YouTube algorithm update and video length: why ultra-long fell

For a hot minute, the meta was simple: make it longer, get paid more, build a cult, win YouTube.

Then the recommendation faucet moved. And a bunch of creators learned (the hard way) that "format strategy" is not a strategy. It's a bet.

What happened

A major YouTube channel operator (Little Dot Studios) analyzed multi-year performance across 800+ channels it manages (930M subscribers total, 11.2B monthly views). Their headline: audiences still follow human-led storytelling and fandoms... but YouTube's incentives around length have been wildly unstable.

In 2023-2024, ultra-long videos (120+ minutes, including 3+ hour uploads) punched above their weight. In their dataset, a 120-minute upload averaged about 100x the revenue of a 20-minute one. Ultra-long also skewed toward more watch time and revenue coming from YouTube Premium viewers (ad-free people who can actually survive a three-hour binge without throwing the remote).

That wave peaked around mid-2024: the ultra-long format was pulling roughly 30M views per month from May through November 2024. Then the air came out.

From December 2024 to May 2025, the average length of the platform's most popular videos dropped about 21% (roughly 35 minutes down to ~28). Little Dot ties that shift to a "significant" recommendation update around January 2025, claiming ultra-long viewing fell over 90% within three months, and that the average length of suggested videos effectively halved (around 80 minutes down to 40). By later 2025, they saw views per video normalize across short-form, long-form, and ultra-long. (Translation: format advantage evaporated, then the ecosystem kind of... rebalanced.)

Meanwhile, YouTube itself keeps pouring fuel on both ends of the fire: TV viewing and Shorts. In YouTube CEO Neal Mohan's 2026 letter, YouTube says Shorts average 200B daily views, and that YouTube has been #1 in U.S. streaming watchtime for nearly three years. ([blog.youtube](https://blog.youtube/inside-youtube/the-future-of-youtube-2026/))

And the money behind all this is not subtle. Alphabet's Q4 2025 results (reported Feb 4, 2026) put YouTube annual revenue at over $60B across ads and subscriptions. ([theverge.com](https://www.theverge.com/news/874161/google-400-billion-revenue-q4-2025-earnings))

Subscriptions are also getting "lighter" (and more mainstream). YouTube rolled out Premium Lite in the U.S. in March 2025 at $7.99/month, and noted YouTube Music + Premium had passed 125M subscribers globally (including trials). ([theverge.com](https://www.theverge.com/news/624598/youtube-premium-lite-announced-us))

One more "quiet" shift creators keep noticing: the home feed. Some ad/creator analysts documented late-2025 UI changes that reduced long-form slots on the homepage and gave more space to Shorts, which can make long-form discovery feel like trying to get booked at a club that replaced half the stage with a vending machine. ([ppc.land](https://ppc.land/youtubes-home-feed-quietly-kills-long-form-video-discovery/?utm_source=openai))

Big picture context: we're in peak content saturation. Omdia research said YouTube hit about 29B videos as of Dec 2025, with Shorts representing over 90% of new uploads. ([omdia.tech.informa.com](https://omdia.tech.informa.com/pr/2026/jan/youtube-reaches-29-billion-videos-as-music-and-shorts-dominate-viewing?utm_source=openai))

Why creators should care

Attention: If YouTube is nudging viewers toward shorter sessions (or at least shorter "entry points"), your 2-3 hour masterpiece can still win... but it may not get as many "first dates" from Browse/Home as it used to. That changes how you package the idea, where you put the payoff, and whether you need a shorter on-ramp.

Distribution: Discovery is increasingly algorithmic, not subscription-led. So if the platform starts recommending 40-minute videos instead of 80-minute ones, your content doesn't become "bad." It becomes harder to be surfaced unless you adapt the wrapper (topic framing, structure, episodic cuts, titles/thumbnails, chaptering).

Monetization: Premium-heavy watch time is a different beast. More paid viewers (and more ad-avoidance options like Premium Lite) can shift where your revenue actually comes from. If your whole business assumes "more midrolls = more money," you're building on sand.

Workflow: Ultra-long only works if you can produce it without dying. The winners aren't the people who can make one legendary 3-hour video. It's the people who can build a repeatable system: research, scripting, editing, release cadence, and smart recycling into Shorts/clips without turning into a content mill.

What to do next

• Build a "two-door" version of your idea. Keep the long video if it's your brand - just add a shorter entry product: a 10-20 minute cut, a "Part 1," or a tight companion video that funnels people into the big one.

• Audit where your views actually come from. If Browse/Home is shrinking for you but Suggested is stable, that's a packaging problem. If Suggested is down too, that's usually an audience-match or retention problem. Don't guess - track it weekly.

• Optimize for "living room patience." TV watch is real, and TV viewers tolerate different pacing. Use chapters, visual resets, and recurring segments. Make it feel like a show, not a lecture.

• Stop betting the company on one upload shape. Do one season of ultra-long. Then one season of tighter long-form. Keep Shorts as the distribution layer, not the whole identity. Your goal is adaptability, not righteousness.