Dream11 · Mobile Product Designer · 2026

Watch Along Landscape.

Our live stream watch time was dead in the water at 2.3 minutes. Throwing more engagement features at it didn't help. The real fix? Making people turn their phones sideways.

The problem

Portrait mode trained users to browse and bounce. Watch time was stuck at 2.3 minutes and more features weren't moving the needle.

The solution

A posture change. We rotated the phone to landscape, gave video 65% of the screen, and moved interaction beside it. Watch time jumped to 15.2 minutes.

Watch Along - portrait to landscape experience
15.2MIN
Landscape watch time
2.3MIN
Portrait baseline
6.6×
Watch time lift

01. Approach.

People were bouncing fast. More features weren't fixing the problem.

My team owns the metrics for whether users are actually watching or just passing through. Watch time was stuck at 2.3 minutes, and leadership wanted it up.

Our first instinct was more polls and gifts to give people a reason to stay. Activity went up; watch time didn't budge.

A PM and I were just chatting when it hit us: why is every other long-form live app in landscape, but we're forcing portrait?

beat · 01
The metric
Average watch time: 2.3 minutes.
beat · 02
The first bet
Shipped more features. Usage increased, watch time didn't.
beat · 03
The spark
If serious viewing is landscape everywhere else, why are we fighting it?

Every other long-form live app treats landscape as the default. The only portrait holdout was TikTok - and that's for short loops, not a 90-minute cricket match. We didn't need a layout refresh; we needed a posture change.

Watch Along, in three acts.

Our product was acting like a casual TikTok feed - terrible for long-form sports. We had to shift the user's posture from "browse and leave" to "rotate and watch."

act · 01
Portrait
Portrait trained users to quickly browse and bounce.
act · 02
Landscape
I owned the transition to a fully rotated mobile landscape mode.
act · 03
The bet
If we change how they hold the phone, watch time will increase.

Portrait was training them to leave.

Our features weren't bad. The format was just wrong. We were putting 90-minute matches into a highly interruptible container. The issue was purely posture.

Three signals from portrait mode.

signal · 01
Not Sticky
Portrait feels disposable. One swipe takes you away.
signal · 02
Wrong Format
Serious sports viewing happens in landscape everywhere else.
signal · 03
Lost Center
The stream never felt like the main event.
Portrait Watch Along - stream view and the inherited gesture map
Portrait Watch Along - stream view and gesture map Portrait Watch Along - stream view and gesture map

02. Insight.

We already had the signal. We just needed conviction.

This wasn't a months-long research sprint. We just looked at the top platforms and realized the pattern was undeniable.

The answer was already visible.

Every platform built for long sessions defaults to landscape and side panels.

YouTube Live
platform · 01
Landscape
YouTube Live - 25–40 min sessions, side-panel.
Twitch
platform · 02
Landscape
Twitch - 30–95 min sessions, side-panel.
Kick
platform · 03
Landscape
Kick - 35–80 min sessions, side-panel.
TikTok
platform · 04
Portrait · outlier
TikTok - 8–12 min sessions, overlay-first.

Landscape · 16:9 · side-panel interaction - the default for serious viewing.

Rotate the phone, increase the watch time.

Watch time was our true north star. Chat and activity health were our secondary reads.

Business readout · north-star
  • Watch time - the one number that mattered.
Secondary read · supporting
  • Chat health · messages per active viewer.
  • Activity health · polls, predictions, reactions.

Design principle - video first. Interaction beside it.

03. Role.

I owned the mobile product translation.

I owned the fully rotated landscape mode - making the behaviour-shift thesis a reality.

What I owned
  • The fully rotated landscape mode - 16:9 viewing.
  • Side chat & activity panel - interaction beside video.
  • Feature UI - Gifts, Squad Goal, On Stage in landscape.
  • Overlays - Hit and Miss, Polls in the new posture.
  • The hierarchy - deciding what stays and what shifts.
How I worked
  • PM - turning our hypothesis into shippable decisions fast.
  • Engineering - rotation handling and the OBS pipeline.
  • Ads - squeezing ads onto a much tighter mobile canvas.

