Myntra · India

Review Trend Tracking on Myntra for a D2C Apparel Label: A Myntra Review Scraping Case Study

Myntra review scraping helped a D2C apparel label track customer sentiment by product and month, find fit and fabric issues and cut size-related returns.

The challenge

The client is a fast-growing D2C label that sells most of its volume through Myntra. Its team asked iWeb Data Scraping a simple question: "We need our brand's customer reviews and how they are shaping over time." Ratings had slipped slightly over two seasons, returns were rising, and the team could not tell whether the cause was fit, fabric, colour accuracy or delivery.

Reading reviews manually on the marketplace was slow, and the seller dashboard did not show trends by product, theme or month.

  • Volume. The brand had more than 1,200 live styles with thousands of reviews between them.
  • Short, mixed-language text. Many reviews were brief ("size runs small", "colour different") or written in Hinglish.
  • Linking reviews to product attributes. To act, the team needed to know which fabrics, fits and categories drew complaints.
  • Competitor context. A 4.0 rating means little without knowing how comparable brands perform in the same category.

What we built

Step 1: Product list and attributes

iWeb Data Scraping collected all live styles for the brand and five competitors, with category, fabric, fit, colour, price and size range.

Step 2: Review collection

For each style, collectors captured public star ratings, review text, review date, size purchased where shown, and helpful votes. Reviewer names were excluded.

Step 3: Theme and sentiment tagging

Each review was tagged with themes using an AI classifier trained on fashion vocabulary, including Hinglish: fit (too small, too large, length), fabric (thin, see-through, rough, shrinkage), colour (different from image, fading), stitching and finish, comfort, value for money, and delivery or packaging.

Step 4: Trend tables

Ratings and theme shares were calculated by month, category, fabric and fit, for the brand and for each competitor.

Step 5: Alerts

Any style whose rolling 30-day rating fell below 3.8, or where "size runs small" exceeded 25% of reviews, triggered an alert to the product team.

Results

Metric Outcome (illustrative)
Reviews analysed in first run 48,000
Styles flagged for fit issues 37
Change in "size runs small" share after size-chart fix 22% to 9% in women's dresses
Average rating change for fixed styles over 3 months +0.3 stars

The biggest finding was that one rayon dress block ran small across 37 styles. The team updated its size chart, added a "size up" note on product pages, and corrected the pattern for the next production run. Returns linked to fit fell noticeably in the following season. A second finding, that two colours looked brighter in photos than in reality, led to a change in product photography.

How it's delivered

The record below is illustrative.

{
  "brand": "Example Label",
  "style_id": "MY-2938471",
  "category": "Women > Dresses",
  "fabric": "Rayon",
  "fit": "A-line",
  "review_date": "2026-09-18",
  "stars": 3,
  "size_purchased": "M",
  "review_text": "Design is nice but size runs small, had to return and order L",
  "themes": ["fit_too_small"],
  "sentiment": -0.3,
  "helpful_votes": 12
}
Category Brand Rating Competitor Median Top Complaint Theme Share of Reviews with That Theme
Women's dresses 3.9 4.1 Fit too small 22%
Women's tops 4.2 4.1 Colour different 9%
Men's shirts 4.0 4.0 Fabric thin 11%
Men's joggers 4.3 4.0 Length too short 6%

How the product team uses the data each week

Every Monday, the product team receives a short report with four parts: styles that crossed an alert threshold, the top three complaint themes by category, new themes that appeared for the first time, and positive themes worth using in marketing copy (for example, "soft fabric" and "true to colour"). The design team uses the fit data when grading new patterns, and the photography team checks colour complaints before each shoot. Customer support uses the same theme tags to prepare answers to common questions on product pages.

The outcome

Reviews linked to attributes showed patterns no single review could reveal.

Hinglish-aware tagging captured complaints that keyword searches would miss.

Competitor benchmarks separated brand problems from category norms.