Glossary
A/B Test
A/B test is used to measure the impact of changing a single element on a site page. For example, testing the incremental benefit of having a green add-to-cart button rather than an existing orange button.
Advanced Merchandising
Advanced Merchandising enables you to manually manipulate the recommendations that the Algonomy engine returns at runtime when the site has very specific use cases, such as desiring to deplete inventory of specific products or boosting a new product line.
Attributable/Rec Sale
The sale of an item is attributable if it is preceded by a click on the item in a Algonomy recommendation. Typically, attributables are counted within a one-day lookback window-from midnight to 11:59:59 on the same calendar day.
Average Order Value (AOV)
Average Order Value (AOV) is calculated by dividing sales revenue by number of orders.
Blacklisting
For information about blacklisting, refer Recommendation Restrictions.
Boosting
Boosting is a merchandising rule use case that increases the frequency by which a product, category, or brand is recommended in context-relevant situations.
Category
A category is a hierarchical grouping of either similar products or sub-categories which contain similar products.
Conditional Search Landing Placement
This is a placement that is displayed on a landing page when a shopper enters the site via the results from a search engine query. This is typically utilized on Item pages and contributes to bounce reduction.
Conditional Probability (CP)
Conditional Probability (CP) refers to strategies that rank and recommend products based on the probability that the customer buys a product in a specific context (for example, viewed a seed product). CP strategies impose a minimum probability of 0.0003 for product qualification.
Conversion Rate
The conversion rate is the percentage of customers who purchase something on the site.
Dashboard/Portal
Dashboard is the Algonomy control panel in which products and merchandising rules are configured. The dashboard also includes Sales and Site Analytics performance reports.
Display Mode
Display Mode is the period when recommendations on merchant sites are visible to customers and ongoing optimization occurs to ensure the best performance strategies get the most impressions.
Dynamic Pricing
Dynamic Pricing enables for the price of items in recommendations to be determined in real time, ensuring that the price of a product on the item page matches the value the customer saw in the recommendation.
Feed
A feed is a structured file in the XML format, that contains data of the merchant's product catalog. Feeds are the mechanism by which the Omnichannel Personalization gains information on merchant inventories, for example, products, pricing, availability, and product attributes.
Force Display
Force Display is a tactic for viewing recommendations using an URL parameter while a site is in Listening mode.
To force the display of recommendations, append "?r3_forceDisplay=true" to a merchant URL on which placements are configured. If existing parameters are attached to the URL, use "&" instead of "?".
Integration Environment
The integration environment is a merchant staging environment, located at integration.richrelevance.com, which enables testing different recommendation zones, enabled strategies and messaging, merchandising rules, and site configurations before publishing settings to production.
JavaScript Injection
JavaScript injection allows Algonomy to dynamically inject new HTML or JavaScript into a merchant website.
JavaScript Instrumentation
JavaScript Instrumentation refers to the JavaScript that lives in the HTML of a retailer's website. The merchant populates related fields in the corresponding JavaScript objects, transmits the fields to the Algonomy server using HTTP or HTTPS, and receives a dynamically-generated JavaScript file that modifies the retailer site to display relevant product recommendations.
King of the Hill
King of the Hill refers to the ongoing optimization process by which all the strategies enabled for a page type on a merchant site receive exposure, and the resulting top performers are awarded the greatest display frequency. To avoid overexposing customers to suboptimal strategies for testing purposes, Algonomy closely manages the traffic being driven to these strategies.
Layout
Layout is a collection of HTML or JSON elements which acts as a template, governing how recommendations are returned from Omnichannel Personalization by replacing placeholder tags within the layout code with information from the stored product catalog. Layouts are attached to placements.
Lift
Lift is the incremental value created by Omnichannel Personalization. Lift can be ascertained through an MVT or A/B test.
Listen Mode
Listen Mode is a two-week period of required analysis during which Omnichannel Personalization collects transaction and customer behavioral data through deployed JavaScript on the merchant site. This real-time data collection enables to understand customers’ aggregate behaviors on which to base recommendation models. Recommendations are not displayed during Listening, though they can be conditionally viewed using the force display parameter.
