Snapchat Spotlight's 30-second earning floor
Spotlight videos must run 30 seconds or longer to earn. That floor is real for eligible video — not a workaround for wholly AI-generated Spotlight.
Snapchat publishes a duration floor for Spotlight that most video pipelines quietly fail. Per Snap's monetization program page, Spotlight videos must be at least 30 seconds long to earn. Not to be recommended, not to rank well — to earn.
That floor only applies to video that is otherwise in the game. On 31 July 2026 Snap's newsroom stated that wholly AI-generated videos are no longer eligible for recommendation on Spotlight. The recommendation-eligibility quality page adds that ranking rewards human-made content over wholly AI-generated content created outside Snapchat, even when disclosed; content generated inside Snapchat's own tools remains eligible. Stitching four off-platform generations into a 32-second file clears the duration floor and still fails that bar.
The duration number is awkward for a second, narrower reason: it is roughly three times what a single text-to-video generation gives you. That matters for human-shot and hybrid Spotlight — generated inserts, b-roll, a held end card — not as a path to earn on a fully synthetic clip.
The floor is a payout condition, not a ranking preference
This is worth stating plainly because most duration advice is soft. "Longer videos tend to hold better" is a hypothesis. "Videos under 30 seconds do not earn on Spotlight" is a published condition with a hard edge on it. A 28-second Spotlight video that performs beautifully is a 28-second Spotlight video that does not generate revenue.
It also stacks with everything else. The 30-second floor is per-video. The eligibility thresholds that govern the account — 50,000 followers, 15,000 hours of view time over the last 28 days with at least 3,000 of those from Spotlight, being 18 or the age of legal majority, Snap Star status, residence in an eligible country, and the content rules — are separate and account-level. Clearing the account gates and then shipping sub-30-second Spotlight videos means qualifying for a program those videos cannot earn under.
Why one generation almost never clears it
We looked at what the catalog can actually produce in a single pass. Across the 149 active video models on Versely, 125 publish a duration list. Their maximum single-generation lengths break down like this:
| Longest single generation | Models |
|---|---|
| 7 to 10 seconds | 55 |
| 12 to 15 seconds | 50 |
| 20 to 25 seconds | 15 |
| 30 seconds | 5 |
Five models out of 125 can produce 30 seconds in one pass, and those five hit the floor exactly, with no headroom for the trim you will inevitably want at the head or tail. Seedance 2.5 and Grok Imagine are in that group; we wrote separately about what a 30-second single-pass generation actually buys you in terms of continuity. If the whole Snap is that file, it is wholly AI-generated content created outside Snapchat.
For the other 120 models a 30-second piece is an assembly job. That is the design constraint for hybrid Spotlight: human-shot footage plus generated inserts, or several short generated beats cut into a piece that is not wholly synthetic. The question is not "which model goes longest," it is "what structure gets me past 30 seconds without the last 15 feeling like filler."
A four-beat structure that clears the floor
Padding is the obvious failure. Holding on a static frame, slowing the last clip down, or repeating a beat to buy seconds all produce a video that technically clears 30 seconds and demonstrably does not hold. Snap also separately disqualifies Snaps that are minimally distinguishable from one another, so "add ten seconds of the same shot" is not a neutral move.
The structure below is built around the arithmetic for a hybrid or human-shot Spotlight that still has to clear the floor. At the default 25 fps, 30 seconds is 750 frames. Four beats plus a held end card gets you comfortably past that with real content in every second.
| Beat | Target | What it does |
|---|---|---|
| 1. Hook | 3 to 5s | States the claim or shows the anomaly. Nothing else. |
| 2. Setup | 7 to 9s | Establishes what we are looking at and why the hook is true |
| 3. Turn | 7 to 9s | The reversal, the reveal, or the second half of the comparison |
| 4. Payoff | 7 to 9s | Resolves the hook. This is the beat people skip when they run out of budget |
| 5. End card | 2 to 3s | Static or near-static. The only place a hold is honest |
Four beats at the 8-to-10-second ceiling most models offer lands somewhere between 32 and 40 seconds before the end card, before trims and transitions eat frames. Add the card and you want to be trimming down to 32 seconds, not stretching up to 30. Generated beats belong here as inserts or as one layer of a hybrid Snap, not as the entire Spotlight.
