CPM calculator

What CPM counts, and what it does not

CPM is cost per mille, and mille is a thousand impressions rather than a thousand people. An impression is one render of the ad, so a viewer served the same creative four times is counted four times and charged for four times. That single distinction is behind most arguments about whether a CPM was good.

• Cost per thousand people reached is a different figure: spend divided by unique reach, times 1,000. It is always higher than CPM, and the gap is your frequency.
• Viewable CPM, sometimes shown as vCPM, charges only for impressions that met a viewability standard, which is commonly half the pixels in view for one second on display and two seconds on video. A vCPM and a CPM from the same flight are not comparable numbers.
• Programmatic reporting often quotes an effective CPM, or eCPM, which back-calculates a thousand-impression price from a campaign bought on clicks or actions. It is a comparison device, not a rate anybody bid.
• CPM is an input price. It tells you what attention cost, and never whether the attention was worth buying.

The exact sums this page runs

One identity, arranged three ways. Pick the quantity you are missing and the page asks only for the other two, so the sum printed under the result is always the sum that produced it.

• CPM = (Ad spend ÷ Impressions) × 1,000. The default arrangement: what you paid, per thousand times the ad was served.
• Ad spend needed = (CPM × Impressions) ÷ 1,000. Budget an impression target: the arrangement a media plan with a rate card needs.
• Impressions you can buy = (Ad spend ÷ CPM) × 1,000. The inverse most people arrive wanting: how much exposure a fixed budget actually buys.

Pulling spend and impressions from the same place

Almost every wrong CPM comes from two numbers that describe different things, rather than from the arithmetic. These steps get both figures onto the same basis before you divide.

1. 1Fix one date range and apply it to both figures. A month of spend against a week of impressions is the most common way to produce a CPM four times too high.
2. 2Take spend from the delivery report rather than the invoice. Invoices cover billing periods, and often carry taxes and platform fees the delivery report leaves out.
3. 3Read impressions from the same report row, and check the column header. Meta shows impressions and reach side by side, and they are not interchangeable.
4. 4Decide whether agency fees, creative production and tooling belong in the spend. Media-only CPM is the industry default; total-cost CPM is the truer number for a board. Whichever you pick, write it down beside the result.
5. 5For video, confirm the buying basis. YouTube and TikTok both sell inventory on more than one basis, and only the impression-billed lines belong in a CPM.
6. 6Split the result by placement before you act on it. A single blended CPM across Feed, Reels and Audience Network hides the placement that is actually setting your average.

When one CPM can be compared with another

A CPM is only meaningful against another CPM bought on similar inventory, in the same country, at a similar time of year. Outside those bounds the comparison mostly measures the auction rather than your work.

• Auction prices rise sharply in the fourth quarter in most consumer markets, so a November CPM against a June CPM measures the calendar as much as the campaign.
• Narrow targeting raises CPM by design. Paying twice as much per thousand to reach people who might actually buy is usually the correct trade, and a cheap CPM against the wrong audience is the most expensive thing in advertising.
• Placement mix moves CPM more than creative does. Feed, in-stream video and off-platform display are different markets, and a blended average across them describes none of them.
• A CPM cannot be compared against a CPC or a CPA at all. They price different events, and converting between them needs a click-through rate and a conversion rate you have actually measured.