The task
The store had basic product cards (brand, model, price) but lacked detailed specifications — movement, case and bracelet materials, water resistance, dimensions, reference numbers, quality descriptions and images. Filling that in manually across thousands of models takes far too long, and incomplete product pages convert poorly and are less likely to rank well in search.
What we did and what data we collected
We collected watch specifications from several external sources and linked them to the store's products through product matching — primarily by reference numbers and by brand plus model. For every item we assembled a unified set of attributes and completed the card.
Challenges and how we solved them
The main challenge was matching. The same reference number is written differently across platforms, and some models come with no reference at all. We matched items by a combination of signals (brand, model, reference, key attributes) rather than a single field.
The second was contradictions between sources: one model, different values for the same attribute. We defined source priorities and value-selection rules so that only consistent data landed in the card.
The third was inconsistent units and formats (millimeters, sizes, water resistance). Everything was mapped to unified dictionaries and units of measurement.
The result
The store received an enriched catalog with complete, consistent specifications — the cards are more informative, and filling in new items now takes a fraction of the manual work.
Services in this case study: Product Data Enrichment · Product data scraping · Product Matching · Data for E-commerce
Useful reading: Data enrichment · Data normalization