Overview

Recommend provides tailored personalized recommendations on eCommerce site to the customers, as they navigate through the e-commerce site. Recommend engine constantly learns from customer behavior to determine which products and services are of interest to the costumers and then adjusts, in real-time, to provide highly relevant recommendations with messages, letting the customer choose the best-suited recommendation to their needs and interests. Recommend dashboard provides merchandisers unparalleled flexibility to monitor performance and to implement new merchandising strategies including product selection, messaging, and inventory management – each informed by unique individual business priorities.

Recommend ingests data from the retailer in various feeds that it uses to recommend products to end-users in placements on the retailer's website, using a complex mix of strategies, rules, and an engine that continuously works to display the most relevant recommendations in real-time.

Implementation Team

The team must include stakeholders who perform implementation and stakeholders authorized to do sign-off after each phase of implementation. The implementation team consists of stakeholders from the following departments.

• Merchandising

• eCommerce

• Finance

• Marketing

• IT or Engineering UI (User Interface/Usability)

Project Owner

An effective approach is to choose a single project owner from any of the departments, mentioned above, which creates a single point of contact for the consolidation of information, management of project timelines, and determination of deliverables. The following teams can be built from the Engineering department.

Deployment Team

The following teams should be built from the Engineering department. Having these teams onboarded and fully engaged for Recommend platform deployment ensures efficiency, quality, and results.

• Web Developers/Engineers
  • Develops the code
  • Integrates the Algonomy code
• Build Engineers
  • Performs and verifies production push
• Quality Assurance Team
  • Tests pages
  • Checks feeds across the browser matrix
  • Performs load testing

• Security Team
  • Ensures that personally identifiable information is not being transferred to Algonomy Usability Team

Usability Team

• UI Engineers
  • Defines best practices for usability across the website Implementation Guide for Merchants
  • Provides guidance
  • Performs testing
• Design Group
  • Defines and approves look-and-feel Marketing Team to Deliver referral management information
  • Provides guidance

Merchandising Stakeholders

• Provides guidance and approves look-and-feel of recommendations o Approve strategic recommendation messages

• Determines product catalog and any product “black lists”

• Leverages Algonomy merchandising controls in support of specific merchandising strategies

The Recommend Implementation Process

The Algonomy implementation process is designed to be highly consultative and efficiently structured, with mutual checkpoints along the way. Each stage progresses with a focus on what works best for the client’s ecommerce site, customers, and sales.

Stage 1: Project Kick-Off

During the initial phase, the optimal integration approach is identified in support of business objectives and success metrics. IT support and security actions are identified, respective owners assigned, and schedules developed. Initial UI mockups are approved.

Stage 2: Initial Integration (Code Complete)

At this stage JavaScript or web services integration is implemented. The look-and-feel of the recommendations is tested. Any IT actions determined in Stage 1 will also be implemented for testing, as will verification of the data-collection process.

Stage 3: Listen Mode

During Stage 3, the JavaScript implementation is deployed into production; however, during this phase, recommendations are not visible to customers. Customer behavior data is gathered, the initial models are built and verified, and the HTML/JS in the browser matrix is tested using a forced-display mode.

Stage 4: Live

During Stage 4, recommendations are now visible to half of your customers. This allows for the further refinement of the models as sales lift begins to ramp up.

Stage 5: Live Optimized

During Stage 5, the models refined in Stage 4 are extended to full audience. Lift will continue to climb during this period. Constant feedback loop is maintained to ensure sustained results and adapted to capitalize on underlying behavioral trends.

The Recommend Deployment Schedule

Stage	Summary	Retailer Tasks	Retailer Team	RR Tasks
1 (1 week)	Project Kick-off	Assign project owner Approve Recommendation look & feel	Merchandisers Designers QA Team Security	Confirm integration path: pure JavaScript, JavaScript + feed Deliver Integration schedule
2 (1 week)	Initial Integration	Implement initial JavaScript Integration: Use Algonomy "Staging Server" and test "Dummy" recommendations (Optional) Set up data feed QA look & feel	Project Owner Engineering Designers	Implement look & feel Validate Data Collection
3 (1 week)	Listen Mode: No recommendations will be visible to users.	Production Deploy: Deploy JavaScript to production Initiate Listen Mode	Project Owner Engineering Build Team	Validate Data Collection Build Initial models Verify model quality
4 (1 week)	Live Recommendations are now visible to users. Sales lift begins	Sales lift ramps up Fix Data errors: expect 1-3 data errors.	Project Owner Engineering	QA HTML: Execute browser matrix. Verify Recommendation quality.
5	Live-Optimized	Sales lift improved further	Project Owner	Constant feedback loop maintained and adapted to capitalize on underlying trends
	Ongoing	Notify account manager of any feed changes, design updates. Notify of any advertising or promotional strategies Update on staff/contact changes	Project Owner Merchandisers	Continuously monitor micro-trends. Continuously update data models

 

Pre-Implementation Questions

In order to align the right resources for your project, the Algonomy team has developed a series of pre-implementation questions developed to reveal your core business opportunities and align our collective teams. Your answers will help us direct future discussions to better ensure quality of implementation, ongoing operations, customer experience, and results. The questions are grouped into three areas: Your Customers, Your Catalog and Your Site.

About Your Customers

• Do you track users across sessions with a cookie?
• What percentage of your users are registered users?
• Do you require that a user be logged in before purchasing an item? If no, what percentage of purchasers are registered users who are logged in?

About Your Catalog

• How many SKUs are in your catalog?
• How often does product data (such as price, availability) change? (For example, does your catalog change every month, daily, every Monday and Wednesday, or rarely?)
• How do you manage merchandising? (For example, do you use a service such as Omniture?)
• Are there category owners designated to manage promotions within specific categories/subcategories? (We need to understand how many stakeholders will have an ongoing operational interest in recommendations.)

About Your Site

• Does your pricing vary based on region, codes, customer type, or other variables?

• Confirm estimated timeframe for deployment of recommendations.

• What programming language/eCommerce platform is your site built on?

• Is site development managed in-house or outsourced?

• Please describe your build/deploy cycle. How often does it happen?

• What is your site traffic volume? 

  • Unique users per day:
  • Page views per day:
  • Purchases per day:

• What metrics do you use to measure your performance?

• Do you use additional software that monitors your site?

• What metrics reporting software do you use? (Knowing that you use a vendor such as Omniture allows us to help your metrics software seamlessly track recommendations through our service.)

• Which email service provider do you use?

• Would you like to integrate recommendations into your email campaigns? (we can tie recommendations into your emails without incremental integration – recommendation code is simply inserted via the email service provider.)

• Is there a new version of your site currently in development?

• Do you develop you eCommerce site in-house or is the work outsourced?

• What eCommerce platform are you running; what language is it programmed in?

Data Feeds

Another factor in expediting the Recommend implementation process is the delivery of merchant product data feeds. This basic information about your product inventory, including product description, genre, and category, ensures optimal display of the appropriate recommendations on your site. There are several types of feeds that are required.  To see detail about those feeds and what is required in each, see our Developer documents.