Radical Tempo
GuidesLog inSign up

Why did my views drop?

Views down on TikTok, YouTube Shorts or Instagram Reels? Five checks that explain most drops, with real channel data, and how to fix each one.

Joe White · · 8 min read

Short answer

Before blaming the algorithm, check five things in this order: was last month one big hit, has your audience already seen this footage, is the clip on the wrong platform, did the opening lose people, and did the platform change how it counts. The first four cause most drops, and all four are fixable.

The numbers live in four apps and none of them explain themselves

A drop shows up as a smaller number in YouTube Studio, TikTok, Instagram and Facebook, each with its own way of counting. Instagram doesn't show you a retention curve at all. So most creators see the fall, guess at a cause, and change the wrong thing.

And low numbers are normal. About three in four TikTok posts get under 1,000 views. What matters is whether your views fell against your normal, and why.

We ran these checks on three channels: PrankU, a celebrity prank channel; Goombay Kids, a children's show from the Bahamas; and Ignite Talks, a library of five-minute talks going back to 2009. Every number below is from their accounts.

Work through them in order

1. Was last month one big hit?

The most common "drop" is a return to normal after one post took off. Totals hide it, because one post can be almost all of them.

Channel Total views The one hit Everything else
Goombay Kids on YouTube, last 90 days 5.01M 97%: one Short, 4.88M views, 98% watched 94 posts, 134K
Ignite Talks, all platforms, Jul 25 – Aug 7 130.6K 74%: one 78-second clip 13 clips, 33.7K

Take that Ignite clip out and its TikTok views for the fortnight go from 99,065 to 6,094. Nothing broke. One talk from 2009 found a huge audience, and the next fortnight was always going to look like a fall.

What to do: Compare your median post, not your total. If the median held, your channel is fine and you're looking at the hit, not a drop. Then make more of whatever the hit was.

2. Has your audience already seen it?

Repost the same moments, cut the same way, and the people who watched last time scroll past. PrankU's NLE Choppa prank shows how fast this happens: re-cutting the same moments the same way months later got 14× fewer Instagram views (a typical post went from 97K in 2024 to 7K in 2025).

The fix isn't new footage. It's a new way into the same footage. PrankU posted one JoJo Siwa scene to Instagram three times:

Posted Views The cut
Oct 2024 460K Launch cut. Took 11 seconds to reach the accusation.
Mar 2025 6.1K Re-cut told the old way.
Sep 2025 3.13M Opens on the accusation. Title asks a question.

The September cut opens on the accusation ("Your name's Joelle, I take it."), flips the title from victim to suspect ("Did JoJo Siwa steal someone's identity?") and ends unresolved. It became PrankU's biggest JoJo reel on Instagram, eleven months after launch.

What to do: If a repost flopped, don't post it again. Cut a new opening from a different moment, and give it a new title, ideally a question.

3. Is it on the wrong platform?

The same file can do millions on one platform and almost nothing on another, on the same day. If you post one cut everywhere, a "drop" might just be the platform that never suited it. One 27-second NLE Choppa clip, posted the same day:

Platform Views
Instagram 20.22M
TikTok 9.98M
YouTube Shorts 12.4K

Each platform wants something different. On YouTube Shorts, PrankU's winners tell the whole story start to finish and get replayed: a 41-second NLE Short was watched 106% through on average and drew 666K views. On TikTok, the 10M cut stated the emergency in two seconds and stopped at the peak.

What to do: Look at the drop one platform at a time. Then cut a version for each platform rather than posting one file four times.

4. Did the opening lose people?

If views fell but you're posting the same kind of content, look at the first seconds. A slow start gets swiped, and the platform shows the clip to fewer people.

On PrankU's 9:30 NLE Choppa episode, viewers left at 0:17–0:52 and 3:04–3:39, during greetings and small talk. They held from 0:52 to 3:04, from "I'm having contractions", and again from 5:28 to 9:07. On Ignite Talks' YouTube clips, the ones that open straight on the idea were watched more than 72% through. The ones that open on intro music or applause stayed under 62%.

Both channels show the same thing. Viewers leave during set-up and stay once something is happening. The worst Ignite opening started on "this", pointing at a slide nobody could see. It was watched 32% through.

What to do: Find the second your viewers leave, across every recent post, not one video at a time. Cut future clips to start where viewers stayed.

5. Did the platform change something?

Sometimes it really is the platform. But check what actually changed before you assume it. Here's what happened in the last 18 months, and what didn't.

  • 31 Mar 2025: YouTube started counting a Shorts view on any play or replay. Totals rose about 20–30% with no change in audience, so a comparison across that date can look like a drop or a jump that isn't real.
  • Jan 2026: TikTok's new US joint venture said it would retrain the algorithm on US data. TikTok told NPR on 4 Feb that no algorithm changes had been made, and as of October there was no public evidence of a switch.
  • Late Jan, 3–4 Mar 2026: TikTok outages caused posting delays and view-count problems. A drop in those weeks is an outage, not a ranking change.
  • Mar 2026: Facebook added "qualified views" to show which views count toward earnings.
  • 30 Apr 2026: Instagram stopped recommending accounts that mostly repost other people's content, and said speed changes or credit overlays don't count as making it new.

