Non-intrusive piece of delightfully personalized information to continually engage users
E-Commerce・Recommender Systems・Conversational Interfaces

Work In Progress

Our Elements guide is still in progress, and therefore lacks full visual and technical assets. We hope to release them by summer of 2020. Thanks for reading Lingua Franca!


The re-engagement element is designed to add a touch of empathy to human interactions with AI tools such as voice assistants, recommenders, intelligent dashboards, among others. It involves placing a small, optional recommendation that is shown or announced infrequently (such as once a day or upon opening the app). This kind of optional engagement provides the user with a momentary delight that may also draw them back to the service.


Like asking the bartender for a suggestion, there are times where a ‘tastemaker’ or ‘expert’ is desirable. Tasteful suggestions can bring clarity an familiarity to a product, providing easy-stepping stones for new users. However, a suggestion is not always desired, nor is it always helpful. If the recommendation is too generic, users will not see the value. If the recommendation is inaccurate, users may lost trust and disengage.

A valuable way to frame a re-engagement is the metaphor of ‘while you were gone’ or ‘since you were last here’. This clarifies the purpose of the re-engagement as optional assistance, while also giving the user a sense that the system is perhaps ‘thinking’ about them.


UX designers may find it preferable to structure the re-engagement along a fixed frequency (e.g. daily) in order to create predictability and encourage users to stay active on the platform.


An infrequent recommendation can be framed as a collaborative filtering[1] task, where the AI compares a user’s history to other users’ histories to determine a recommendation that is missing for that user. System architects may want to consider the role of human curation in this re-engagement system’s recommendations. For example, perhaps humans curate the recommendation set, and an AI only selects from that human-curated pool. This can be implemented either as a re-weighting step, or by using classification to place users into different categories for recommendation.



  1. Collaborative Filtering on Wikipedia ↩︎