Quick answer
CPM (cost per mille) is what an advertiser pays for 1,000 ad impressions: spend ÷ impressions × 1,000. It prices reach, not results, so a low-CPM channel can still cost more per purchase; compare channels only after converting to cost per session and cost per conversion.
Formula: CPM = spend ÷ impressions × 1,000
What is CPM?
In advertising, CPM stands for cost per mille, Latin for "cost per thousand". It is the price an advertiser pays for 1,000 impressions of an ad, whether or not anyone clicks. Google Ads describes CPM bidding as paying "per one thousand views (impressions)" on the Google Display Network, and social platforms such as Meta sell most of their inventory the same way.
That is all CPM tells you: what reach cost. It says nothing about whether the people who saw the ad visited your site, signed up or bought. A channel can have a low CPM because its impressions are cheap for good reasons (a broad audience, low-intent placements, small or brief ad slots) and still be the most expensive way to win a customer.
Two variants are worth knowing:
- Served CPM counts every impression the platform delivered. Google Ads counts an impression "each time your ad is shown on a search result page or other site on the Google Network"; Meta counts an impression the first time an ad is on screen.
- Viewable CPM (vCPM) charges only for impressions that could be seen. Google Ads treats a standard display ad as viewable when at least 50% of its area is on screen for at least one second, and a video ad when at least 50% is visible while it plays for at least two seconds, which follows the Media Rating Council guideline.
vCPM is always higher than served CPM for the same campaign, because the same spend is divided by fewer impressions. Compare like with like.
The CPM formula
CPM = spend ÷ impressions × 1,000
The formula rearranges three ways, which is useful when you plan a budget instead of reading a report:
| You know | You want | Formula |
|---|---|---|
| Spend and impressions | CPM | spend ÷ impressions × 1,000 |
| CPM and impressions | Spend | CPM × impressions ÷ 1,000 |
| Spend and CPM | Impressions | spend ÷ CPM × 1,000 |
| CPM and click-through rate | Cost per click | CPM ÷ (1,000 × CTR) |
The last row is the bridge to the next denominator. If a placement costs $4.00 CPM and 0.67% of impressions turn into clicks, each click costs $4.00 ÷ (1,000 × 0.0067) ≈ $0.60. That is the first step of the chain this article argues for: per thousand impressions, then per session, then per conversion.
CPM vs CPC vs CPA
All three divide the same spend. They differ only in the denominator, and therefore in what risk the advertiser carries.
| Metric | Denominator | Who carries the risk | What it answers |
|---|---|---|---|
| CPM | 1,000 impressions | Advertiser: you pay even if nobody clicks | What did reach cost? |
| CPC | One click (or one session, in your own analytics) | Shared: platform must earn the click, you must convert it | What did traffic cost? |
| CPA | One defined action: sign-up, order, subscription | Mostly the platform, if you buy on CPA | What did a result cost? |
| ROAS | Revenue ÷ spend (a ratio, not a cost) | n/a | Did the spend pay back? See ROI and ROAS |
A rule of thumb: buy on whatever unit the platform sells, but judge on the denominator you actually care about. A brand-awareness campaign can be judged on CPM and impressions; a campaign meant to sell must be judged on cost per purchase.
Worked example: the cheap CPM that costs more per purchase
Sample data: a coffee-subscription store runs two paid channels for one month. Channel A is paid social bought on CPM; channel B is paid search bought on CPC. Impressions and clicks come from each ad platform; sessions and purchases come from the store's own analytics.
| Channel A: paid social | Channel B: paid search | |
|---|---|---|
| Spend | $2,400 | $800 |
| Impressions | 600,000 | 25,000 |
| CPM | $2,400 ÷ 600,000 × 1,000 = $4.00 | $800 ÷ 25,000 × 1,000 = $32.00 |
| Clicks (platform) | 4,000 | 1,000 |
| Sessions (site) | 3,320 | 870 |
| Cost per session | $2,400 ÷ 3,320 = $0.72 | $800 ÷ 870 = $0.92 |
| Purchases (first orders) | 40 | 26 |
| Conversion rate (purchases ÷ sessions) | 40 ÷ 3,320 = 1.20% | 26 ÷ 870 = 2.99% |
| Cost per purchase | $2,400 ÷ 40 = $60.00 | $800 ÷ 26 = $30.77 |
Read the table from top to bottom:
- Per thousand impressions, A is eight times cheaper ($4.00 vs $32.00). On CPM alone, you would move budget to A.
