Additional Considerations

Overview

There is no “one-size-fits-all” approach with best practices application, even within the same vertical. This information is intended for the associates who are implementing Recommend on the retailer's site in the most advantageous way to guarantee success. Following the guidelines on the pages that follow helps to safeguard performance of Recommend on a retailer's site.

All sites are different, and what works best for one is not guaranteed to translate into gains for another. As such, treat this guidance as general best practices, a foundation for optimization from which site-specific tuning should occur. Understand the logic behind our optimization methodology, the questions to ask, so that you can adapt and apply as appropriate.

Merchandising Rules

Rather than attempting to document what we should specifically do in each situation (an impossible task), it’s more effective to be aware of the factors that should influence your configuration decisions. The answers to the following questions should inform our approach:

• Are the products configurable—size, color, etc? Such catalogs require configurability built into the recommendations layout, for example, Quickview. Absent this, customers engaging with recommendations deep in the funnel will be diverted from the checkout process which puts Conversion at risk.
• Do the products have visual appeal? Aesthetic verticals elevate the importance of product imagery and potentially de-prioritize names/descriptions.
• Does cross-selling have strict criteria? Is compatibility critical? Is it important that we recommend products outside of the category represented by what’s in the Cart? Unless there is a large corpus of purchase co-occurrence data from which to model recommendations, it’s unlikely that behavioral cross-sells will consistently ensure a compatibility match or present complementary buys—which leads to a frustrating customer experience, missed sales opportunities, and merchandise returns. For verticals such as Electronics and Appliances, it’s important that we consider supplementing or supplanting our out-the-box behavioral recommendations with other solutions for improving compatibility—things like Advanced Merchandising, offline transaction data, or custom strategies. This is a critical optimization opportunity that can mean the difference between neutral (or even harmful) cross-sell recommendations and ones that materially boost cart values.
• Small or expansive catalog? Customers shopping broad catalogs have the biggest “pain” finding what they want due to the “paradox of choice”. There are too many options, and smart personalization can help quickly navigate them to the right product(s). The smaller the catalog, the lesser our ability to help. Tactically, this means fewer placements per page since the quantity of relevant options for recommendation is reduced.
• Is it a flash retailer or are the products short-lived? If the retailer has a fluid inventory, we’re probably lacking a robust set of shopping data around each product. For example, if an item only lives on the site for 7 days, our cross-sell models, which have a lookback of 60 days, are not likely to be very strong. Consequently, we will need to be more thoughtful with our recommendations— opting for fewer placements, potentially with a higher level of curation.
• Do price and availability change frequently during the day? If so, a catalog feed that’s processed only once per day will be insufficient for keeping recommendations current, and we run the risk of suggesting unavailable or mispriced items. To avoid creating a frustrating experience, consider multiple or delta feeds, or a pricing API.