Published 29 September 2026 · Figures measured 15 September 2026 · Every figure below carries the number of Shorts behind it
The field gives two answers, and they contradict each other
Search this question and you land in the middle of an argument. We read these pages on 29 September 2026.
One camp says a Short keeps earning, and has by far the biggest dataset. socialinsider.io, published 27 January 2026, reports interval view counts from "the analysis of 3M YouTube Shorts, coming from 56K accounts of all sizes and regions active from January to December 2025." Its figures run 833 views in days 1 to 5, 1,786 in days 5 to 10, then 150, 555, 500, and 3,310 views in the 25 to 30 day interval. Its conclusion is that Shorts get their "biggest spike nearly a month after they go live."
The other camp says the opposite, and it is about very large channels. In December 2025, tubefilter.com and searchenginejournal.com both reported what analysts had nicknamed "the flattening": Shorts older than roughly 28 to 30 days appearing to lose reach. It was raised by retention strategist Mario Joos, corroborated by creator Tim Chesney, and drawn from "channels ranging from 100 million to one billion monthly views." YouTube has not confirmed it, and Joos himself called it "a carefully constructed working theory and not a confirmed fact."
Notice what those two have in common before you notice that they disagree. One is an average over three million Shorts from accounts of every size. The other is a handful of channels doing a hundred million views a month. Neither of them is a measurement of what happens to a Short that got 200 views. That is the gap this page is in.
What our catalogue actually did, week by week
We hold repeated view snapshots per Short, so consecutive readings give us how many views arrived in between. Here is that, pooled, for every Short in the base with at least 100 views and at least two readings: 310 Shorts, measured 15 September 2026.
| Age of the Short | Intervals | Shorts | Gained anything | Views per Short per day |
|---|---|---|---|---|
| 0 to 7 days | 299 | 93 | 79.3% | 405.52 |
| 7 to 14 days | 558 | 113 | 37.5% | 28.95 |
| 14 to 21 days | 517 | 111 | 19.9% | 0.54 |
| 21 to 30 days | 399 | 77 | 15.5% | 0.46 |
| 30 to 45 days | 68 | 52 | 39.7% | 0.53 |
| 45 to 60 days | 86 | 56 | 36.0% | 0.21 |
| 60 to 90 days | 60 | 34 | 56.7% | 0.46 |
| 90 to 400 days | 112 | 59 | 17.0% | 0.28 |
405 views per Short-day in the first week. 29 in the second. Half a view a day in the third. Whatever a Short on these channels was going to become, it had become it by around day 14, and by day 21 the figure is close enough to zero that the rest of the table is a flat line with noise on it.
One column needs explaining before anyone reads a recovery into it. The "gained anything" column climbs after day 30, and that is an artefact of how we collect, not a rebound. Our sync refreshes a Short on every pass while it is under 30 days old and weekly after that. Longer gaps mean more days inside each interval, and more days means more chance of catching a single view. The per-day column corrects for exactly that, and it does not climb. We are pointing this out rather than dropping the column, because a reader who found it on their own would be right to wonder what else was tidied.
The same question measured a second way
Pooling intervals from different Shorts has a weakness: a handful of busy Shorts can dominate it. So the same question was asked per Short instead. For each one, what share of its total views arrived in the last fortnight we observed, and how old was it at that point? That pass covers 164 Shorts across 10 channels, and it is the one we lean on.
| Age at last reading | Shorts | Median share of its views from the last fortnight | Gained nothing at all |
|---|---|---|---|
| 14 to 21 days | 14 | 14.48% | 7% |
| 21 to 30 days | 28 | 0.49% | 14% |
| 30 to 45 days | 8 | 0.11% | 38% |
| 45 to 60 days | 37 | 0.00% | 68% |
| 60 to 90 days | 23 | 0.06% | 30% |
| 90 to 180 days | 54 | 0.00% | 76% |
The middle Short in our catalogue took 0.49% of its lifetime views in the fortnight before day 30, and 0.11% in the fortnight after it (n = 28 and n = 8). From day 45 onwards the middle answer is zero.
