Visual Similarity feature is currently on Closed Pilot release for a limited number of clients only. We do not wish to accept new pilots, thank you for your understanding.
Visual Similarity is a pre-defined, product recommendation logic which enables you to successfully recommend brand new products for the customers who are most likely to buy. This product recommendation works without having any browsing or purchase history. You can select the predict recommendation type listed here:
Content > Personalization Rules > Create Personalization Rule > Visual Similarity for New in Stock
Prerequisites
The prior conditions and the requirements are the same that applies to the Personalization rules. See, Personalization rules' prerequisites.
How does it benefit me?
Visual Similarity is a powerful recommendation tool in Personalization Rules, which helps you create true, one-to-one personalization. It enables you to promote your newly arrived collections for those customers who are most likely to be interested in the product.
Use cases
Visual Similarity is based on the New In Stock product recommendation logic which enables customers to provide 1:1 recommendations based on their taste, even without existing user behavior data related to the new items.
Recommending items visually similar
You can create a campaign for your products let's say in the fashion industry, because customers are the one who are most likely interested about visually similar products.
Recommending items without history
You can create a campaign for completely new products where you do not have any browsing and/or purchase history.
Interpreting the main logic
Rules within Visual Similarity
- What is considered to be new: You can set up your own rule about what is considered to be new item/product (between 1-30 days).
Item level matching: It keeps the recommendation within a certain category that the customer was previously browsing. Brand and Custom fields can be used.
Known limitations
Currently we do not have support for the following:
- Missing Parent-child level matching: One of the elements of our calculation is item-to-item similarity. This means that we might show the same item multiple times in one recommendation, just in different sizes, or color.
- Localized Fields: Not available yet.
- Like other Personalization rules, Personal Logic can only be used in batch type email campaigns. It cannot serve Transactional use cases (for e.g. abandoned cart).
- You can only send product personalized email campaigns to those customers who received any email from you in the last 180 days.
- Product Availability needs to be selected and filtered manually.