YouTube RPM Benchmarks by Niche: Every Published Range, Compared and Sourced

The short answer

The more useful finding is what sits underneath. Revenue is RPM multiplied by views. We sampled four niches through the OutlierKit API on 19 August 2026, and inside a single niche the average views per video across sampled channels varied by 30x to 716x. The widest published RPM gap between niches is about 7x. The niche you pick moves your earnings less than where you land inside it.

Worked out with real numbers from that sample: a personal finance channel at the sample median of 2,666 views per video, using the most generous published RPM of $25, earns about $67 per video from the platform. A gaming channel at its sample median of 47,836 views per video, using a pessimistic $2 RPM, earns about $95. The low-RPM niche wins, using the numbers most favourable to the high-RPM one.

What YouTube officially defines

These are the only hard numbers in this report. Everything else on this page is either a third-party claim or our own sample, and it is labelled as such.

Those definitions come from YouTube's own RPM and CPM help page. The revenue split is published too. Creators receive 55 percent of net revenue from watch page ads on long-form video and 45 percent on Shorts, and YouTube Premium is paid from a separate pool distributed by member watch time, per the partner earnings overview.

Every published benchmark, side by side

These are other people's numbers. We are reporting what each source publishes, with the date and the stated method, not endorsing any of them. Note that only the first table is actually measuring RPM.

Line them up on one niche and the problem is obvious. For personal finance, TubeAnalytics publishes $8 to $35 RPM, Virlo publishes $15 to $25 CPM, and our own niches guide publishes $18 to $45 CPM. Those are three different claims in two different units, and a reader comparing them is comparing nothing.

Why the tables disagree

1. CPM gets labelled RPM. CPM is the bigger, more quotable number, and it travels through blog posts until someone puts it under an RPM heading.
2. Geography is not held constant. The same video earns very differently depending on where the audience sits, so a range built from a mixed creator pool describes the pool, not the niche.
3. Shorts and long-form are mixed together. They pay from different pools at different splits, so blending them produces a number that describes neither.
4. Sample sizes are almost never disclosed. TubeAnalytics says the figures are creator-reported but gives no count. Virlo gives no method at all. Our own page gives figures without linking an outside source, which is the same gap, and we are fixing it.

What we can actually verify: view supply by niche

Views are public. Revenue is not. So this is the half of the equation anyone can check, and it is the half that moves most. Each row below is a semantic search of the OutlierKit channel index run on 19 August 2026. The exact query is printed so you can repeat it.

In personal finance, the sampled channels ran from 442 to 13,199 average views per video, a 30x spread. In consumer tech the spread was 716x, from 1,284 to 919,326. The single biggest video in the tech sample had 4,191,113 views. The biggest in the finance sample had 240,727. Same platform, same month, wildly different economics, and none of it shows up in an RPM table.

This is a small sample and it is a sample of channels our index matched to each query, not a census of the niche. It is enough to establish the spread, which is all we are claiming from it. If you want the same numbers for your own category, the OutlierKit API runs the identical query.

Turning an RPM assumption into a revenue estimate

The formula is simple. Platform revenue equals views divided by 1,000, multiplied by RPM. What matters is being honest that RPM is an assumption you chose, not a number you measured.

Every one of those revenue figures is a floor, not a ceiling. Sponsorships, affiliates, memberships and owned products routinely exceed platform revenue on an established channel, which is why the monetization mix matters more than the RPM you were quoted. Our trending niches report measures the other half, which niches are actually producing breakout videos right now. The drama sub-niche RPM breakdown shows how far the ranges move inside one category.

What changes on 1 February 2027

Any RPM benchmark published before August 2026 is missing this. YouTube announced changes to the Partner Program that take effect on 1 February 2027.

• Creators need 10 million qualified Shorts views over the previous 90 days to remain eligible for ad and subscription revenue sharing on Shorts. Below that, long-form monetization continues but Shorts earnings stop.
• New applicants need 8,000 qualified watch hours in the last 365 days, or 20 million qualified Shorts views in the last 90 days. Existing members are not affected by the new entry bar.
• YouTube Premium Lite expands to every country where Premium is offered, paid from a pool at 60 percent of net subscription revenue.

Details are on the official YouTube blog announcement and the Shorts monetization policies page. We broke down the clauses that got less coverage in the full Partner Program changes analysis. If your plan was to reach the Partner Program on Shorts volume, the route just got harder.

Frequently asked questions

Method and limits

First-party figures come from the OutlierKit channel search endpoint, pulled on 19 August 2026. Each niche row is one semantic query, printed above, returning between 6 and 10 channels. Median, minimum and maximum are calculated across the average views per video that the endpoint returns for each channel. Two channels in the gaming sample returned no view average and are excluded from that row's statistics but counted in the sample size.

Third-party figures are quoted from the pages linked beside them, with the publication dates those pages carry. We did not verify their underlying data, because none of them publish it.

What this report cannot tell you: any specific channel's real RPM, revenue, watch time, or audience geography. Those live in that creator's YouTube Studio and are not available to any third party, this one included. Anyone who tells you otherwise is modelling and not measuring.