YouTube 2.0:
fixing long-form discovery.
Long-form viewing on YouTube is not declining. It is being displaced by faster formats. This is a set of product interventions to win that attention back, not by creating new surfaces, but by reducing friction inside the feed that 491 million Indian users already use every day.
This page is the full Product Requirements Document for YouTube 2.0, a conceptual redesign covering discovery research, 171 survey responses, four product interventions, and success metrics. It is a design exercise, not a shipped product.
Project overview
OverviewLong-form content still accounts for 70%+ of total YouTube watch time, it is not in structural decline. The problem is feed displacement: Shorts have absorbed the mobile browse real estate that used to surface long-form videos. This PRD proposes four interventions, grounded in a 171-response primary survey, that fix discovery inside the feed users already have, not a new destination they'd have to learn.
Built by
Gaurav Gupta, independent product exercise, not affiliated with or commissioned by YouTube/Google.
Not decline.
Displacement.
01 · Problem & Platform
The central problem
With the rise of short-form video, long-form videos (10+ minutes) get surfaced less in the mobile feed. Shorter attention windows and competing short-form platforms have left long-form content underused, even though the underlying appetite for it hasn't gone away.
The core thesis: Long-form content is not in structural decline. It still accounts for 70%+ of total YouTube watch time. The problem is feed displacement, Shorts are absorbing mobile browse real estate that used to surface long-form videos.
Why this matters for the business
YouTube is a two-sided marketplace where creators, viewers, and advertisers are connected by a matchmaking engine. Any gap in that matchmaking is a gap in value for all three sides.
- Top creators (>5M subscribers) historically drove maximum engagement through long-form content. Post-Shorts, these creators have seen a measurable decline in long-form views.
- Declining creator motivation restricts supply of good long-form content, compounding the discovery problem over time.
- Long-form earns $3–$6 RPM (per 1,000 views) versus $0.03–$0.07 for Shorts, a 100x gap in per-view monetisation.
How the recommendation system evolved
2005–2011
Optimised for: Clicks. More clicks meant more recommendations, created a clickbait epidemic. Users felt tricked.
2012–2023
Optimised for: Watch time. Clicks alone didn't mean someone enjoyed the video. Views dropped 20% at the shift, they kept it anyway.
2025 onward
Optimised for: Satisfaction-weighted discovery, not just “did they watch?” but “did they feel the time was worth it?”
Five discovery surfaces, one critical gap
YouTube does not have one feed, it has five distinct surfaces, each doing a different job: Home (interest expansion), Suggested/Up Next (keep the session going), Search (declared intent), Shorts (new audience discovery), and Subscriptions (loyalty reinforcement).
The critical design gap: Since mid-2025, the Shorts and long-form recommendation systems are fully decoupled, different engines, different signals, different goals. A user's interest expressed through Shorts is invisible to the long-form system. This is the core technical problem our solutions must address.
Three structural blind spots
- Mood vs. history: The system knows what you watched last week, not what mood you're in right now. A finance viewer on a lazy Sunday still gets finance content.
- The mid-length gap: Videos between 4 and 8 minutes fall into no man's land, too long for Shorts optimisation, too short for long-form signals.
- Cold start for new creators: A new creator has almost no chance of a long video getting recommended without subscriber data. Shorts offer a much easier path.
171 viewers said
the same thing.
02 · Research
This primary research is based on a quantitative and qualitative survey distributed to 250–300 individuals, capturing 171 responses, designed to understand behaviour, intent, and friction points specifically around long-form (10+ minute) consumption.
Demographics
- 25–34 years: ~54%
- 35–44 years: ~25%
- 18–24 years: ~12%
- 45+ years: ~8.5%
- Gender split: ~55% male, ~45% female
Research themes
- General usage and platform intensity
- Long-form engagement motivation
- Drop-off triggers and abandonment causes
- Discovery entry points and search behaviour
- Feature validation for navigation improvements
What viewers said
The clearest message: they want long videos, but YouTube isn't helping them find or finish them.
People are not giving up on long-form content. They are giving up on YouTube's ability to show it to them. The interest exists. Discovery and retention are failing.