What to do: Check whether your drop lines up with one of these dates, and whether it shows on every platform or just one. If it's only one platform and no change lines up, go back to checks 1 to 4.

Run all five checks without opening four apps

You can do every check above by hand. It takes a spreadsheet, four dashboards and a lot of scrubbing through retention graphs. Radical Tempo, an AI video platform that clips, publishes and measures short-form video, does it from your connected accounts and then acts on what it finds.

Check By hand With Radical Tempo
1. One big hit? Export each platform, sort, work out medians Ranks every post across all platforms and shows totals and medians side by side
2. Already seen? Remember which clips came from which footage Groups every clip with the video it came from, so you see each footage's history
3. Wrong platform? Match the same clip across four apps Compares the same clip platform by platform, and against similar posts on each
4. Weak opening? Watch retention graphs one video at a time Finds where viewers leave across all your posts, and ranks each against similar-length videos
5. Platform change? Read the news and guess Shows whether the drop is on one platform or all of them, from the same data

Then it closes the loop. Radical Tempo cuts the next clips from the moments where viewers stayed, writes a post for each platform, and checks the results the following week. What it learns about your channel goes into a playbook it uses on every clip after that. That's how PrankU's average views per post rose about 3× between the first and second half of 2025.

It compares you against more than your own history too. Radical Tempo holds about 10 billion views of performance data, so it can tell you whether a clip is weak for your channel or weak for anyone posting that length of video.

Or just ask why

Connect Radical Tempo to Claude and you can ask the question in plain English. Claude does the reasoning. Radical Tempo supplies what Claude can't see on its own: your posts on every platform, where viewers left each one, and what worked for similar creators.

You: Our TikTok views are way down this fortnight. What happened?

Claude: Your channel didn't drop. Last fortnight was one big hit. "The Hoodoo Trap", a 78-second cut of a 2009 talk, did 96,888 views, 74% of everything you got. Without it, TikTok did 6,094 views across 13 clips, which is your normal.

One thing to fix: clips that open on the podcast intro or applause finished under 62% on YouTube. Clips that start straight on the idea finished over 72%. Want me to cut five new clips from your archive that open on the first concrete idea, and show you them before anything posts?

Illustrative conversation. Numbers are Ignite Talks' real results for 25 Jul – 7 Aug 2026.

Prompts to try

Check 1

Using Radical Tempo, compare my median views per post this month with last month on each platform. Was last month one big hit?

Check 2

Using Radical Tempo, list my clips from the last 60 days that came from footage I had already posted, and how each did against the first time.

Check 3

Using Radical Tempo, find clips I posted to more than one platform and show where each one did best and worst.

Check 4

Using Radical Tempo, find where viewers drop off across my last 20 videos and what the clips that held them have in common.

Setting it up takes about two minutes: add Radical Tempo as a connector in Claude with this address, then sign in.

https://mcp.radicaltempo.com/mcp

Step-by-step instructions are in our guide to using Claude as your social media manager.

Sources

Frequently asked questions

Did the TikTok algorithm change in 2026?

TikTok's new US joint venture said in January 2026 that it would retrain the recommendation algorithm on US data. TikTok told NPR on 4 February 2026 that no algorithm changes had been made, and as of October 2026 there was no public evidence of a switch. Outages in late January and on 3–4 March 2026 did cause posting delays and view-count problems.

Why did my YouTube Shorts views drop?

The usual causes are a single outlier in the previous month, an opening that loses viewers in the first seconds, or a Short that tells only part of the story. YouTube also changed how Shorts views are counted on 31 March 2025, so comparisons across that date mislead. PrankU's best Shorts tell the whole story and get replayed: one 41-second Short was watched 106% through on average and drew 666K views.

Why are my Instagram Reels views down?

Since 30 April 2026, Instagram has stopped recommending accounts that mostly repost other people's content, and it says speed changes or credit overlays don't count as new. Also check whether your audience has already seen the footage: PrankU's same-way re-cuts got 14× fewer Instagram views months later, while a new opening on the same JoJo Siwa scene got 3.13M.

Does reposting the same clip hurt views?

Re-cutting the same moments the same way usually does worse, because the audience has seen it. A proven hit reposted to the same platform months later with a new title can still work: PrankU reposted a Zedd clip four months after launch through Radical Tempo and it did 1.79M views on Instagram, more than its own launch post (1.12M).

How do I find where viewers stop watching?

YouTube Studio shows a retention curve for each video and TikTok shows partial retention, but Instagram shows none. Radical Tempo finds the drop-off moments across all of your posts at once and ties each one to what was said and shown at that second, so you can cut future clips from the moments that held viewers.

How many views is normal for a TikTok post?

About three in four TikTok posts get under 1,000 views, according to Gigapay's 2026 creator economy report. Compare a post against your own median and against similar posts, not a fixed number.