1.1 Product Feed Introduction

General Introduction

As a global sizing leader, the world's best apparel and footwear companies trust us to solve sizing and help them sell smarter. Our Fit Services have been proven to boost conversion rates and slash returns by leveraging the data that we collect about products and shopper behavior.

The importance of high data quality in maximizing the impact cannot be overstated. In particular, precise and accurate data in your product feed allows us to

  • enable our services for more products, thus increasing shoppers' exposure to our services,
  • correctly tailor the user experience,
  • improve the accuracy of our recommendations, and
  • integrate smoothly, allowing Fit Services to go live faster.

Purpose

This product feed specification provides the guidelines and rules to which the feed should conform. It is a guide to help you understand the mandatory data requirements for your product feed and it defines in detail what the format of the data should be with examples and best practices. This document is meant to be the data specifications support material from the sales deck of the Fit Finder Standard service. The product feed is the most critical part of the Fit Finder integration and requires high quality information and data structure. A high impact product feed allows Fit Analytics to import garment information to the Fit Finder database.

Good data quality ensures a smooth Fit Finder integration and leads to strong size recommendations, ensuring the best tailored experience to users.

1.2 Product Feed Requirements

Feed Format

Your product feed should be shared via an HTTP(S) or (S)FTP URL hosted by you pointing to a CSV or XML (RSS/Atom) file using UTF-8 encoding. There must be a separate feed entry for each size variant, and the attribute names must match the names in the "Attribute" column of the tables below.

If the previously mentioned methods are not feasible for you, we can also accommodate data feeds hosted on your Azure Blob Storage, AWS S3, ​​or Google Cloud Storage bucket.

If you already have a Facebook or Google product feed, you’ll likely only need a few adjustments, the main one being the addition of an item_subgroup_id attribute.

If you have multiple online stores or versions of your store with different sizing, product availability or languages, we require a separate product feed for each.

  • Example 1: If you have French and Dutch language versions of your Belgian store, we require separate French and Dutch feeds.
  • Example 2: If your European stores all share the same sizing, product availability and language (e.g. English), then it’s fine to provide a single European feed.


Feed Mandatory Data

Fit Services like Fit Finder can only be enabled if all of the following attributes are provided for the product in question. If you can’t provide the values as requested, please contact us so we can suggest how you could solve this.


Mandatory Data Table

+
Attribute Short description Examples
id An ID that uniquely identifies a size variant of a garment.
        12345-250-11
        12345-BLACK-S
      
item_subgroup_id An ID that uniquely identifies a color variant of a garment, grouping together all sizes for a specific combination of color, material, pattern, etc. For example, all sizes of a given red dress should have the same item subgroup ID, but the same dress in blue (or in a different material or pattern) should have a different item subgroup ID.
        1234567
        12345-250
        12345-BLACK
      
item_group_id An ID that groups together all variants of a garment. Variants are a group of similar products that only differ from one another by product details like the size, color, material or pattern.
        12345
      
brand The product’s brand name, manufacturer or designer, which allows us to customize recommendations by brand.
        Esprit
        Nike
        Adidas
        Puma
      
gender The Fit Finder flow differs between men’s and women’s products, e.g. recommendations for women require bra information. The [gender] also allows us to tailor size recommendations, e.g. men’s sizes typically fit differently than women’s sizes. Products should be classifed as unisex if that product can be recommended for women and men.
        male
        female
        unisex
      
age_group The demographic that your product is designed for. Correctly distinguishing between adults’ and kids’ products is important. A kids’ product that is mistakenly marked as for adults might lead to inaccurate size recommendations for the product in question and might pollute our machine learning data, skewing recommendations for other products.
        kids
        adult
      
size The localized size string that is displayed to shoppers on the product detail page. We recommend adding the size system in the size strings on the PDP and in the product feed, e.g. UK 10 rather than just 10, to reduce confusion for your shoppers.
        32
        UK 6.5
        Small
        XXXL
      
size_system The country’s sizing system that your product uses.
        UK
        US
        EU
      
size_type The body type for which the garment is designed.
        regular
        petite
        plus
        tall
        big
        maternity
      
link A URL pointing to the product detail page on the shop website.
        http://au.exampleshop.com/TZZ12345-105-51.html
      
image_link A URL pointing to the main product image.
        http://au.exampleshop.com/TZZ12345-105-51.jpg
      
title A brief description of the product.
        Adidas trainer
        Long sleeve blouse
      
fb_product_category or
google_product_category

The product category according to the Facebook product taxonomy (preferred) or the Google product taxonomy. At least one of the two must be provided, since the product category affects the Fit Finder user experience and size recommendations. For example, shoppers are asked to input bra information for tops, but not for bottoms, and a size M for a dress may fit differently than a size M for shorts.

