The Observatory × Dialogue AI · Product / UXR Sprint Brief

What to build for Love Island viewers

The one insight

Retention lives in the known-person social layer, not the anonymous online community. Build for co-viewing and spoiler-safe participation. Treat discourse as reach, not as the product.

Five findings

1

The community question is really two questions

Co-viewing with known people and anonymous online discourse are different things. One is load-bearing for loyalty, the other is not. Any 'community' feature has to pick which one it serves.

"It would absolutely lose its spark for me."P22 · 16:27
2

Online discourse is a job-doer, not a driver

When viewers use the online layer it validates their reaction, decodes the edit, keeps the show top of mind, and feeds gossip. Its main cost is spoilers. It grows engagement; it does not retain.

"I usually just look at reviews first, see what the common consensus are on it. And then afterwards, I watch it and then browse Reddit again to see if my views on the show or the episode are consistent."P10 · 04:16
3

Catch-up viewers are locked out of live agency

Voting is the existing agency hook, but roughly half the sample catches up later and cannot vote. They already get spoiled by their feed anyway. There is a whole mode of participation missing for them.

"Usually, I already see it, and I'm not gonna be able to get spoiled 'cause I already kind of see it when it comes across my feed."P48 · 12:03
4

The engine is decompression, not romance

The dominant reason to watch is switching off after a long day. Features that add effort or homework fight the core job. Romance-forward framing misreads the room.

"It's more or less just like how you turn off your brain and you watch it, and if you turn off your brain and you watch it, it is enjoyable."P11 · 05:17
5

A loyal minority is attached and hiding it

A small group form a real bond with the cast and treat it as their most private admission. They are the most loyal users. Design that respects the attachment (without exposing it) protects the base.

"It's a bit sad to say, but I'm a stay-at-home mom, so I don't have much friends or interaction. So seeing reality TV really makes me feel connected to people."P28 · 01:08

Jobs to be done

When I finish an episode
I want to check my read against other fanswithout getting spoiled on episodes I haven't seen yet.
Post-episode reaction space, gated by watch progress
When a new episode drops
I want to watch it with my peopleeven when we aren't in the same room.
Co-watch scheduling and shared reactions for known groups
When I've fallen behind
I want to catch up and still participatenot feel locked out because voting already closed.
Spoiler-safe async agency for catch-up viewers

User needs

DecompressSwitch off, demand nothing, escape the day"Just a release. They allow you to relax without having to think too much. If it's been a…" · P45
BelongShare the reaction with my people, feel part of it"It would absolutely lose its spark for me." · P22
ValidateConfirm I'm reading the show right"I usually just look at reviews first, see what the common consensus are on it. And then …" · P10
ControlChange the outcome, feel some agency"I actually get to change things up as if I'm like controlling the simulation." · P31
ProtectGuard my experience from spoilers and toxicity"I tend to think that communities by default ruin shows." · P35

Action split

This sprint

  • Prototype a spoiler-safe post-episode reaction space, gated by each user's watch progress.
  • Test co-watch scheduling for known groups (a partner, a friend group, a chat).

Not now

  • Building anonymous-community / subreddit-style features as a retention play. The evidence says that layer is reach, and some of the most engaged users avoid it.