Several years ago I was on a team working on customer-facing analytics for a service. The product manager for the team approached things with what seemed to be the tech playbook du jour for growth and laid out some typical engagement metrics to optimize. As with many other goals on that team (and more widely) how some of these things would translate to user value fell apart with a couple pops off of the why stack (some wider thoughts around modeling and validation will dig deeply into this topic).
One sizable issue in my mind was that we were looking to build a single pane of glass to present the most valuable information to users, but one of the functional KPIs the manager had identified was the session duration with the goal of driving that number higher. If we were realizing the goal of surfacing insights for ready consumption…why would the user be spending more time viewing it? It certainly seemed more plausible that longer sessions suggested the information was not being efficiently expressed and therefore the user needed time to find or understand it. In such cases it feels like the gauge of an optimal UX would be frequent, but very short engagements (implying we were providing very valuable information which was being regularly used). This is analogous to services like a search engine - more time spent on the results page suggests that there are issues with result relevancy or presentation such that what is actually desired cannot be readily located.
This line of thought was largely prompted due to some discussions around the use of services like social media. Unfortunately we've drifted into a mindset that normalizes use of technology independently of what it is doing for you, and then such services may simply be seen as competing against each other for your attention rather than providing clear value to the users. In terms of getting value out of technology a basic assertion would be that the length of time necessary to achieve desired results is inversely proportional to the net benefit of the technology.
A likely problem is that people tend to rely on increasing converged information channels - the same technology that is used for distraction/entertainment is also often used for both important information (i.e. people getting their news through TikTok) and connections back to reality (where, per TBL, many services have drifted away from that being a priority). For individuals this normalizes habits like doom-scrolling and incidentally consuming what is presented which inhibits recognition of the nature of how such information is consumed (could I have found this more quickly? is this information trustworthy?), and from the perspective of building a solution prospective the goal of engagement can subsume that of satisfying actual information needs. This feeds the vicious cycle of the use of some technologies for their own sake rather than as a tool serving a purpose that can be validated.
One guard against this (one I've started to adopt before this idea crystallized) is having specific purposes for different services and systems that I use. This keeps me from falling into some proximal vortexes and also provides a way to have a clearer sense for what time I'm spending where and how much I'm getting out of that time.