YouTube Trending Niches Report: Where the Outliers Actually Are, August 2026

The short answer

The second finding matters more than the ranking. In the two worst-scoring niches, the handful of videos that did break out were tightly clustered in one sub-format. Every qualifying fitness video was about exercise after 60. Almost every qualifying cooking video was slow-living village content or meal prep, not recipe tutorials. The niche label was the wrong unit of analysis.

How this is measured

For each niche we ran one live search through the OutlierKit refresh endpoint on 19 August 2026, with identical settings: three pages of live YouTube results, and a filter keeping only videos scoring 10x or higher on their strongest outlier signal. That signal compares a video against its own channel's normal performance, so it measures breakout, not size.

The hit rate is simply the number of videos that cleared the bar divided by the number of live videos the endpoint fetched. A high rate means the topic is currently rewarding videos well beyond what their channels usually get. A low rate means supply is already meeting demand.

The ranking: outlier hit rate by niche

What is actually breaking out inside each niche

The ranking is the headline. The composition is the useful part.

Fitness: the outliers are all about being over 60

Only three fitness videos cleared the bar out of 120. All three were aimed at older audiences, with titles built around what one exercise does for you after 60 and whether strength beats cardio for staying alive. General workout content produced nothing. If you read the niche as fitness, it looks dead. If you read it as healthy ageing, it looks like the only part still moving.

Cooking: slow living beats recipes

Fourteen of 120 cooking videos cleared the bar, and the pattern was consistent. Rainy village cooking, traditional preparation filmed as ASMR, countryside diaries, and weekly meal prep, from Indian, Vietnamese, Afghan and African channels. The one conventional cooking outlier was a chef turning his apartment into a restaurant, which is a stunt format, not a recipe.

Faceless automation: the loudest niche with the least evidence

Three of 97, and every one came from a channel under 2,500 subscribers, including one with 19 subscribers and one with 136. At that size a single ordinary video produces a huge ratio against a tiny baseline, so even those three are weak evidence. Meanwhile the content itself was almost entirely about how to make money with faceless channels rather than a faceless channel succeeding at something. That is a market selling shovels.

Personal finance: the top rate, and it is not American

Twenty of 120, the best in the panel. The striking part of the sample was its geography. A large share of the qualifying videos were Indian finance content, covering SIPs, salary management in rupees, and money habits framed culturally. That is one sample on one day and not a claim about YouTube as a whole, but it is a pointed reminder that the finance opportunity is not only the US market that the RPM tables obsess over.

AI: high noise, middling breakout

Ten of 100, which puts the loudest topic on the platform in the middle of the panel. The qualifying videos clustered around agent tutorials, full-length courses, and free-tool workarounds. Notably, the single biggest outlier in the AI sample at 59x was a video about YouTube banning faceless niches, which is creator-economy commentary rather than an AI tutorial.

Search demand does not predict breakout rate

We pulled monthly search volume for each niche head term on the same day. The two measures do not line up, which is the point.

Cooking has 45 times the search volume of personal finance and a lower breakout rate. Search tells you how many people want the topic. It says nothing about whether the existing videos are already satisfying them. That gap between demand and confirmation is the same one we take apart in the RPM benchmarks report.

How to use this when picking a niche

1. Rule out, do not pick. A low hit rate is solid evidence a topic is already well served. A high one only tells you the topic rewards good work, not that you will make it.
2. Drop one level before you judge. Fitness looked dead until the sub-format appeared. Run the same measurement on the specific format you would actually publish, not the category name.
3. Discount outliers from tiny channels. A 40x ratio on a 19-subscriber channel is arithmetic, not evidence. Look at what broke out on channels with a real baseline. We took this apart in the sleep and ambient audit, where applying a 1,000 subscriber floor cut the hit rate from 15.8 percent to 6.7 percent.
4. Check the language and region of the winners. Several niches here are being carried by non-US audiences, which changes both the competition and the earnings maths.

For the standing guidance on choosing between categories, the niches guide covers the evergreen trade-offs, and the faceless channels hub covers what still works without being on camera. You can run this measurement on any topic yourself with the Outlier Finder.

Frequently asked questions

Method and limits

Six queries, run once each on 19 August 2026 through the OutlierKit refresh endpoint at three pages per query, filtered to videos scoring 10x or higher on their strongest outlier signal. 677 live videos were fetched in total. The exact query strings are printed above so the run can be repeated.

Three limits worth stating plainly. First, the AI row fetched 100 videos rather than 120 and the endpoint reported an incomplete fetch, so its denominator is smaller than the others. We ran it twice and got 9 and then 10 qualifying videos out of 100, so the rate is stable even if the fetch is not. Second, the filter applies to whichever outlier signal is highest for a video, which on very small channels can be inflated by a low baseline. That is why the faceless row is discussed rather than simply reported. Third, this endpoint does not return publish dates, so the report measures what is breaking out now and cannot show a trend line against last quarter.

Six niches is a panel, not a census of YouTube. Treat the ranking as evidence about these six queries on this date, which is exactly what it is. The next edition repeats the same six so the numbers stay comparable.