Search relevance and product data
Native speakers rate how well your search results match real queries in each market, and write, translate and fill in the product data your catalog needs: titles, attributes and categories.
zapatillas de running para mujer
Zapatillas de running para mujer, amortiguación
ExactZapatillas de running para hombre
Irrelevant: wrong genderMallas de running para mujer
Related: not shoesZapatillas de trail para mujer, impermeables
Related: trail, not road
Example tasks for a Spanish storefront.
- Search work
- Result ratings, query intent labels and side-by-side ranker comparisons
- Product work
- Attributes, titles, descriptions and category mapping
- Markets
- Raters and writers from each country you sell in
- Output
- CSV, JSON or your catalog's import template
Relevance and product data
Search relevance rating means people judging how well each result answers a query, so you can measure your search engine and train its ranking. Product data enrichment means filling in and correcting the titles, attributes and categories behind each listing, so the right products show up for those queries.
Both depend on knowing the market. In Mexico, people search for running shoes as "tenis", not "zapatillas". A rater from Spain would mark the right results wrong.
Search and catalog work
For marketplaces, retailers and search teams launching in new markets.
Search result rating
Each result rated against the query on your scale, by raters who search the way people in that market do.
Query intent labels
Queries sorted by what the person wants: a product, a brand, a category, help or a store.
Side-by-side comparisons
Two rankings for the same query compared, to see whether a new ranker beats the current one in each market.
Attribute extraction
Size, color, material, fit and other attributes pulled from titles, descriptions and images into structured fields.
Titles and descriptions
Product copy written or translated for each market, following the marketplace's title rules and your style guide.
Category mapping
Products placed in the right node of your taxonomy, or mapped to a marketplace's own category tree.
Where local knowledge changes the answer
Each of these looks small. Across a catalog, they decide what shoppers find.
- Words"Tenis" in Mexico, "zapatillas" in Spain. "Trainers" in the UK, "sneakers" in the US.
- SizesEU, UK and US shoe and clothing sizes converted and labeled correctly.
- UnitsCentimeters and inches, liters and fluid ounces, set to the market.
- Colors and materialsNames shoppers actually search for, not literal translations.
- Seasons and eventsBack to school, Diwali, Ramadan or Black Friday, each with its own timing and products.
How a relevance project runs
Raters calibrate on your guidelines before production, and agreement is tracked throughout.
Guidelines
Your rating scale or attribute schema, with examples. We add market-specific cases.
Calibration
Raters work the same sample, and we compare their decisions with each other's and yours.
Agreement review
We fix the guideline wording where raters split, then lock it with you.
Your sign-off QAProduction
A share of items is rated twice, and a senior reviewer checks a sample of each batch.
Delivery
Ratings or enriched data in your format, with agreement figures and open questions.
Formats
- Ratings
- CSV or JSON with query, result, rating and comment, or in your rating tool
- Catalog data
- CSV, XLSX, JSON or your marketplace's import template
- Taxonomies
- Your own, or a marketplace's category tree
- Markets
- UK, US, EU countries, Latin America, the Middle East, India and East Asia, confirmed per project
Scope of search and catalog work to confirm with client
Related services
Questions about search and product data
Can you rate search results in our markets?
Yes. Raters from each market judge results against real queries on your scale, so ratings reflect the words, sizes and expectations of local shoppers.
Can you write and translate product attributes?
Yes. We extract attributes from titles, descriptions and images, fill in missing fields, and write or translate titles and descriptions for each market.
Which rating scale do you use?
Yours, if you have one. If not, we suggest a three or four-point scale with written definitions and examples for each level, and agree it with you during calibration.
Can you map products to a marketplace's categories?
Yes. We place each product in the right category in your taxonomy or in a marketplace's category tree, and flag products that don't fit any node cleanly.
How do you keep raters consistent?
Calibration on your guidelines before production, a share of items rated twice, and a senior reviewer checking a sample of each batch. You get the agreement figures with every delivery.
Start with a pilot batch
Send your guidelines and a sample of queries or products. We rate or enrich a pilot set and quote for the rest.
- Phone
- +44 20 3290 0116
- Phone hours
- Weekdays 9:30 to 16:00, weekends 9:30 to 12:00, UK time