Comments and Likes by Duration: Search Sample Correction
Correction, 2026-11-16: We withdraw the general advice to make long videos for comments or judge Shorts by views and likes instead of comments. Duration-grouped search results are not a verified Shorts sample or a format-effect experiment. The raw dataset and aggregation code were unavailable for independent recomputation; original tables and charts remain as previously published observations.
Shorts classification also depends on aspect ratio and upload date. Aspect ratios were not verified in this sample. See YouTube’s official Shorts classification guidance.
Summary
- In this search sample’s record, the median comments per 1,000 public views was highest for the over-three-minute group: 0.81 in Korea and 1.09 in the US, against 0.17 and 0.09 for videos of 60 seconds or less.
- The recorded median likes per 1,000 views was highest for the 1–3 minute group: 16.2 in Korea and 25.6 in the US.
- The question-mark-title groups had lower recorded median comment rates. This was not an experiment testing the effect of adding or removing punctuation.
Likes and comments per 1,000 views by length
The original record covers 260 top search results from the preceding seven days across seven Korea-region and six US-region topics. Its stated method is to calculate each video’s public likes or comments ÷ public views × 1,000, then take the group median. That is not total reactions divided by total views. The raw observations were unavailable for independent recomputation in this correction.
| Group | Videos | Likes | Comments |
|---|---|---|---|
| Korea, 60 s or less | 67 | 9.4 | 0.17 |
| Korea, 1-3 min | 33 | 16.2 | 0.40 |
| Korea, over 3 min | 40 | 8.3 | 0.81 |
| US, 60 s or less | 73 | 14.3 | 0.09 |
| US, 1-3 min | 25 | 25.6 | 0.26 |
| US, over 3 min | 22 | 16.6 | 1.09 |
These rates are not probabilities of engagement
The denominator is the public view count at collection, not unique viewers or Studio engaged views. See YouTube’s public statistics definitions. One person can watch repeatedly or leave multiple comments, so the rate does not say how many of 1,000 people commented. Adding likes and comments does not produce a share of people who engaged.
The video counts describe the original groups, not the metric-specific valid sample sizes after hidden likes or disabled comments were excluded. Group-level exclusions and missing-data handling code could not be verified. Korea and US label search regions, not viewer nationality. Topics, channels, upload times and exposure were not controlled.
Short videos did not lack views. US videos of 60 seconds or less had a median of about 720K views (724,626), yet only 0.09 comments per 1,000 views. In Korea, videos over three minutes had a median of about 570K views, higher than 60 seconds or less (313,357) and 1–3 minutes (268,225), and they also had the most comments per 1,000 views.
South Korea by topic
All numbers are medians per 1,000 views. Vlogs (0.84 comments) were well ahead of every other topic (0.17 to 0.27). Games had the most likes (15.7) and personal finance the fewest (5.3).
| Topic | Likes | Comments |
|---|---|---|
| Budget cooking for one | 11.9 | 0.22 |
| Mukbang (eating shows) | 10.5 | 0.27 |
| Vlog | 10.4 | 0.84 |
| Games | 15.7 | 0.24 |
| Diet | 9.3 | 0.25 |
| Personal finance | 5.3 | 0.23 |
| Study | 11.1 | 0.17 |
US by topic
In the US-region searches, AI tools (1.27) and personal finance (0.46) had the most comments, while workout (0.07) and minecraft (0.09) had the fewest. Meal prep had the most likes (24.6).
| Topic | Likes | Comments |
|---|---|---|
| meal prep | 24.6 | 0.37 |
| minecraft | 15.9 | 0.09 |
| vlog | 21.5 | 0.12 |
| AI tools | 17.1 | 1.27 |
| workout | 12.5 | 0.07 |
| personal finance | 14.0 | 0.46 |
Recorded comment rates for question-mark titles
A common tip says putting a question in the title invites comments, so we split them out. In this sample, videos with a question mark in the title had 0.20 comments per 1,000 views in Korea and 0.16 in the US, no more than the other videos (0.31 and 0.18).
| Group | Comments |
|---|---|
| Korea, question-mark titles (15) | 0.20 |
| Korea, other titles | 0.31 |
| US, question-mark titles (14) | 0.16 |
| US, other titles | 0.18 |
On the popular chart, music and gaming differed
We ran the same calculation by category on the YouTube popular chart (Korea, US and Japan), pulled separately at 8:02 p.m. Korea time the same day. The post linked below used a chart pulled at 1:33 p.m., so its counts differ slightly. In the US and Japan, music had more likes and more comments than gaming. In Korea, music had more likes (16.8), but gaming had more comments (0.90 vs 0.42). The chart's makeup is covered in Why YouTube's Popular Chart Shows Mostly Games and Music.
| Group | Videos | Likes | Comments |
|---|---|---|---|
| Korea, music | 60 | 16.8 | 0.42 |
| Korea, gaming | 128 | 4.7 | 0.90 |
| US, music | 42 | 53.0 | 3.52 |
| US, gaming | 129 | 23.4 | 1.53 |
| Japan, music | 32 | 36.6 | 2.21 |
| Japan, gaming | 140 | 12.2 | 0.92 |
Examples
Videos with many comments per 1,000 views. Titles and views on the cards are current YouTube figures.
- Loading…11 min 48 s vlog: 4.0 comments and 56.8 likes per 1,000 views (about 1.96M views on Oct 23)
- Loading…62-second game clip: an exception among short videos, with 2.8 comments per 1,000 views (about 2.56M views)
- Loading…21 min 43 s meal-prep video: 2.5 comments per 1,000 views (about 201K views)
Questions and optional experiments for your own channel
- Define why comments matter. If your goal is questions or community conversation, record the content of comments and whether they receive replies. This sample does not justify prioritizing long videos or making a video longer.
- Choose metrics for the goal, not the format. Comments still matter for Shorts when conversation is the goal. Reach, continued viewing and reactions answer different questions; views or likes alone do not establish success or failure. Start with your own videos of the same format, topic and time window.
- Find your next idea in the comments of popular videos. Search your topic in Rising videos, click a video's comment count and read what viewers asked about or wished for.
- Punctuation does not measure the content of a question. A title without a question mark can still ask a question. If testing a relevant question, change one element at a time and record impressions, traffic source and time since publication. A before-and-after difference alone does not establish causation.
Method and limits
- On October 23, 2026 at 04:33 UTC we ran one YouTube search per topic (region South Korea or US, last seven days, by views). Korean topics were budget cooking, mukbang, vlog, games, diet, personal finance and study; US topics were meal prep, minecraft, vlog, AI tools, workout and personal finance. We kept the top 20 per topic (140 Korean and 120 US videos).
- We left out the Korean ‘AI’ search because only 3 of its 20 titles were in Korean. Many US-region results came from Indian creators.
- Likes and comments are the public counts at collection time. According to the original record, videos with hidden like counts (1 in Korea, 10 in the US) were left out of the like figures, and the video with comments turned off (1 in the US) was left out of the comment figures. Counts change over time, and comments YouTube filters as spam may not be counted.
- The popular-chart figures come from a separate sample pulled at 11:02 UTC the same day.
- We looked only at the top 20, so the sample includes only videos that already did well. Videos that stayed lower in the ranking are not included.
- ‘3 min or less’ uses duration only and can differ from YouTube's Shorts classification.
- Length and comment rates appeared together; this does not show that videos got more comments because they were long. Differences in topic and channel are mixed in.
To check recent videos in your own topic, search a keyword in Rising videos.