3,765
Scripts analyzed
2.25x
Higher 1M+ hit rate for clean videos
12.6x
Max-view gap (69.8M clean vs 5.5M profane)
500
Profane scripts in sample
There's a piece of creator advice that just refuses to die: "be authentic, swear if you swear in real life, the algorithm rewards realness." It's wrong. The platforms quietly throttle profane content in ways most creators don't realize because the throttling is invisible at the median.
We pulled the transcripts of 3,765 short-form videos across TikTok, Instagram, and YouTube Shorts. We tagged each one for the presence of profanity (fuck, shit, damn, hell, ass/asshole, bitch, crap, bullshit, piss). Then we compared the view distributions.
The median is the same. The ceiling is not.
The ceiling, not the median
Profane and clean videos have nearly identical median views (17,386 vs 15,827). The throttling doesn't show up at the typical post. It shows up at the top: clean videos hit 1M+ at 2.25x the rate of profane videos (4.96% vs 2.20%), and the maximum clean video in our dataset (69.8M views) is 12.6x larger than the maximum profane video (5.5M). Profanity doesn't kill your typical post. It caps your viral one.
Here's the full study.
Finding 1: The 1M+ rate gap
Profane and clean videos look almost identical in the long tail. They diverge sharply as you climb into the viral tier.
% of videos crossing 1M views
With profanity
2.20%
Clean
4.96%
Max views in sample
With profanity
5.5M
Clean
69.8M
Average views per video
With profanity
119K
Clean
363K
The median view count is roughly the same (clean videos median 15,827, profane videos median 17,386, actually slightly higher for profane). That's where the "swearing doesn't hurt me" creator anecdotes come from. Most posts perform similarly whether you swear or not.
But the average tells a different story. Clean videos average 3x more views than profane videos (363K vs 119K). The reason: a small number of clean videos break out and pull the mean up. Almost no profane videos break out.
Finding 2: The word-by-word ceiling
Not all profanity is equal. We tagged each script by which specific word appeared and ranked them by median views.
Median views by which profane word appears (min 25 videos per word)
- 01
hell
123 videos · mildest, highest median
56.2K - 02
damn / goddamn
90 videos
53.9K - 03
ass / asshole
38 videos
39.6K - 04
shit / bullshit
165 videos
18.3K - 05
fuck (any form)
235 videos · biggest sample, lowest median
12.4K - 06
piss(ed)
32 videos
8.6K
The hierarchy is clean. Mild profanity (hell, damn) is tolerated by the algorithms. Moderate profanity (ass, shit) gets throttled meaningfully. Strong profanity (fuck) gets throttled hardest. F-bombs in your script cut your median by 4.5x vs videos with only "hell" in them (12,421 vs 56,192).
Two notes on the data:
'Hell' is barely profanity. The high median for 'hell' (56,192) is partly because phrases like 'what the hell' and 'hell of a' are colloquial and platform classifiers ignore them. If you treat hell + damn as 'soft profanity', this category is roughly clean-equivalent.
'Fuck' is 5x more sampled than 'piss' (235 videos vs 32) which gives more confidence in the f-bomb finding. The piss/fuck ranking is reversed if you weight by sample size.
The word-strength rule
The harsher the word, the lower the ceiling. F-bombs cap median views around 12K. Mild profanity (damn, hell) doesn't measurably hurt. Most "swearing is authentic" arguments don't survive the data when you split by which words are actually being used.
Finding 3: The mega-viral exceptions
Out of 500 profane videos in our sample, only 11 crossed 1M views (2.20%). They share a clear pattern.
Alright so everybody's heard of Italian brainrot, right? Pretty much it's taken over the world.
@evantheguardian · 5.5M views
The single biggest profane mega-viral. Profanity is incidental: appears mid-script, not in the hook. The video is a viral pop-culture commentary that scaled despite the profanity, not because of it.
FanDuel is running one of the biggest scam companies in the entire world...
@ev_handd · 3.3M views
Outrage-driven exposé. Profanity comes later in the script as the ranting intensifies. The hook is clean and curiosity-driven.
Cristiano Ronaldo just reminded everybody how poor you are.
@evhandagain · 2.2M views
Mild profanity ('bro') in on-screen text only. Sports drama topic with cultural permission for casual language.
The three traits the profane mega-virals share:
Profanity is INCIDENTAL, not in the hook. The first 3 seconds of every profane mega-viral in our sample are clean. The swearing happens later, in the ranting/payoff zone where the algorithm has already committed to amplifying the video.
The word is MILD when it does appear. 'Damn', 'hell', 'ass' show up. 'Fuck' and 'shit' rarely make the mega-viral cut, and when they do, they're spelled out only in the audio (not the captions or on-screen text).
The TOPIC has cultural permission for casual language. Sports drama, breaking news commentary, viral pop culture moments. Niches where strangers expect frank language. Generic productivity or business content with profanity gets crushed.
If you're going to swear in a video and want it to break out, here's the unspoken rule: clean hook, mild words later in the script, high-velocity topic, audio-only profanity. Profane hooks ("F**k this advice") have a near-zero mega-viral rate.