Two honest notes on that table. The 30 to 45 day row rests on 8 Shorts and should not be asked to carry anything on its own. And the "gained nothing" column is not monotonic: it runs 68% at 45 to 60 days, drops to 30% at 60 to 90, then rises to 76%. On 37, 23 and 54 Shorts from different channels, that is the kind of wobble small samples produce, and we are not going to build a story on the dip.
There is an obvious objection to all of this: a Short with 120 views that gains one view reads as 0.8%, so tiny Shorts could be manufacturing the result. So the same pass was re-run on only the Shorts above 1,000 views, where a single view cannot move a percentage. Same shape: 3.83% at 14 to 21 days (n = 6), 0.17% at 21 to 30 days (n = 16), 0.17% at 30 to 45 days (n = 4) and 0.00% at 45 to 60 days (n = 27). Those are small numbers of Shorts and we are printing them with their sizes rather than quoting the percentages on their own.
We went looking for the late spike on purpose
A collapsing average could still hide the thing everyone is actually asking about: the one Short that sits quiet for three weeks and then takes off. So we hunted for it directly, in the subset where we can see both sides of the line. 26 Shorts were observed both at or before day 21 and again after day 30.
- 2 of those 26 gained 20% or more of their views after day 21. One went from 111 views to 158. The other went from 66 to 106. Both are on a channel we run ourselves.
- The largest gain after day 21 anywhere we can measure is 61 views, on a customer Short that had 1,973 views at the time (n = 26).
So late movement is not zero. It is real, and on our catalogue it is measured in tens of views. A Short that stalled at 200 and came back to life as a Short with 20,000 views is not something we have observed once.
So is the much bigger study wrong? No, and this is the important part
It would be easy, and cheap, to put our 310 Shorts next to their three million and imply we have corrected them. We have not, we cannot, and we are not claiming to. Their dataset is roughly ten thousand times the size of ours and it covers accounts of every size, which ours does not.
The resolution is that the two numbers are probably not answering the same question.
First, the statistic, which the article does not state. Its figures are one view number per interval, with no note saying whether that number is a mean or a median, and no inclusion criteria published either. It matters enormously. A mean over three million Shorts is very easy for a small number of very large Shorts to dominate: if a few thousand Shorts in that dataset were picking up six-figure view counts in week four, the interval mean would read 3,310 while the typical Short in the same dataset sat at zero. Our figures are medians, which are built not to do that. So if theirs are means, the two results are compatible and nobody is wrong. If theirs are medians, they genuinely disagree with us, and we would want to know that. We cannot tell which, and we are not going to assume the reading that happens to suit us.
Second, and this one holds whichever way the first goes: the populations are different, and not slightly. Their own figures sum to roughly 2,600 views in a Short's first ten days. On our side, the median ShortLens customer's best Short ever, of any length, sits at about 1,200 views (measured across 631 Shorts on 28 customer channels, 9 to 12 September 2026). So either the average Short in their dataset does more in ten days than our typical customer's best Short has ever done, in which case these are plainly not the same creators, or that 2,600 is itself being carried by a small number of very large Shorts, which is the first explanation arriving a second time. There is no version of this where their dataset is describing a channel like ours.
Which is why we would put it like this. Not "Shorts stop earning after two weeks." Something narrower and, we think, more useful: on a small channel, a Short that peaked in the hundreds or the low thousands is finished well before day 30, and the 25-to-30-day argument is happening in a window where a Short like that has already stopped moving. Whether a Short that reached six figures keeps earning is a real question and it is not one we can answer.
Three limits, and the one observation that would end this page
We cannot test the large-Short case at all. Roughly 1 Short in 100 in this catalogue has ever passed 10,000 views. We hold exactly one Short that passed 100,000, it took 178,003, it is on a channel we run ourselves, and it was already 24 days old the first time we observed it, which is the wrong side of the line this page is about. So we have no evidence about big Shorts, in either direction, and nothing here should be read as extending to them.
