This page explains how a Mapp Fashion integration fits together: which data you provide, what Mapp Fashion does with it, and how recommendations reach your website and your emails.
For a product introduction, see What Is Mapp Fashion? For business scenarios, see the Mapp Fashion use cases.
How Mapp Fashion Works
Mapp Fashion combines three types of data:
Product data describes your catalog. Mapp Fashion enriches it with fashion-specific attributes.
Transaction data covers orders and returns. It is used for model training, analysis and recommendations.
Interaction data comes from tracking on your website. It shows what each shopper views and buys and is used to personalize recommendations.
The data flows in four steps:
You provide product data and transaction data.
Mapp Fashion enriches each product. Stylists, supported by machine-learning models, label physical and contextual attributes. Derived attributes are calculated from these by rules.
Tracking on your website records how shoppers interact with products and recommendations.
Mapp Fashion combines the enriched attributes, purchase history and shopper behavior to generate recommendations. They reach your website through hosted widgets, your own frontend or the API, and your campaigns through emails.
For details on the attribute model, see How Enriched Product Attributes Improve Product Discovery.
Note
The quality of the recommendations depends on complete, consistent and up-to-date data. Product and customer identifiers must be identical across product data, transaction data and tracking. For how identifiers map across product data, tracking and the API, see Product and Customer Identifiers.
Recommendation Types
Each recommendation type starts from a different context:
Type | Starting point | Typical placement |
|---|---|---|
Outfits | One product (anchor item) | Product detail page |
Partner outfits | One product, combined into looks you supply yourself | Product detail page |
Similar items | One product | Product detail page |
Top items | Shopper profile, no product needed | Homepage, landing pages |
Basket items | Products in the basket | Basket, wishlist, checkout |
Faceted recommendations | Filters such as category, brand or price | Product listing and discovery pages |
Themed recommendations | A theme agreed with Mapp | Campaign and merchandising placements |
How Outfits Work
Each outfit is built around one anchor item, usually the product a shopper is viewing. Mapp Fashion combines it with complementary garments, shoes and accessories. Outfits are grouped by occasion, such as casual, evening or business.
The styling logic is defined together with Mapp's stylists for each brand. It decides which product types are paired and which combinations are valid.
Partner outfits are looks you supply yourself, also known as Shop the Look or Model Wears. Products that share a Set ID form a Fashion Set and are prioritized as pairings. For details, see How Personalized Outfit Recommendations Help Shoppers Discover Complete Looks.
Integration Building Blocks
A typical integration consists of the following building blocks. Which ones you need depends on your use cases. The matrix in Prerequisites for Mapp Fashion Onboarding shows the requirements per use case.
Onboarding and Domain Setup
Mapp Fashion runs on a dedicated host, usually a subdomain of your website that you delegate to Mapp. Mapp provides the host and your namespace ID during onboarding. For details, see Getting Started.
Product Data
New implementations manage product data in the Product Catalog in Connect, using a catalog in Fashion-enabled mode. Once the Product Catalog is available for your account, it is the required source for product data.
Implementations that started before the Product Catalog deliver product data through the Product Feed.
Transaction Data
Orders and returns are delivered through the Transaction Feed. This applies to new and existing implementations.
Tracking
Tracking runs on Mapp Intelligence Smart Pixel. The Mapp Fashion extension adds fashion-specific events and the retrieval of recommendation content, so no separate tracking script is needed. For the implementation, see Tracking Setup.
Displaying Recommendations
There are three ways to show recommendations on your website:
Option | Use it when | Documentation |
|---|---|---|
Hosted widgets | You want ready-made components. Mapp Fashion renders the widget and records interactions automatically. | |
Smart Pixel retrieval | You render recommendations in your own frontend. The Smart Pixel requests the content. | |
Server-side API | You render recommendations on the server, in a headless setup or in a backend integration. |
Note
If you render recommendations yourself, set up Recommendation Tracking so that clicks and other interactions are recorded.
Emails
Themed, BAU and outfit emails embed Mapp Fashion recommendations as images in your email templates. For details, see Emails.
Measuring Performance
To compare recommendation performance against a control group or a competitor, see A/B Testing for Recommendation Performance. Metrics such as attributed revenue are defined in Metric and Dimension Definitions.
Existing Standalone Implementations
Implementations that use the separate dressipi.js script are documented in Standalone Integration (Legacy). To move to the Smart Pixel, see Migrate from Standalone Tracking to Smart Pixel.