Stop guessing which words are throttling your reach.
Connect TikTok or Instagram. We tag every word of every video on your account, flag the ones platform algorithms throttle, and write a 30-day script calendar in language patterns that travel safely across feeds.
47,598 creators·No credit card required·60 seconds
Finding 4: Why this happens
The throttling is real, the throttling is silent, and the throttling has two structural causes.
Cause 1: ad revenue protection. Advertisers refuse to run ads adjacent to profane content. Major brands have explicit "brand safety" thresholds in their ad platforms (TikTok Ads Manager, Meta Ads Manager, Google Ads). When a video contains f-bombs, the platform classifies it as "limited monetization" and reduces its distribution to protect the ad inventory. Your video stays up. The For You Page just stops sending it to as many people.
Cause 2: brand-safety classifiers for younger users and restricted mode. Both TikTok and Instagram have "restricted mode" for younger accounts. Profane videos are excluded from those feeds entirely. That's a structural cap on your audience size: you literally cannot reach the under-18 segment of the platform if your script contains strong profanity.
The combined effect: a profane video can still go viral, but it has to overcome a 30-50% reach reduction the algorithms apply silently. Most don't make it.
How to use this
If your goal is reach, default to clean scripts. The 12.6x max-views gap is the clearest finding in our profanity data. The reach ceiling on profane content is real even if the median performance looks similar.
If you do swear, swear softly. 'Damn' and 'hell' are functionally clean by algorithm standards. F-bombs and s-bombs are the strongly-throttled words.
Keep the hook clean even if the body has profanity. Every profane mega-viral in our dataset opens clean. The first 3 seconds of your video are where the algorithm decides whether to amplify it.
Bleep audio profanity and asterisk-substitute caption profanity. We don't have direct comparison data, but auto-moderation classifiers flag spelled-out profanity more aggressively than 's**t' or bleeped audio. If you're committed to the take, defensively obscure the word.
Don't optimize for 'authenticity' if reach is the goal. The 'be yourself, swear naturally' advice produces long-tail-tier videos. Authenticity is a long-tail strategy. Reach requires self-editing.
Save profanity for niches that have permission. Sports drama, news commentary, viral pop culture moments. In niches where casual language is expected (most career, productivity, finance, health content), profanity is a measurable reach penalty.
The bottom line
The "swearing is fine, the algorithm rewards realness" advice is folklore. The platforms have quietly built profanity-throttling systems into their distribution algorithms. Those systems don't kill your typical video. They just cap your ceiling.
If you're posting weekly content and most of your videos sit in the 5K-50K view range, profanity probably doesn't matter much for you. Your ceiling and the algorithm's profanity ceiling don't intersect. But if you're trying to break out into the 100K-1M range, profanity is one of the cleanest single-variable changes you can make. Cut the f-bombs from your hooks and your ceiling goes up.
The bigger creator brands figured this out years ago. Watch their TikToks: bleeped audio, asterisk captions, mild words at most. The reach reflects that discipline.
Get a script playbook tuned to platform algorithms.
TCL audits your account plus your top competitors, flags every algorithm-throttled phrase, and writes 30 scripts in language patterns that travel safely across feeds.
47,598 creators·No credit card required·60 seconds
Methodology
Dataset: 3,765 short-form video transcripts drawn from our analyzed video corpus on 2026-04-28 (combining first-party analyzed content and competitor-scrape analyzed content). Platforms: TikTok, Instagram, YouTube. The 3,765 sample is the subset of our 10,718-video corpus with full transcripts populated. Videos without transcripts (sound-off skits, music-only edits, older un-transcribed content) are not included.
Profanity detection: Whole-word, case-insensitive regex match against the following word groups: f-bombs (fuck/fucked/fucking/fuckin/fucker), s-bombs (shit/bullshit/shitty), damn (damn/damnit/dammit/goddamn), hell, ass (ass/asshole/dumbass), bitch (bitch/bitches/bitchin), crap, piss (piss/pissed). A transcript is "profane" if at least one whole-word match occurs across any of those groups. The classification is loose; mild expletives like 'hell' qualify, even though many platforms don't actually treat them as profanity for moderation purposes.
Per-word breakdown: A video is counted in a category if its transcript contains any whole-word match for that category's word group. A single video can appear in multiple categories (e.g., a script that contains both "shit" and "fuck"). The per-word numbers are not mutually exclusive.
Why median + max + 1M+ rate, not just one metric: The headline finding (clean and profane perform similarly at the median) requires reporting the median to be honest. The ceiling finding requires the max and the 1M+ rate. Reporting only one of these would mislead.
Known limits:
- Transcripts are auto-generated and have minor accuracy issues. Mishears that flip clean → profane (e.g., "duck" → "fuck") are rare but possible. Mishears in the other direction are also possible.
- We don't have direct data on bleeped audio vs spelled-out profanity vs asterisk-substituted text. The "bleep your audio" advice is inferred from platform behavior, not directly tested in this study.
- Cross-posted videos appear once per platform. A video posted to both TikTok and Instagram counts twice.
- The dataset is skewed toward English-language content. Non-English profanity is not detected by our regex.