Two structural notes that save re-generations:
- Put the weakest beat third, not last. If one of the four comes back soft, you want it buried in the middle where momentum carries it, not sitting at the end where completion is decided. A retention curve that sags at 60% recovers. One that dies at 85% costs you the completion.
- Write the beats before you generate any of them. Generating clip one and then asking what should follow produces four clips that are each fine and do not add up. Write all four as a single sheet first — it is much cheaper to rewrite a beat than to regenerate one.
Assembly is where the cost decision lives
Once you have four clips, getting to one 30-second file is an edit, not a generation.
Versely's editor is EDL-based: one timeline that stays re-renderable. Stitching several clips into one video on that timeline means the assembly can be revised without re-cutting from scratch, and a single charge applies to the final export regardless of how many clips are on the timeline. There is also a free 480p preview pass so you can check the timing before committing to a full render — it carries a short per-user cooldown, so it is a check-your-work tool rather than something to lean on iteratively. The preview and final export mechanics are worth understanding before you build a habit around them.
Build the timing on the timeline, verify the delivered duration at preview resolution, then export once. Do not export four times chasing a duration the preview pass would have shown.
If a beat lands short and you need a couple of seconds rather than a whole new clip, extending an existing video is usually cheaper than regenerating the beat and often reads better than a hold. Extend models publish their own duration ranges — check the range before you assume a two-second pad is in bounds.
Measure delivered duration, not intended duration
The gap that catches people is between the duration you asked for and the duration in the file. A model set to 10 seconds does not always return exactly 10 seconds. Transitions consume frames; a crossfade at each of three cut points can eat a second or more. Head and tail trims to remove a soft first frame take more. Four "10-second" clips can land at 34 seconds or at 29, and only one of those earns.
Read the actual runtime of the exported file before it goes near a scheduler. If Spotlight is a recurring format, saving the assembly as a reusable workflow with the beat targets baked in answers the duration question once at design time.
A floor is a floor. Thirty seconds is where earning starts, not where it optimises. Nothing in Snap's published policy suggests 31 seconds is a target worth hitting precisely — it suggests that below 30, the question does not arise.
FAQ
Does the 30-second floor apply to Snaps posted outside Spotlight?
Snap's published requirement is specific to Spotlight videos. Stories and other surfaces have their own place in the monetization program, and the 15,000-hour account threshold counts view time across surfaces with at least 3,000 hours specifically from Spotlight. The 30-second rule is the Spotlight-specific one.
Can I hit 30 seconds by slowing a shorter clip down?
Mechanically yes, and it is a bad idea. Slow motion on generated footage exposes temporal artefacts that are invisible at normal speed, and the resulting video reads as padded, which is precisely the shape Snap's minimally-distinguishable and editorial-judgment clauses are written to catch. Generating a fourth beat is the better spend.
How many clips does a 30-second Spotlight video usually need?
For a hybrid piece whose generated inserts come from a model that tops out around 8 to 10 seconds, four beats plus a short end card is the reliable shape. If you are on one of the handful of models that reaches 20 seconds in a pass, two beats plus a card also works and gives you fewer cut points to manage. Four generated clips and nothing else is wholly AI-generated content created outside Snapchat, which Snap says is not eligible for Spotlight recommendation.
Does a longer video earn proportionally more?
Snap does not publish a per-second rate, so nobody can answer that with numbers. What the program does measure is total view time over a 28-day window, and view time is a product of views and how long each viewer stays. A 45-second video that people abandon at 12 seconds generates less view time than a 32-second video watched to the end.