How the YouTube Shorts Algorithm Decides, for Founders Cross-Posting

You already post on TikTok, you cross-post the same videos to YouTube Shorts, and the two platforms return different numbers for identical files. The explanation starts with what the Shorts feed measures. Shorts are served in a swipe feed where the next video simply plays, so the system's core signal is what the viewer does in the opening seconds: stay or swipe away. YouTube surfaces this in analytics as the share of viewers who stayed to watch past the initial seconds, and everything downstream, from the next round of distribution to whether your Short earns a search tail, flows from that decision.

The founder diagnostic for Shorts that get no views covers what to do when the number is stuck at zero. This piece is the layer underneath: how the feed picks winners, how that differs from the long-form YouTube you may already know, and what changes for a founder whose Shorts strategy is mostly a TikTok export button.

How does the Shorts feed decide what to serve?

YouTube has answered this on the record. In Creator Insider's Shorts algorithm explainer, the question went directly to Todd Sherman, then the product lead for Shorts, and Sherman's framing is the one worth keeping: the audience is the algorithm. The system's job is to get videos to people who will value them, so it serves your Short to viewers whose past behavior predicts interest and then reads their response as the verdict.

Sherman also explained what a view means on this surface. YouTube wants a view to encode intent, a person deciding to watch rather than a frame flickering past mid-swipe, which is why analytics gives you viewed versus swiped away as its own metric. In a separate Creator Insider conversation with the creator Jenny Hoyos, Sherman went further on the inputs: watch time carries weight as a proxy for whether the viewer valued the video, likes, comments, and shares add signal when there's enough volume, and the stay-or-swipe decision matters precisely because it happens on every single view. He also said the system works to normalize for duration, so a sharp 15-second video isn't structurally outcompeted by a 60-second one. The machine learning figures out how strong each signal is against one defined objective: did people enjoy what they were served.

For a founder the practical reading is short. Your Short gets tested against strangers regardless of your subscriber count, the test is heavily weighted toward the first seconds, and no amount of tagging, timing, or caption strategy overrides an audience that keeps swiping.

How is Shorts distribution different from long-form YouTube?

Long-form YouTube is a marketplace of choices. A viewer sees a thumbnail and a title, and the click is the selection signal, which is why long-form channels obsess over impressions click-through rate. The Shorts feed removes the choosing. The video plays without being picked, so there's no thumbnail to optimize and the hook absorbs the entire job the thumbnail used to do.

The second difference is where Shorts travel. YouTube's own analytics documentation breaks Shorts traffic into the Shorts feed, YouTube search, channel pages, browse features such as the Home feed where the Shorts shelf appears, and external sources. The feed delivers the burst, while search and channel pages deliver the tail. A Short that answers a question people search keeps collecting views long after the feed has moved on, which is a discovery channel TikTok barely offers, and it means titling a Short like a search answer costs nothing and keeps working for as long as people keep asking the question.

The third difference is the floor under your channel. Shorts sit on the same channel as your long-form videos and your subscribers see both, so a founder who eventually wants longform, product explainers, launch videos, a podcast, is compounding one asset. YouTube's analytics even reports viewer overlap across formats, because the company expects the formats to feed each other.

What changes when you cross-post from TikTok?

Mechanically, almost nothing: both platforms run interest-based swipe feeds that test openings against stranger behavior. That overlap is the economic argument for cross-posting at all, since one well-made vertical video gets judged by two independent feeds for the marginal cost of an upload. The accounts we run at JOLT post seven videos a week across all platforms on exactly this logic, and the mechanics of repurposing between TikTok, Reels, and Shorts are worth doing properly rather than by export button.

Properly means three things on the Shorts side. Upload a clean file, because a TikTok watermark bouncing across the frame tells both the viewer and the platform the video is a secondhand repost. Cut the opening for a cold audience, because a hook that leaned on your TikTok context, a running bit, a reply to a comment your Shorts viewers never saw, opens with a debt the new audience won't pay. And give the Short a real title, because search is a Shorts surface and a title written like an answer is free distribution the TikTok version never gets.

The deeper adjustment is patience with divergence. The same video can run on TikTok and stall on Shorts, or the reverse, because each feed is sampling a different audience with different learned interests. Wide swings between uploads are the design working as intended. Judge the system on a run of uploads rather than any single one, and read the stayed-to-watch share before concluding a video failed.

What should founders watch in Shorts analytics?

Skip most of the dashboard. Three reports do the diagnostic work.

Viewed versus swiped away. The percentage of viewers who stayed past the initial seconds. Sherman has described creators comparing this across their own uploads to find which openings grab and which get swiped, and that comparison matters more than any absolute number. Your best Short is your benchmark.

How viewers find your Shorts. If the Shorts feed is your only meaningful source, you're renting all your reach from the feed's mood. Views arriving from YouTube search and channel pages are the compounding kind, and their share should grow as your library does.

Engaged views over raw views. YouTube counts a view the moment a video starts to play, and bases Partner Program earnings on engaged views, the ones where viewers stayed past the initial seconds. The gap between your raw and engaged numbers is the price of your current hooks.

What you're looking for across all three is the pattern we saw run at full scale on the Real American Beer campaign: 90 days of daily content, zero paid media, 15M organic impressions measured across Instagram, Facebook and X, and over $1M in sales, all built on openings that strangers chose to keep watching. The feed mechanics reward the same behavior at founder scale, with smaller numbers and the same shape.

What about the money questions?

A big share of searches around this topic are payout math: how many views for $10,000, what's the rate per thousand. There is no fixed answer, because Shorts monetization runs through a revenue-sharing pool whose per-view value moves with region, niche, and month, and YouTube bases the earnings on engaged views rather than raw plays. For a founder, chasing the payout is a category error. More than 200 founders run our playbook, and the value shows up in their pipeline, their hiring funnels, and the investors who recognize them on the first call. The Shorts check, when it exists at all, rounds to zero next to one closed customer.

The same goes for the 7 second rule people ask about. YouTube has never published a rule with that name. What it has published is the metric that inspired the folklore, the share of viewers who stay past a Short's initial seconds, and what Sherman has said on the record is that grabbing people in the first few seconds decides performance at any video length. The rule is a mnemonic for the metric that does the real grading.

Where to go from here

The feed's verdict is fast and indifferent, which makes it a cheap source of truth about your content. Pull up your recent Shorts, rank them by stayed-to-watch, and rewatch the top one and the bottom one back to back. The difference will almost always live in the first sentence. Rewrite your next opening accordingly, upload a clean file with a searchable title, and hold the cadence long enough for the feed to learn who should see you. If Shorts is one lane of a broader push, the shortform video pillar maps how the platforms fit together, and if your Shorts are stuck at zero, start with the diagnostic instead. The feed grades every upload. Give it something strangers would choose to finish.