Redesigning Discovery and Engagement on Netflix

✳The problem they went after

Netflix runs on a mission to entertain the world, and this teardown checks how the end-to-end experience holds up for three different people: the first-time sign-up, the returning viewer, and the occasional user. Splitting them apart matters, because friction a daily viewer absorbs without noticing is exactly what pushes a casual user toward cancelling.

✳What they actually did

The friction comes out specific. Recommendation carousels sit buried under heavily promoted content, download prompts only appear once a title is open, there is no weekly or daily subscription for casual viewers, and the desktop exit flow is more complex than it needs to be. Credit goes where due: streamlined onboarding, multi-profile support, crisp content descriptions. Reverse-engineering then covers personalised recommendations, regional top-10 lists, and quick-watch curation.

✳What you can take from it

The transferable bitThe reverse-engineering section explains each feature's retention and licensing value rather than just describing what it does. The segmentation then splits binge watchers, casual series viewers, movie consumers, and games users by discovery need, so every pain point stays attached to a specific person.

✳The problem they went after

Heavy weekend viewers hit decision fatigue scrolling a library too large to sort by eye. Research found they turn to Google, Reddit, and Twitter to decide what to watch rather than trusting in-app recommendations. Content disappearing without notice and a fragmented multi-platform landscape chip away at whatever trust is left.

✳What they actually did

Three journeys carry most of the friction: deciding what to watch, watching sports, and finding authentic regional content. Each is examined across hardcore OTT lovers, time-constrained parents, and casual weekend viewers, all of whom hit recommendation misalignment and platform-switching overhead. Opportunities that surface include notifying users before content is removed, pulling third-party review signals into browsing, short-form previews in the style of YouTube Shorts, and pricing changes such as annual plans and weekend entertainment passes.

✳What you can take from it

The transferable bitUsers going elsewhere to make a decision is a product signal, not user behaviour to work around. Once you accept that Reddit is doing the recommendation job, integrating outside review signals stops looking like a compromise and starts looking like the obvious fix.

✳Everything else they shared

✳The problem they went after

Netflix viewers struggle to find things they enjoy, worn down by excessive scrolling and recommendations that miss. Research narrowed the mess to four problem areas: irrelevant suggestions, scroll fatigue, originals promoted over everything else, and a rating system giving the algorithm weak signal. The feed buries good content beneath originals and low-quality titles.

✳What they actually did

Nine solutions come out of SCAMPER. The strongest pull IMDb and Rotten Tomatoes into the browsing experience, add a TikTok-style swipe interface for fast content decisions, and replace the infinite-scroll home screen with a persona-mapped layout capped at genre-specific top-five lists. A conversational panel asks two questions about mood and available time, reaching hyper-personalisation without a long onboarding quiz. RICE scoring across six high-confidence ideas puts ratings accessibility and review aggregation first.

✳What you can take from it

The transferable bitGoing wide before going narrow is the shape here. SCAMPER produces volume, RICE cuts it, and the confidence column does real work by excluding ideas the research cannot support. Note also that the winners are the cheap ones: fixing rating signal and aggregating reviews beat rebuilding the home screen.

✳Everything else they shared

✳The problem they went after

The persona is a busy working professional who scrolls twelve to fifteen content categories before finding something to watch, and often abandons the session or opens a competing platform instead. That is decision fatigue caused by too much content, poorly prioritised, and it costs Netflix a session it had already won.

✳What they actually did

Three solutions are specced in detail. A post-viewing feedback popup captures a rating when a show ends and feeds it back into recommendation recalibration. Trust signals from IMDB and Rotten Tomatoes appear on content thumbnails as clickable badge overlays. A short-form vertical scroll preview, inspired by YouTube Reels and paired with Tinder-style swipe, lets users preview and shortlist quickly. Each arrives with screen flows, UX intent per screen, and tech layer notes.

✳What you can take from it

The transferable bitRead this for what a PRD carries beyond the feature description: RICE-prioritised selection, milestone definitions, pitfall mitigation, and measurement built on click-through rate for recommended content, session length, and in-app engagement. The UX-intent-per-screen section is what keeps design and engineering reading the same document.