What the product signals showed
- After watching a Short, only 15% actively look for a longer video from the same creator, but 41% say they would watch one if it were easy to find. The demand exists; the pathway doesn't.
- 22% stop watching a long video because a suggested video on the side caught their eye. The platform competes against itself.
Market evidence backs it up
Creators observe per-view economics, not per-hour parity: long-form pays $3–$6 RPM vs. $0.03–$0.07 for Shorts, but Shorts grow faster. Meanwhile 19% of all users abandon a session without watching a single video, rising to 29% among 18–24 year olds. Shorts interactions actually fell 47% YoY in 2025 despite post volume surging 71%, short-form is showing fatigue, not long-form indifference.
The competitor
graveyard is clear.
03 · Landscape & Persona
How attention is being fragmented
TikTok
Captures casual, idle scrolling via an endless short-form feed, conditions users toward low-commitment viewing before a long session can begin.
Netflix
Captures intentional evening sessions through lean-back, premium long-form content, taking dedicated, high-intent viewing time away from YouTube.
AI tools
Capture informational intent through instant responses, reduces the need to watch a 20-minute explainer for an answer.
What other platforms tried: and what failed
| Attempt | Outcome | Lesson |
|---|---|---|
| Instagram IGTV | Separate long-form tab. Killed March 2022. <7M downloads vs. 1B+ users. | Users don't adopt new in-app destinations for long-form. |
| Netflix “Play Something” | Random content button. Killed Jan 2023. | Randomness without context reduces trust. |
| Netflix Autoplay | Auto-play next episode. Still active, increased consumption. | Reducing friction within format works. |
| TikTok long-form | Extended duration to 60 min. Creators >1 min gain 5x more followers. | Even short-form platforms pursue long-form. |
The architectural constraint that shaped every solution below: every bolted-on cross-format discovery feature has failed within 2 years. A solution must be native to the feed users already have, not a new surface.
Three problems, one compounding loop
P1: “I can't find what I came for.”
Most users have to manually search every session. When finding content feels like work, users settle for Shorts or leave.
P2: “What if this wastes my time?”
A 20-minute video feels like a commitment users can't easily judge. Thumbnails look similar; intros are often slow.
P3: “Why spend 15 hours for uncertain reach?”
Creators see unpredictable algorithmic visibility for long-form and shift effort toward Shorts, reducing supply, which makes discovery even harder.
The home feed stops showing long videos → users don't find them → those who do leave early because content is padded → creators see low completion and poor earnings → they stop making long videos → the feed has even less to show. The four solutions below are designed to break this loop at three points simultaneously.
Primary persona: The Intentional Learner
Who they are
Age 18–34, mobile-first, purpose-driven. Watching to learn or understand a topic in depth, not passive browsing.
Their core frustration
Time wasted finding the right video. They know what they want to learn but can't trust the home screen to surface it. Search is the fallback, not the exception.
Their risk
Switches to an AI tool or exits the app entirely, taking the ad impression, the creator view, and the engagement signal with them.
Four interventions.
Zero new destinations.
04 · Solutions
Solution 1: Deep Dive Tab
Vertical solution, directly addresses P1.
A dedicated tab positioned second in the nav bar, built entirely around intentional long-form consumption. Users declare interest categories, PM, Podcasts, Finance, Football, and the tab organises long-form content around them. No Shorts appear anywhere inside it.
Continue Watching
Pinned strip for videos started but not finished, with a progress bar and resume timestamp.
New from Your Creators
Latest videos from subscribed creators in this category, chronological or ranked.
Deep Picks
Quality-ranked recommendations from creators not yet followed, each with chapter previews and a trust signal like “81% of viewers finish this.”
Creator Spotlight
Rotating weekly to introduce one unfamiliar creator's top three long-form videos.
Key innovation: Deep Picks ranks by completion rate and rewatch signals, not view count. High views is a popularity signal; completion is a quality signal. For long-form, these frequently conflict, and every existing surface optimises for the wrong one.
Solution 2: Long-Form Viewing Streak
Lateral solution, addresses P1 through behaviour change, not a new surface.