This finding is one counter-example from revision, and saying so is the point. A single Short anywhere in our catalogue gaining 1,000 or more views after day 30 would end this conclusion. In 164 observed Shorts it has not happened yet. If it does, we will say so here rather than quietly leaving the page up.
The oldest and newest buckets are thin. The 30 to 45 day row of the second table is 8 Shorts, the 14 to 21 day row is 14, and the late-spike hunt rests on 26. The rows carrying the conclusion are the ones with 28, 37 and 54 Shorts in them, and the 310-Short pooled table. Read the thin rows as consistent with the rest rather than as evidence on their own.
And the window that decides everything is the one we cannot see. Our earliest reading of any Short is around day 3, and the analytics data we are given lags what a creator sees in YouTube Studio by several days on top of that: on one Short on 10 September 2026 we read 96 views where Studio showed 1,942. Every claim about what happens in a Short's first 48 hours describes a window our data starts after. Worth noting that this particular limit runs against our own conclusion rather than for it. A lag pushes early views into later buckets, which makes late accrual look larger than it really is, and we still measured late accrual at roughly nothing.
One thing YouTube announced, at exactly the level the announcement supports
This is going to come up, so here it is with the source attached. On 23 September 2026, in its Made on YouTube announcements, YouTube described a Studio assistant that looks at your older uploads. The primary announcement on blog.youtube says this, in full:
"Ask Studio can even run in the background to review your older videos to suggest and test refreshed thumbnails, helping classic content reach new audiences."
That sentence is worth reading closely, because a good deal of the coverage around it is broader than it is. It covers older videos, thumbnails, and suggesting and testing. It does not say Shorts, it does not say titles, and it does not say anything is applied on your behalf. We have quoted the whole of what the source says on the point rather than the summaries of it.
We have not tested this feature and we hold no observations of it. It was announced after every figure on this page was measured. We are not going to tell you whether it works, in either direction. What our own data says is a separate and narrower thing: on the small-channel Shorts we measured, there was very little still arriving after day 14 for any refresh to catch. That is a description of what we observed on our catalogue. It is not a prediction about a product.
Run this on your own catalogue in about ten minutes of actual work
The measurement in the second table needs no tool and no access to anything you do not already have. It needs two weeks of patience.
- List your Shorts that are already older than 30 days, with today's view count against each one. YouTube Studio gives you this on the Content tab.
- Write down today's date and leave them alone. No edits, no new descriptions, nothing.
- Two weeks later, record the same view counts again.
- For each Short, divide the new views by its total. That is the share of its life arriving in a fortnight, at its age, on your channel.
- Take the middle value. That is your answer, and it is the only version of this number that describes you.
On our 164 Shorts across 10 channels the middle answer was 0.00% from day 45 onwards. Yours may be different, and if it is, yours is the one that matters. Do this on Shorts only. Nothing on this page was measured on long-form video, and we would not assume it transfers.
What this page does not say
- Not "YouTube stops showing your Short after two weeks." We measured views arriving. We have no view into what the platform decides, we cannot observe it, and we are not going to guess at it. Every sentence here is about view counts we can see.
- Not "old Shorts never get anything." 2 of the 26 Shorts we could observe on both sides of day 21 gained a fifth of their views afterwards. The amounts were 47 views and 40 views.
- Not a correction of the larger study. Three million Shorts from 56,000 accounts is a better dataset than ours by every measure except one: it does not tell a creator with 200 views what happened to Shorts like theirs.
- Not "delete it and reupload it", and not "leave it up and wait" either. We have never measured a reupload. What we can say is only that waiting, on this catalogue, did not pay.
- Nothing about Shorts that reached six figures. See the limits above. We hold one and we started watching it too late.
And the thing we genuinely cannot see
The metric most serious accounts of Shorts distribution point at is viewed versus swiped away, the share of people who were shown the Short and stayed rather than flicking past. It is visible in YouTube Studio and it is not in the data a creator can authorise us to read, and there is no honest proxy for it. It matters here specifically: a Short that stopped earning on day 6 and a Short that was barely shown to anyone look similar in a view count and are different problems. We will not publish an estimate of it and we will not tell you whether you clear anybody's threshold for it. We explain what happened to the people who started. Studio tells you how many never did.
