing it. The combination was built to:
Classify their data automatically across two systems
Report on the model’s performance
Automatically normalize and standardize their product data
Fine-tune the data models based on human-in-the-loop feedback
The 70,000 SKU Pilot
Pivotree and Sonepar Canada ran the implementation in phases, and phase one was a test with a real exit. Both teams chose a set of products, loaded it into DIVE, and compared DIVE's predictions against the models Sonepar Canada was already using. If the predictions weren't better, the project would stop there.
DIVE ingested about 70,000 products and mapped them into a usable taxonomy in roughly three hours, at 90% accuracy. Sonepar Canada's manual process was estimated at 32 weeks and 17,280 hours of human effort for the same catalog.
Phase 2, the full DIVE implementation, started from there.
In 24 hours, Sonepar Canada did what would previously have taken almost 3 years and 88,020 hours of human effort to accomplish.
163 weeks of work, done in a day
Phase 2 was the whole catalog. Roughly 350,000 products were uploaded, classified, and written into Sonepar Canada's MDM platform in 24 hours, at 95% accuracy once Pivotree had tuned the model. The same work done using the legacy process was estimated at 163 weeks and 88,020 hours, which is three years of calendar time with 13 full-time people on the job.
As organizations mature, they should focus not just on categorizing data, but also on being able to leverage it for greater purposes like predicting and building out what customers want.