How a Dutch E-commerce Brand Extracted 100K+ EAN-Based Reviews from Bol.com for Product Sentiment

Introduction

Tagline: A Netherlands-based e-commerce brand extracted product-level reviews at EAN-level granularity from Bol.com — building a product sentiment intelligence layer their competitors couldn't access.

The Challenge

The client operated as both a seller on Bol.com (the dominant Dutch e-commerce marketplace) and a multi-channel consumer brand selling through their own DTC and other retailers. Customer reviews — particularly product-level reviews on Bol.com — were a critical input for:

• Product quality intelligence — what specific issues customers were reporting across the product catalog
• Competitive sentiment — how competing products in their categories were performing on reviews
• Category-level trends — what attributes customers cared about most (and least) across product categories
• Strategic merchandising — which products were gaining or losing review velocity

Bol.com has substantial review depth across millions of products, but accessing structured review data at scale required custom extraction. Standard sentiment monitoring tools either didn't cover Bol.com or covered it too sparsely to support product-level analysis.

The Approach

Actowiz Solutions built an EAN-based review extraction pipeline for Bol.com:

• EAN-driven product identification — used EAN (European Article Number) as the unique product identifier, allowing precise matching across the client's catalog
• Initial review backfill — captured the first 100,000+ reviews for the client's priority products and competitive set
• Review attribute extraction — review text, rating, date, verified purchase flag, helpful counts, language
• Sentiment analysis integration — output structured to feed the client's sentiment analysis tools cleanly
• Ongoing refresh — daily updates capturing new reviews as they appeared

The Solution Architecture

Bol.com's review pagination and structure required careful handling — reviews are paginated, sometimes incomplete in the default view, and have language variants. The extraction pipeline handled the full review depth per product, with quality validation against rating distributions and review volume sanity checks.

Output included raw review data plus a structured sentiment dashboard with product-level and category-level views.

Results

• 100,000+ reviews extracted across the client's products and competitive set
• Product-level sentiment scoring for the full priority catalog, with category-level rollups
• Identified 8 specific products with declining sentiment trends — informing operations and merchandising responses
• Competitive sentiment intelligence revealed where competing products had quality reputation gaps the client could exploit in marketing
• Daily refresh kept sentiment intelligence current as new reviews arrived
• Reusable infrastructure — same pipeline now extended to track reviews across Amazon NL, Coolblue, and other Dutch e-commerce platforms

Why This Matters For You

If you sell products in the Netherlands or broader EU markets, Bol.com's reviews are foundational customer voice data — but accessing it at scale requires custom extraction. The same applies to Coolblue, Amazon NL/DE/FR/UK, Otto.de, and dozens of regional European e-commerce platforms. Building a sentiment intelligence layer with proper review data — at the product or EAN level — turns customer feedback from a reactive customer service signal into a proactive product strategy input.

The same pattern works globally: Amazon (all geographies), AliExpress, Tmall, Mercado Libre, Lazada, and any major review-rich e-commerce platform.