        2271
        Apparel & Accessories > Clothing > Dresses
        clothing & accessories > clothing > women's clothing > dresses
      
product_type The product type according to your own product categorization system.
Home > Women > Dresses > Maxi Dresses
availability Whether a product is in or out of stock.
        in_stock
        out_of_stock
      


Feed Optional Data

The following attributes help with detecting feed data quality issues that could affect size and style recommendations, and tuning for which products Fit Services are enabled.


Optional Data Table

+
Attribute Short description Examples
gtin The Global Trade Item Number (GTIN), also known as the EAN or UPC, that uniquely identifies a size variant of a garment in a globally standard way.
3234567890126
color The product’s color.
Green
material The main fabric or material that your product is made of.
Cotton
pattern The pattern or graphic print on your product.
striped

1.3 Product Feed Example

Find a product feed example here for download.

1.4 Feed Attributes Details

Color variant ID [item_subgroup_id]

Use the color variant ID [item_subgroup_id] attribute to group together all sizes for a specific combination of color, material, pattern, etc. For example, all sizes of a given red T-shirt should have the same item subgroup ID, but the same T-shirt in blue (or in a different material or pattern) should have a different item subgroup ID.

  • Required for each product
  • Format: String (Unicode characters. Recommended: ASCII only); 1–70 characters
  • Examples: 1234567, 12345-250, 12345-BLACK
  • Matching item subgroup IDs must be available on product detail pages (PDPs) and order confirmation pages (OCPs) on your website and mobile apps.
  • The item subgroup ID must not be reused later for a different garment. It must be unique in the history of your shop.
  • The item subgroup ID must be locale-independent. For example, if 12345-BLACK appears in the product feed for your US store, then it must also appear as 12345-BLACK (not e.g. 12345-NOIR) in your French store.
  • Avoid changing item subgroup IDs. Doing so may lead to an interruption of Fit Services if for example the IDs on PDPs are updated before a product feed with the new IDs has been ingested into the Fit Analytics database. It may also lead to a gap in training data for our size recommendation models if we aren’t able to link Fit Finder inputs, purchases and returns due to mismatched IDs.

2.1 Returns Feed Introduction

To benefit from the complete Fit Finder experience and the machine learning process of the recommendation engine, Fit Analytics requires a returns feed. That feed allows us to import returns data to our database and matches the return to the purchases for your shop. When sharing the file, it’s important to pay attention not only to the attributes and format but also to the data quality:

  • Good data quality leads to a quicker, smoother import of data into the database.
  • Good data quality leads to better size recommendations, ensuring a good user experience.

Fit Analytics requires an HTTP(S) or (S)FTP URL pointing to a CSV feed using UTF-8 encoding. If your shop is live in multiple countries, we require returns for all locales where Fit Finder is live. Important details regarding the format, frequency and availability of the feed:

  • Returns should ideally be exported on a daily basis. Weekly or monthly exports are also supported, but there will be more of a delay until the returns are reflected in reports and in size recommendations.
  • The filename must include the export date in the format YYYYMMDD.
  • The files should only include new returns that came in since the previous export.
  • Files should remain available for some time (ideally at least a month), so that Fit Analytics can re-ingest them in case of an error.

2.2 Returns Feed Details

Returns Feed Details Table

+
Attribute Short description Examples
return_id An ID that uniquely identifies the return event.
82934601982
timestamp The date and time of the return event.
2024-01-01T10:00:00Z
order_id An ID that identifies the original purchase order that included the item now being returned. The order ID in the return feed must match the order ID reported from the order confirmation page when the item was originally bought.
1234567
item_id An ID that uniquely identifies a size variant of a garment.
12345-250-11
quantity The quantity of items being returned.
2
return_reason The reason why an item was returned. Used for improving the recommendation system and for reporting. Only required if a shop collects return reasons and if the purchase was not canceled.
        big
        small
        fit
        style
        other
      
is_canceled Whether or not the purchase was canceled, as opposed to having been received and then returned. Only required if the user has the option to cancel a purchase on your website.
        true
        false
      

2.3 Returns Feed Example

Find a return feed example here for download.

Questions?

We're happy to help. Send your question to support@fitanalytics.com.