Netflix “Tonight”
Redesigning netflix.com discovery so members spend their evening watching, not scrolling — by asking what fits tonight before showing the catalog.
A self-initiated side project. Not affiliated with or endorsed by Netflix. All titles, rankings and poster art are fictional and original to this concept.
From utility bills to Friday nights.
By day I design products in the utilities space — field work and outage tools for crews and storm-duty teams who need clarity when the stakes are real. Netflix is the product I open every night, and the kind of product I most want to help shape next.
I took this on myself, outside of work, to show how the skills I use every day carry over to streaming and entertainment.
Designing for everyone
Utility customers span every age, ability and device — so does Netflix’s audience. Accessibility is a default in my work, not an add-on.
Making complex choices simple
I turn dense information — rates, plans, usage — into clear decisions. It’s the same muscle as turning 40 rows into three confident picks.
Measuring what ships
At work I track task success and task time — at National Grid, testing lifted task success from 68% to 95%. Here I bring the same rigor with a real metric and test plan.
The web home page is a wall of near-identical rows with no context about time, mood or company — so choosing takes longer than it should, and some sessions end with nothing watched.
A “Tonight” bar (mood, time, who’s watching) that reshapes the page into three picks, each with a plain-language reason, plus fewer rows that each state their intent.
Cut median time-to-play to under 3 minutes and reduce sessions that end without playing. Targets to validate in an A/B test, not shipped results.
More choice made choosing the hardest part of watching.
Netflix knows this. In 2024 its product team described members doing “eye gymnastics” across the home screen, and in 2025 it shipped a TV redesign with reason tags like “#1 in TV Shows” and mood-responsive recommendations. In the web captures I reviewed, netflix.com still uses the older pattern: a billboard and dozens of horizontal rows.
Six things slowing the decision down.
I reviewed the logged-in web experience using captured screens on Mobbin (browse rows, billboard, title modal, hover cards), scoring each issue by how directly it delays a confident choice.
{{f.t}}
{{f.d}}
Not personas — evening modes.
The same person opens Netflix in different states. Designing for the situation, not the demographic, told me what the home page should ask first. These are proto-modes from secondary research, to validate in interviews.
{{m.t}}
“{{m.q}}”
…help members find something that fits their time, mood and company — and feel sure it’s worth it — before the scroll wins?
{{p.t}}
{{p.d}}
Ask one question first. Then show less, better.
The Tonight bar
Replaces the autoplay billboard as the first thing on the page. Three optional inputs — mood, time available and who’s watching — instantly re-rank the picks below. It remembers last night’s choices and never blocks: skip it and the page still works.
- Single tap per dimension; no form, no submit button
- Time filter uses episode or movie runtime
- Fully keyboard and screen-reader operable (radio groups)
Three picks, each with a reason
Recommendation quality isn’t only the algorithm — it’s whether people trust it. Every pick shows one evidence tag (trending, critics, a friend’s rating) and one sentence tying it to the member’s own history. No hover needed; it works on touch and with a keyboard.
“Not tonight” dismisses a pick and asks one optional reason, which feeds the next three.
Six rows that say what they’re for
Below the picks, the catalog stays — but capped at six rows by default, each named for a job rather than a genre. Duplicates across rows are removed. “Browse everything” is one click away for people who love to explore.
Picking together
When “Partner” or “Family” is selected, each person can add their mood from their own phone. The page proposes overlap picks instead of making one person scroll for everyone — targeting the couch negotiation directly.
| Moment | Today on netflix.com | Tonight concept |
|---|---|---|
| {{c.m}} | {{c.a}} | {{c.b}} |
Faster to play, without watching less.
{{x.t}}
{{x.d}}
Validation plan. Six moderated sessions with recent “gave up and closed the app” members, then an unmoderated first-click test comparing today’s home to the Tonight home on “find something for a 40-minute evening.”
What this taught me — and why I want to do it full-time.
Crossing domains: utilities taught me that clarity earns trust when a mistake costs money. In entertainment the cost is a wasted evening, but the principle holds — cut the effort, explain the why, respect people’s time.
Trade-off I made: removing the autoplay billboard likely costs promotional reach for new originals. I’d test a compromise — a slim billboard inside the picks when it matches tonight’s inputs.
Biggest risk: asking anything adds friction. That’s why every input is optional and defaults to last night’s answer.
Next: extend the Tonight bar to TV, where Netflix’s 2025 redesign already introduced reason tags — the two ideas reinforce each other.
Clear choices, whether it's a storm call or a Friday night.
I’m looking for a product design role on a team that measures design by outcomes — tools people depend on at work, or products they choose every night. If that’s your team, let’s talk.
Reelgood & Learndipity Data Insights survey of 2,000 US adults (via TheWrap/Yahoo) · UserTesting & Talker Research streaming study (via BGR) · Nielsen Total Audience Report Q1 2019 (via TV Insider) · Reuters on Netflix’s 2024 TV test · Netflix Tudum, “Netflix Is Getting a New Look,” May 2025 · Screens: Mobbin.