Manual Recommendations
Manual Recommendations are merchandising rules that force specific, merchant-defined cross-sell opportunities during the merchandising of products or categories. Manual Recommendations should be used only for very specific use cases, and they override other recommendation sets from the Algonomy engine.
Model
Model refers to the back-end calculation that determines the products to be displayed for each recommendation strategy. Models are typically built at least once per day per site.
Multivariate Test (MVT)
Multivariate Test (MVT) is a test that provides the ability to measure the performance of various treatments of Algonomy recommendations.
Overlap
Overlap refers to the sales from recommendations that would have occurred regardless of Algonomy recommendations. Algonomy recognizes that in some cases, our recommendations merely provide a navigation path to a product that the customer would have eventually found anyway. Therefore, Algonomy uses overlap to exclude these sales and calculate the direct impact of recommendations on sales.
Page Area
Page Area is a named area on a page which when combined with a page type constitutes a placement.
Page Types
The various page types on a merchant site have different designs, business objectives, and customer use cases that prompt very specific placement and strategy configurations. While some merchants choose to insert recommendations on more pages than others, typical Omnichannel Personalization implementations use real estate on several standard page types.
Placements
Placement is associated with an individual location where recommendations are displayed on the site. It is defined as the combination of a page type and a page area. Attached to the placement is one or more Layouts, which define how recommendations are returned to the browser. Individual placements may be configured to prefer specific strategies.
Recommendation Restrictions
Recommendation Restrictions refer to a merchandising rule use case that forces recommendation strategies to not recommend or exclusively recommend certain products.
RecPipe
Recommendations pipeline; decides amongst a group of strategies which one to use.
Revenue per Session (RPS)
Revenue per Session (RPS) refers to sales/visit. If you take your site sales and divide that by your session traffic, that will be your baseline or benchmark to start from. When Algonomy is live, we'll look to that metric to inform our performance and use AOV and Conversion Rate as levers to increase RPS.
For example, AOV can be sacrificed for the sake of conversion, so using that as the primary metric can be a false indicator of performance. The same goes for conversion rate. If you increased prices on the site, Conversion Rate may suffer but AOV would be up. Algonomy uses RPS because it is a normalized way to measure sales without being incorrectly informed by conversion or AOV alone. Here's an example of how the numbers can look when using AOV, Conversion, and RPS:
|
Sessions |
Conversion |
AOV |
RPS |
Revenue |
|---|---|---|---|---|
|
1000 |
5.10% |
$49.00 |
$2.50 |
$2,499 |
|
1000 |
4.90% |
$50.00 |
$2.45 |
$2,450 -- Higher AOV, but less revenue |
Algonomy Anywhere API
Algonomy Anywhere API is a set of HTTP links and query string parameters that can be used to fetch the most current recommendations based on the most current customer behavior.
Session ID
Session ID is the standard web-server generated identifier for the current browser session. Algonomy looks to the retailer to set their standards of session time-outs so that it matches their internal systems.
For server-side API calls from a merchant's e-commerce platform, the user ID and session ID supplied should match and be consistent with the IDs that are used by the web server that a customer interacts with.
Site Analytics
The Site Analytics report is an enhanced sales performance reporting tool that enables merchants to filter sales data across several specific dimensions such as page type, page area, strategy, or strategy family.
Strategy
Strategy is an algorithm that leverages purchase and/or browser behavior to generate product recommendations that align with specific business objectives.
Strategy Family
Groups of strategies that share similar objectives. Strategy families include CategoryBestSellers, CategoryBoughtBought, MoversandShakers, New Arrivals, Personalized, ProductBoughtBought, ProductViewedBought, Product ViewedViewed, RatingsReviews, SearchBought, and SiteWideBestSellers.
Strategy Message
A descriptive message telling a customer why a certain product is being recommend, driving higher customer engagement. For example, "Customers who viewed this item also viewed:".
Über
Über is the Omnichannel Personalization decision engine that determines the appropriate recommendation strategy to display given consumer context.
User ID
A User ID generally connected to an email address, is a direct link to the login credentials a customer uses to log in to and buy products from an e-commerce site. The retailer can use this actual ID or a hash of the ID. The user ID is provided by the retailer through instrumentation and crucial to generate user specific recommendations between sessions.