04. Solution.

Rewriting the hierarchy, not just rotating it.

This wasn't just slapping a chat panel on the side of a video. We had to keep users engaged without letting the UI run over the stream.

Three moves to make it real.

Move 01 · posture - make video worth rotating for
Portrait to landscape - watch posture shift Portrait to landscape - watch posture shift

Make video worth rotating for. Rotate the phone and video instantly takes the center - the portrait feed shifts into a 16:9 watch posture.

Move 02 · layout - a 65 / 35 split
Landscape frame - 65 / 35 video and side panel split Landscape frame - 65 / 35 video and side panel split

A 65 / 35 split. We gave chat just enough room without crowding the stream - video holds 65%, the side panel takes 35%.

Move 03 · interaction - overlays move sideways
Chat panel and activity stack beside the stream Chat panel and activity stack beside the stream

Overlays move sideways. We didn't delete anything, just repositioned it to the side rail - chat, gifts, and polls are still one tap away.

One frame. Video centred. Activity beside it.

The stream holds the left two-thirds. Chat and activities take the right rail. Overlays pop up inside the video so they don't break the watch posture.

A better viewing story required two compromises.

Focusing on the stream forced us to make difficult UI tradeoffs.

tradeoff 01 · ads

Keep ads. Shrink them.

Kept · 320×50 ad unitCut · full-size in-stream

Ads pay the bills, so removing them was a non-starter. We shrunk them to 320×50 units - visible, but out of the way.

320×50 is a standard that already performs well; full-size ads competed with the watch posture.

tradeoff 02 · fantasy & payments

Rotate for team creation.

Watch · landscapeTask · portraitReturn · landscape

Team creation was too heavy to squeeze into a side panel alongside video and chat. So we shift the phone back to portrait for those tasks, then flip back to watch: landscape → rotate to portrait (create team, buy DreamBucks) → rotate back to landscape.

Fantasy was already native to portrait everywhere else, so the rotation kept the mental model intact.

05. Impact.

2.3 minutes → 15.2 minutes of watch time.

The first month validated our bet: this was a fundamentally stronger way to watch.

15.2 min
Launch readout

Average watch time · first-week landscape sessions.

2.3 min
Portrait baseline

Average watch time per user before the orientation shift.

We bought watch time. We paid in interaction depth.

Conversation held steady, but polls softened slightly because we explicitly made the video the star.

chats / user
4.72
Conversation stayed perfectly alive.
reactions / user
1.19
Mostly flat once the hierarchy settled.
predictions / user
1.76
Held flat after the move to landscape.
polls / user
1.72
Slightly weaker in the new layout, which we expected.
DreamBucks / user
45.23
Monetization showed the cost of letting video lead.

06. Reflection.

Turning a directional hypothesis into a shippable decision - fast.

My strength · reading the pattern
  • Reading benchmark patterns fast.
  • Holding the line on the design against revenue pressure.
  • Keeping the system coherent without waiting for perfect certainty.
The tradeoff I made · watch over depth
  • Poll engagement dipped, but by design.
  • Breaking orientation for fantasy flows.
  • Watch time was our north star, and we couldn't hedge our bets.

Orientation is not formatting. It changes intent.

lesson · 01 / product

Posture changes the story.

We didn't fix watch time by decluttering. We changed the story: less browsing, more staying.

lesson · 02 / team

Lean teams reward informed conviction.

We didn't wait for perfect certainty - strong product judgment was enough to move fast.

See more case studies

DreamStream

DreamStream - Designed and shipped Dream11's live sports console in 60 days.

Live Streaming
DreamGuru

DreamGuru - Built a two-sided fantasy guru marketplace, free first then paid.

Creator Economy