LinkedIn's 'Dwell Time' Metric Explained
A key, but often overlooked, metric for content ranking on engagement platforms is "dwell time." An analysis explains how platforms like LinkedIn use the amount of time a user spends on a post without interacting as a powerful signal for content quality and visibility.
LinkedIn's algorithm quantifies "dwell time" in two distinct ways: "on the feed" dwell time, which begins measuring as soon as 50% of a post is visible during a scroll, and "after the click" dwell time, which tracks the time a user spends on content after clicking a link. This dual measurement allows the platform's machine learning models to differentiate between initial interest and deeper engagement. The shift towards this metric addresses the "noisy" nature of explicit signals like clicks and likes. Many users might click on an article and immediately bounce, a behavior LinkedIn calls a "click bounce," which provides a poor signal of content quality. Dwell time was integrated to better account for the ~90% of users who consume content passively without leaving explicit reactions, providing a richer, more nuanced signal of content value. In recommendation systems, treating all clicks equally can lead to optimizing for clickbait. Academic and industry research, such as work presented at RecSys conferences, explores using dwell time not just as a feature, but as a direct prediction target (predicting watch time instead of click-through rate) or as a weight to give more importance to clicks that result in longer engagement. This use of implicit feedback is standard in large-scale recommendation engines. Meta's news feed, for example, uses complex multi-task learning models to predict not just likes and comments, but also the probability of a user dwelling on a post for at least three seconds. These predictions are then aggregated into a single, personalized ranking score for every piece of content. Similarly, Netflix's recommendation engine moves beyond simple ratings by heavily analyzing view duration and other behavioral signals to power its personalization. The system considers a user's "time budget" for evaluation, optimizing the ranked list of titles to maximize the chance of engagement within the limited time a user spends browsing. The application of dwell time, however, is not universal. While Microsoft's Bing has confirmed it uses the time between a search click and a return to the results page as a quality signal, Google has consistently denied that dwell time is a direct ranking factor. Google's engineers state that while user satisfaction is critical, using direct user interaction metrics like dwell time for search ranking is "made up crap" and that the real system is simpler than such theories suggest.
