CPM is the most quoted and most misunderstood number in creator advertising. It looks like a single figure. In practice, it's a moving target shaped by niche, geography, season, video length, and whether you're buying media directly or sponsoring a creator. Here's how the math works in 2026, and how to read the benchmarks without getting fleeced.

Start with the formula (it never changes)

CPM = “Cost Per Mile,” from the Latin for a thousand is simply the cost of reaching 1,000 impressions:

CPM = (Total Cost ÷ Impressions) × 1,000

Work a few examples and it sticks. $500 spent for 100,000 impressions is a $5.00 CPM. A creator earning $4,800 from 820,000 impressions is running a $5.85 CPM. Fifteen dollars from 5,000 impressions is $3.00. The raw unit, cost per single impression (CPI) is just CPM divided by 1,000.

The formula is the easy part. The hard part is knowing which CPM someone is quoting you.

CPM vs RPM: don't let a creator quote you the wrong one

Google's own YouTube Help documentation draws a clean line between two metrics that get carelessly swapped in pitch decks.

CPM is the advertiser-facing number. It reflects ad and YouTube Premium revenue, counts only monetized (ad-shown) views, and is measured before YouTube takes its revenue share.

RPM is the creator-facing number. It bundles every revenue source, ads, channel memberships, Premium, Super Chat, Super Stickers, divides by all views including demonetized ones, and is calculated after YouTube's cut. As one source frames it neatly: CPM is advertiser value; RPM is creator profitability.

That's why RPM always lands below CPM. Vendor pages consistently peg YouTube's cut at 45% (creator keeps 55%), so a $10 CPM translates to roughly $5.50 in the creator's pocket. Worth noting: Google's official page confirms the before/after framing, but the fetched documentation doesn't actually print the 45% figure, that split is corroborated only by secondary sources. Treat it as the working assumption, not gospel.

One more wrinkle that inflates or deflates the number: not every view carries an ad. Estimated ad-impression rates run 40–65% of views, with one popular calculator defaulting to 62.5%. And playback-based CPM (per 1,000 monetized video views) tends to run higher than impression-based CPM because a longer video can serve several ads per view.

The “average” CPM is a fiction, use ranges

Here's where you should get skeptical. Three independent tools quote three different “global averages”: $4.50, $6.50, and $3–$5. None publishes a methodology or a collection date. The honest read is a range of roughly $3 to $6.50, with the true figure entirely dependent on the variables below. Anyone who quotes you a single authoritative global CPM is selling confidence they don't have.

Niche is the single biggest lever

Every source agrees on the shape: finance sits at the top, gaming, and vlogs at the bottom. The magnitudes, however, diverge by 2–3x between sources, so cite the range, never one number.

  • Finance/Investing: the runaway leader. Quoted ranges span $12–$45, $18–$32 (median $24), and $20–$50+, with premium finance content cited “over $50.”
  • Tech/Reviews: roughly $8–$25 depending on source.
  • Business/SaaS/B2B: one source puts SaaS/B2B at $18–$45; others land business/marketing nearer $6–$14.
  • Education: $5–$20.
  • Health & Fitness: $5–$18.
  • Gaming: $1.50–$16, splitting between hardware-focused gaming (higher) and casual gaming (lower).
  • Entertainment/Vlogs: the floor, at $2–$6.

The takeaway for buyers: a finance creator and a vlogger with identical view counts are not remotely comparable inventory. Pay for the audience's commercial intent, not the subscriber number.

Geography can swing CPM 3–5x

Audience location is nearly as powerful as niche. US and Tier-1 audiences (UK, Canada, Australia, Germany) generate 3–5x the CPM of a global average audience. Tier 2 covers markets like Brazil, Mexico, and Poland; Tier 3 includes India, the Philippines, and Pakistan.

The gap is stark: Indian-audience channels are cited as running 70–85% below US-focused channels, and one modelling approach applies “roughly a seventh” of US rates to India. Germany sits at about half of US rates in that same model. The driver is simple, advertiser spend and purchasing power concentrate in high-income regions. Before you sign a sponsorship on raw reach, ask for the audience geography split.

Season matters: Q4 is a different market

Every source agrees Q4 (October–December) is the peak, and January is the sharpest cliff. The size of the swing is contested, one tool models an 80% Q4-over-Q1 lift, another says 40–60%. Illustrative tool figures show December CPMs above $10 collapsing to under $4 in January. If you're buying awareness, budget accordingly: Q4 buys premium reach at premium prices, while Q1 offers the cheapest impressions of the year.

Format: length and Shorts

Video length changes the economics. Videos of 8+ minutes unlock mid-roll ads, lifting CPM by an estimated 40–80%. Shorts, by contrast, earn dramatically less, one source models Shorts CPMs of just $0.01–$0.06 against $3–$15 for long-form, translating to $10–$60 per million views on Shorts versus $1,000–$5,000 on long-form. (Those specific Shorts figures come from a single source, so treat as directional.) The strategic implication for sponsorships: a Shorts-heavy creator monetizes native ads poorly, which is precisely why a brand integration can command a premium over their dashboard rate.

Buying YouTube ads directly? Different numbers entirely.

If you're running media rather than sponsoring creators, one dated August 2026 benchmark table gives the clearest picture:

  • Skippable in-stream (TrueView), CPV: $0.03–$0.12 per view, billed only on 30s+ views or interactions.
  • Non-skippable in-stream, CPM: $8–$18 per 1,000.
  • Bumper ads (6s), CPM: $6–$15 per 1,000.
  • Discovery, CPC: $0.10–$0.40 per click.
  • Shorts ads, CPM: $4–$12 per 1,000 (an emerging, lower-cost format).

The headline: expect $0.03–$0.12 CPV and $6–$18 CPM in 2026. Verticals matter here too, finance, insurance, legal and enterprise software can run 2–3x baseline, while fitness, food and lifestyle sit cheapest. Budget guidance is blunt: put at least $1,500–$3,000/month behind a meaningful test, because below $1,500 the data accumulates too slowly for Google's bidding algorithms to learn.

For context against other channels, one July 2026 source pegs programmatic display at $1–$3, Meta at $8–$15, and LinkedIn at $25–$40, a reminder that YouTube's non-skippable CPMs sit in the same neighbourhood as Meta.

Use CPM to price sponsorships, not just compare them

The smartest framing in the research: creators should reverse-engineer CPM from a sponsorship offer and price like a media property rather than guess, and advertisers should do the same math in reverse. A sponsorship CPM can legitimately exceed dashboard ad monetization because an integration bundles audience trust, niche alignment, creative delivery and in-content placement that a pre-roll can't buy.

Two reporting cautions before you sign off on efficiency numbers. First, CPM, CPI and CPT are not interchangeable in practice. Second, using estimated impressions instead of billed impressions can introduce a 12–18% variance in efficiency reporting (single-source, so verify against actuals). And on measurement: third-party cookies were ultimately kept after the April 2025 reversal, Consent Mode v2 is required for EEA/UK traffic, and the skippable in-stream billable threshold remains 30 seconds.

The one-line takeaway

There is no single YouTube CPM. There's a formula, a set of multipliers, niche, geography, season, format, and a benchmark range wide enough to hide a lot of overpaying. Always ask which CPM you're being quoted, demand the audience geo split, and price against the range, not a headline average.