the metrics that matter: why engagement isn't the goal

you’ve probably heard the terms DAU (daily active users) and session length tossed around in tech circles. they’re the default metrics for measuring success in apps, especially social or entertainment products. if people are spending hours on your app, you’re winning, right?

but when it comes to ai companions, that mindset isn’t just flawed, it’s harmful.

the problem with optimizing for time spent

an ai companion that keeps you glued to the screen for three hours isn’t a companion. it’s a trap. it might be entertaining, or distracting, or even compelling in the moment. but if you walk away feeling drained, anxious, or more isolated than when you started, what was the point?

we’ve seen this before. social media platforms optimized for engagement ended up feeding us outrage and comparison. video platforms kept us scrolling long past bedtime. these designs exploit human psychology for attention, not wellness.

an ai companion should do the opposite. it should help you regulate, reflect, feel heard, then gently encourage you to go live your life. the goal isn’t to become the main character in your day. it’s to be a supportive sidekick.

designing for satisfaction, not session length

at lucy, we think about something we call 'closure.' did the user finish the conversation feeling better? did they feel understood? did they gain clarity or comfort? if yes, then we’ve done our job, even if the interaction was only ten minutes long.

this isn’t just a philosophy. it’s built into how lucy works. our proactive engine, for example, is designed to notice when you’re in a stable or positive state. when that happens, it backs off. it doesn’t push more questions or try to create drama to keep you talking. it might even suggest you take a break, go for a walk, or check in with a friend.

we’re not perfect at this yet. sometimes lucy might misread the mood or overstep. but the intent is clear: we want you to close the app feeling better, not staying on it longer.

the business tension: vcs vs outcomes

here’s where it gets tricky. venture capitalists often push for growth metrics like DAU and session length. they’re looking for hockey stick graphs and proof of 'stickiness.' it’s an easy story to tell: look how much people love our product, they can’t stop using it!

but founders and product leaders in this space have a responsibility to push back. we should be measuring outcomes, not just engagement. did we reduce someone’s anxiety? did we help them feel less lonely? did we support them through a hard day? these are harder to measure, but they’re what actually matter.

if we don’t fight for outcome-based metrics, we risk building products that are addictive instead of helpful. we risk becoming part of the problem, another app that takes more than it gives.

a companion should help you leave

a good friend doesn’t keep you on the phone for hours if you need to go. a good therapist doesn’t drag out a session just to fill time. they help you process, then send you back into the world feeling stronger.

that’s what an ai companion should aspire to. it should be a place you can go to for a quick chat, a moment of validation, or a thoughtful question, and then leave feeling reset and ready.

we’re building lucy with that in mind. it’s a work in progress, but it’s a commitment. to us, success looks like you opening the app less because you need it less, not because you’re bored, but because you’re doing better.

if you want to try a companion that aims to help you feel better, not just keep you online, come check out lucy.

meet lucy