Tell me about a product you love, and how you would improve it.

I'll pick Spotify. A quick check on the lens before I dig in.

Spotify's core job for me: be the effortless soundtrack to my day, and keep me discovering music I'll love without work. Its user is the broad listener who wants both familiarity and novelty at near-zero effort. The genius is that it makes discovery passive, Discover Weekly, autoplay, radio, you find new music without trying.

Credit where it's due: passive discovery is best-in-class. Discover Weekly and the daily mixes are uncannily good, autoplay keeps the music going, and the cross-device handoff is seamless. For "play me something good," nothing beats it. I won't dwell here, because the interesting part is where that same strength creates a gap.

Before I pick a weakness, I'll lay the whole experience out as a loop, so the critique is a located judgement, not a cherry-picked gripe. The job for this user runs as a value loop, and these stages are exhaustive of the core music experience: discover → consume → keep and build taste → share. I'll grade each and go after the one that is both weakest and highest-leverage:

• Discover: best-in-class, as above.
• Consume: excellent, seamless playback and cross-device handoff.
• Keep / build taste: the weak link, and the one I'll target.
• Share: fine, but secondary to the core job.

Discovery being the strongest stage is exactly what makes "keep" the highest-leverage gap: the better the firehose, the more it hurts to let the wins drain.

The biggest weakness, for the engaged-but-not-power-user segment: Spotify is excellent at finding music and poor at keeping what you found. The whole product optimises for the next passive play, not for the loop of "I heard something I loved, help me hold onto it and build my taste."

Concretely: a song plays on autoplay, I love it, and a day later I can't get back to it. Liking a song dumps it into one giant, undifferentiated Liked Songs list. There's no easy "where did I hear this," no lightweight way to organise discoveries into something I'll actually revisit. So discovery is a firehose with a leaky bucket: the platform keeps finding me music and quietly lets the wins drain away.

I'd segment this carefully. Hardcore curators make playlists obsessively and don't feel it; pure lean-back listeners don't care. It bites the large middle: people engaged enough to want to keep what they discover, but not enough to do manual playlist labour.

The fix targets the leaky bucket without adding curation labour:

• A "recently discovered" memory: an automatic, browsable history of new-to-you songs you reacted to, played fully, replayed, liked, with where you heard them. Closes the "I can't find that song" gap with zero manual work.
• One-tap lightweight organisation: from that memory, save into auto-suggested small collections ("songs like your Tuesday run mix") rather than one giant Liked list.
• A discovery-retention nudge: a weekly "keepers from your discoveries" surface, so the wins compound into taste instead of draining.

I'd ship the recently-discovered memory first: it's the smallest piece that tests the core thesis, that engaged listeners want to retain discoveries, and it needs no new content deals or heavy ML.

I'd validate in layers:

• Leading: the share of discovered songs revisited within two weeks, and usage of the new memory surface. A direct read on whether retention-of-discovery is real.
• Lagging headline: retention and listening of the target middle segment, are they listening more and staying because their discoveries stick?
• Guardrails: it doesn't cannibalise the passive-discovery magic (people still let autoplay run), and it doesn't add clutter that hurts the lean-back experience.

Beyond the "do users care" challenge already handled, two more I'd guard against:

It could clutter a deliberately simple product. Spotify's strength is effortlessness; a new surface risks adding cognitive load. So it has to be automatic and tucked away, not another tab demanding work.

Business angle: retaining discovery deepens engagement, which helps retention and premium value, so it likely aligns with the business. But if it pulled listening away from the algorithmic surfaces Spotify steers and monetises, that's a tension I'd watch.

So: I love Spotify for making discovery effortless, its biggest weakness is that the same design lets discoveries drain away for the engaged middle, I'd fix it with an automatic recently-discovered memory and lightweight one-tap organisation, validate with revisit-rate and segment retention, and hold guardrails so the fix doesn't clutter the lean-back magic that makes the product great.