A weekly habit mechanic rewarding consistent engagement with long-form content in declared categories. Watch 3 qualifying videos (15+ min, 75% completion) in one topic per week and the streak advances, living inside the Deep Dive tab, framed as a personal progress tracker.
Week 3: Consistent Viewer
Badge unlock + 1 curated recommendation in the declared category.
Week 6: Personalised Summary
Hours watched, topics covered. Recommendations shift from videos to creator playlists.
Week 12: The Pathbreaker
Graduates to broader category playlists, matching format to demonstrated depth.
Key innovation: Three opt-in notifications only, each firing with the specific recommended video or playlist attached, the user never opens YouTube from a streak notification and has to decide what to watch. The pathway is always pre-loaded.
Solution 3: Shorts-to-Long-Form Contextual Bridge
Vertical solution, directly addresses P1, the format-switching gap.
A behaviour-triggered bridge card inside the Shorts feed once a user shows genuine topic interest, after a 95%-complete Short and 3+ same-topic Shorts in one session. Instead of another random Short, a 45–90 second Deep Dive Preview appears in the same vertical format, with a “Watch Full Video” CTA and visible length.
- Trigger: 95% completion + 3+ same-topic Shorts in one session, both must be true, preventing irrelevant cards after casual scrolling.
- Frequency cap: One bridge card per 5 Shorts maximum.
- Learning system: Exit under 60s reduces similar suggestions; 5+ minutes boosts them.
- Risk control: If Shorts session length drops >5% in A/B testing, bridge frequency is reduced before rollout.
Key innovation: Not a same-creator link, every existing bridge mechanism is creator-dependent. This one is system-level, cross-creator, and behaviour-triggered. The surface is intent-matched, not subscription-matched.
Solution 4: Creator Success Suite
Lateral solution, addresses P2 (padding loop) and P3 (creators quitting long-form).
AI-powered tools inside YouTube Studio that reduce the effort, cost, and skill barrier of making long-form content, plus a pre-publish Content Quality Score that penalises padding before a video goes live.
AI Creation Lab
Built-in toolkit: AI script generator, AI editor suggesting cuts and chapter markers, auto-captions in 12 Indian languages plus English, AI thumbnail generator, SEO-optimised titles and descriptions. Access tiered by subscriber count.
Content Quality Score
Runs before publish, scores hook strength, pacing, depth, and engagement potential on a 0–100 scale, with timestamped suggestions like “trim intro from 1:10–2:00.” Not public, not linked to monetisation, purely advisory.
Key innovation: The Content Quality Score is a pre-publish intervention, the only point in the workflow where the platform can influence quality before it damages viewer trust. Every other mechanism responds after the fact; this acts before the video goes live.
How we'll know
it worked.
05 · Metrics & Trade-offs
Primary metrics (North Star)
Solution-level metrics
Deep Dive Tab
- Tab adoption rate (% of MAU visiting at least once)
- Long-form videos started via Deep Dive vs. Home
- Search time per session (target: reduce from 12 min)
Viewing Streak
- Week 3 retention rate of streak participants
- Week 6 completion rate (milestone unlock)
- Weekly long-form sessions per streak user vs. non-streak
Shorts-to-Long Bridge
- Bridge card tap rate (% of shown cards tapped)
- Post-bridge long-form watch time (>5 min threshold)
- Shorts session length impact (must stay within −5%)
Creator Success Suite
- Long-form upload frequency per creator (before/after)
- Average Content Quality Score at publish
- Creator long-form completion rate vs. control group
What this PRD does not solve
- Ad frequency and unskippable ads: the most cited frustration in open-ended responses, but a monetisation decision requiring a separate workstream.
- TV and desktop experience: long-form thrives on CTV (73% of TV watch time), but that only matters once users can discover content on mobile first.
- The Shorts–Long-Form transition rate: the single most important unverified metric in this PRD. YouTube has never disclosed what % of long-form views originate from a Shorts session, treat this as the #1 assumption to test before broader rollout.
The architectural constraint that shaped all four solutions: users will not leave their existing feed for a new destination, proved by Instagram IGTV, Netflix Play Something, and every other bolted-on long-form feature in the competitor graveyard. Every solution here is native to a surface users already visit.