Questions people ask about this
Taken from public creator forums and search results in September 2026, close to the wording they were asked in. Every answer carries the number of Shorts behind it, and where the honest answer is that we do not know, that is the answer.
My Short stopped getting views after two days. Is that normal?
On the channels we measure, that is close to the ordinary shape rather than a fault. Across 310 Shorts with at least 100 views, measured 15 September 2026, views arrived at 405.52 per Short per day in the first week, 28.95 in the second, and 0.54 in the third. The drop is not a cliff at some particular hour, it is a steep slope through the first fortnight.
What we cannot tell you is why yours stopped where it did. We hold no view into what the platform decides about any individual Short, and the window where that gets decided is the one window our data starts after: our earliest reading of any Short is around day 3.
The thing this does answer is narrower and still worth something. If you have been refreshing a two-week-old Short waiting for it to move, our data says the movement you are waiting for is not something we have observed at any size that would change your week.
Do YouTube Shorts still get views after 30 days?
On our catalogue, technically yes and practically no. The middle Short in our second pass took 0.11% of its lifetime views in the fortnight after day 30 (n = 8 Shorts, which is thin), and from day 45 onward the middle answer is 0.00% (n = 37 at 45 to 60 days, n = 54 at 90 to 180 days), measured 15 September 2026 across 164 Shorts on 10 channels.
There is a widely-reported claim that Shorts older than 28 to 30 days are being deprioritised. We cannot adjudicate it, and the reason is more useful than a verdict would have been: on our catalogue there is nothing left at day 30 to deprioritise. Accumulation has already ended around day 14. A change acting at day 28 would be acting on a quantity that is already rounding to zero for a small-channel Short. Note also that the reports of that shift come from channels doing 100 million to one billion monthly views, which is not the population we measure.
Can a Short go viral weeks after I posted it?
Not on anything we have observed, and we looked for it specifically rather than inferring it from an average. Of the 26 Shorts we could observe both at or before day 21 and again after day 30, 2 gained a fifth or more of their views afterwards, and in absolute terms those gains were 47 views and 40 views. The largest late gain anywhere we can measure is 61 views.
Two scope limits on that, both real. It rests on 26 Shorts, which is not many. And 1 Short in 100 in this catalogue has ever passed 10,000 views, so if late virality is something that happens to Shorts which already travelled, we have almost none of those to watch it happen to. The one Short we hold above 100,000 views was already 24 days old when we started observing it.
So the honest answer is: on small channels, delayed virality is not something our data shows, and our data is not able to rule it out on big ones.
Should I delete a Short that flopped and reupload it?
We have never measured a reupload and we are not going to advise you on one. That would need the same Short posted twice under comparable conditions, and we hold no such pairs.
We can speak to one half of the advice around this question. A common reason given for leaving a flop up is that Shorts resurface later. On our data that is the part that did not hold: across 164 Shorts on 10 channels, the middle Short gained 0.00% of its views in the fortnight after day 45, and the largest late gain anywhere we can measure is 61 views (n = 26). Keeping it up costs you nothing, and our measurement says it is unlikely to pay you anything either.
The one thing we would say without hedging is where the remaining upside sits. On this catalogue, a Short's outcome was settled inside its first fortnight, so the next upload is the one still capable of changing a week.
Three questions this page deliberately does not answer. What a reading above 100% actually means, and what it predicted, has its own measurement and its own page: is over 100% retention good on a YouTube Short? What average percentage viewed is normal for a Short of a given length has a different table behind it: what is a good average view percentage on a YouTube Short? Why a Short with better retention lost to one with worse retention is a wider measurement than anything here and is not about a Short's age at all.
How this was measured
Why we looked at this at all
ShortLens reads the measured retention curve of each of your Shorts and tells you where viewers left, to the second. This analysis is a large part of why everything the product says is pointed at your next upload rather than at reviving an old one. A tool that told a creator to go back and rework a three-week-old Short would be sending them after an audience that, on our own measurements, has already stopped arriving.