In production Industrial distribution and logistics

Intelligent Product Categorization

Grupo Cartes · 4 weeks

3M+

Rows updated

508

Tables processed

327

Brands normalized

27 min

Execution time

NLPFuzzy matchingPythonSQL Server

Context

Grupo Cartes manages a catalog of over 560 product tables from hundreds of manufacturers (MANN, DONALDSON, HENGST, FLEETGUARD, etc.). Over the years, acquisitions and brand changes created a critical problem: the same product can have multiple references.

The challenge

  • Brand name variations (“MANN”, “MANN-FILTER”, “MANN+HUMMEL”) made automatic linking impossible
  • 600K+ uncategorized references
  • Historical catalogs disconnected from current inventory

The solution

Intelligent brand normalization

Over 100 known mappings with fuzzy matching for new variations. AI identifies that “MANN-FILTER” and “MANN+HUMMEL” are the same brand.

Massive categorization

Update of the CATEGORY field in 508 tables, applying unified identifier to over 3 million rows.

Measurable results

MetricValue
Tables processed508
Rows updated3,032,848
Unique categories75,044
Brands normalized327
Execution time~27 minutes

Impact

  • Automated cross-selling: instant identification of equivalents
  • Alternative suggestions when a product is unavailable
  • Historical catalogs linked with current inventory
  • Foundation for future automatic recommendations

Business impact

AspectBeforeAfter
Uncategorized references600K+0
Categorization timeWeeks (manual)27 minutes
Cross-sellingImpossible without clean catalogAutomated
Alternative suggestionsManual, experience-basedInstant, data-driven

Years of acquisitions created catalog chaos. The same product had multiple references across 508 tables. Now unified, enabling automated cross-selling and alternative suggestions.

Status: IN PRODUCTION

Ready to optimize your operations?

Free 30-minute diagnosis. No commitment, no smoke.