# Functionality Overview

Conceptually, Cartographer SaaS application is separated into two primary components:

  • Application runtime stack, which provides integration JavaScript code, processes user input, serves recommendations data, and captures analytics events;
  • Administration component, responsible for data import, project authoring and publishing.

At the high level, Cartographer enables developers and merchandisers to create recommendation experiences by:

  • Data Import: Cartographer imports a product catalog data from sandbox/staging (development stage) or staging/production (production deployment) site using Open Commerce API;
  • Project Configuration: Customers create the recommendation flow by configuring questions, answers, and product mapping in Cartographer;
  • Storefront Integration: Salesforce Commerce Cloud implementation renders the experience using pipelines, content slots, or content assets;
  • Analytics Tracking: Salesforce Commerce Cloud implementation connects add-to-cart and checkout events with Cartographer conversion tracking to identify impact at the individual question and answer level and fine tune the experience.

# Question Types

# Simple Choice

Map a single product attribute to a set of predefined answers. Questions can be rendered in the browser experience as distinct push buttons, sliders, dropdowns, or multiple-choice checkboxes.

# Combination (Hybrid)

Similar to Simple Choice, Combination questions allow you to map multiple products attributes to a single question, exponentially increasing recommendations flexibility.

# Fuzzy Choice

Simple and Combination choice questions typically produce exact recommendations, determined by whether customer selections match question answers. Fuzzy question allows to make Cartographer algorithm less restrictive, and produce potentially better recommendations. For example, you may configure a question asking for customers preferred colors, with "Black," "Black/Red," "Red," "Red/Blue," "Blue" answer options. If the customer selects "Black/Red" answer, but there are products that may be a better fit (as determined by other questions in the project) with "Black" or "Red" colors, they may be recommended as well, albeit with a lower score.


Cartographer provides two similar, yet distinct fuzzy recommendation concepts. Fuzzy matching that can be enabled for Simple and Combination questions works only with numeric product attributes and looks at each answer mapping individually. For example, you could have a Comfort level product attribute, rated 1 to 10. A question answer may require products with a value of 10, but if fuzzy matching is configured, products with similar attribute values would also be considered, at a lower recommendation score. For example, depending on settings, a product with Comfort level 6 could yield a 60% match recommendation. Fuzzy questions allow you to establish a similarity scale of arbitrary attribute values, not necessarily just numbers, for example, colors, names, etc.

# Slider

Allow customers to select the desired value on a numeric scale, rather than a preconfigured answer. Questions of this type would typically map to tangible numeric or physical product attributes, such as price, weight, length, etc.

# Free Text

Free text questions allow collecting unstructured input from customers, such as names. Values entered in recommender sessions can be retrieved later using Cartographer reports. Free text questions do not affect recommendations engine

# Analytics Tracking

Cartographer provides some analytics metrics, including a number of overall loads and impressions, as well as segmentation by full or partial conversions. A successful conversion is determined by the following sequence of events:

  • Customer visits a page with a recommender instance;
  • Recommender is loaded successfully, and customer engages with recommender by selection one or more answers;
  • Product recommendations are generated and displayed within the viewport. Recommendations not shown on the visible area of the screen are not counted as successful impressions;
  • Customer adds a product to the cart within the predefined time interval since the product impression. Products may be added either from recommender experience or another area of the site, for example Product Detail Page;
  • Customer checks out within the predefined time interval after adding a product to the cart.

Cartographer internal analytics tools require integration with two storefront events to provide accurate events reporting:

  • Add to Cart: submits products added to cart and links them to the currently active recommendation session;
  • Checkout Complete: submits products checked out by the storefront customer and links them to the recommendation session.

Event tracking may be implemented in many ways, for example, by direct modification of Salesforce Commerce Cloud storefront code and templates. Some integration options do not require page code modifications, for example, when using a tag manager such as Google Tag Manager (https://tagmanager.google.com) or Tealium. It is also possible to integrate recommendation events collection into third-party analytics tools such as Google Analytics, Omniture, etc.

Refer to the developer documentation for specific integration requirements and code samples.

# Maintenance

Once the initial integration is in place, Cartographer recommender instance can be updated in two ways:

  • Data and flow changes, such as adding or removing questions, answers; additional product launches do not require Salesforce Commerce Cloud updates and can be handled completely in Cartographer backend;
  • Presentation updates can be achieved in most cases by altering content asset HTML and CSS without deploying Salesforce Commerce Cloud code changes