- Per session, A is still cheaper ($0.72 vs $0.92), but the gap has shrunk from 8× to about 1.3×, because search impressions turn into clicks far more often (4.0% CTR vs 0.67%).
- Per purchase, the order flips. A costs $60.00 per first order, B costs $30.77. Search visitors arrived with a need; social visitors were interrupted while scrolling, and converted at 1.20% instead of 2.99%.
The same $3,200 produced 66 purchases in total, a blended $48.48 each. Shifting $800 from B to A at A's current rates would buy about 13 more purchases ($800 ÷ $60) and lose 26, so judging by CPM would cost the store purchases even though it bought more impressions. The CPC and CPA entries reuse this same dataset to show where clicks go missing and how the choice of "acquisition" changes the answer.
This is the whole case for converting CPM into a common denominator before you compare channels: CPM measures the price of attention, not its yield.
What moves CPM
CPM is a market price set in an auction, so it moves with supply and demand for the audience you target. The levers that matter in practice:
- Audience size and competition. The narrower the targeting, the fewer impressions are available and the more advertisers bid against each other for them. If CPM jumps after you add an interest or a lookalike filter, that is the filter, not the creative.
- Placement and format. Search results, in-feed video and premium publishers usually cost more per thousand than small display slots or low-attention placements. Higher CPM is often paid for higher attention.
- Seasonality. Demand for impressions peaks before big retail dates, so the same audience costs more in late November than in February. Compare CPM year on year, not month on month.
- Viewability rules. Moving from served to viewable buying raises the CPM number without changing what you pay overall.
- Ad quality and relevance. Platforms reward ads people engage with, so a fatigued creative can raise the effective price of the same audience. If frequency (impressions ÷ reach) climbs and CTR falls, refresh the ad before blaming the market.
Common CPM mistakes
- Ranking channels by CPM. As the example shows, the cheapest impressions can produce the most expensive customers. Always carry the comparison down to sessions and purchases.
- Mixing served and viewable CPM. A $6 vCPM and a $4 CPM describe the same price when two-thirds of impressions are viewable ($4 ÷ 2/3 = $6). Check the column definition in each platform before putting them side by side.
- Comparing impressions across platforms. Google and Meta count impressions differently, and Search Console counts organic impressions differently again. Summing them gives a number nobody can interpret; see impressions.
- Forgetting the calculator works both ways. Work backwards from the purchase price you can afford. Channel A produced 40 purchases from 600,000 impressions (0.67% CTR × 1.0% of clicks buying). If a purchase may cost at most $40, the CPM you can afford is $40 × 40 ÷ 600,000 × 1,000 ≈ $2.67, well below the $4.00 you pay. That is a planning number you can take into the negotiation; the CPM calculator does the arithmetic.
- Treating low CPM as efficiency in a report to management. Report CPM next to cost per session and cost per purchase, and the marketing ROI the spend produced.
Reading CPM-bought spend in TapCub
TapCub does not buy ads or run auctions, and the Channel ROI page does not calculate CPM; that number stays in your ad platform. What TapCub adds is the part of the chain the platform cannot see: sessions and purchases on your own site. The example above works like this:
- Import the spend. In All-platform analytics › Business › Channel ROI, open the Costs tab and add spend one row at a time or import a CSV with
day,channel,source,campaignandcost(clicksandimpressionscolumns are accepted too). Importing the same day, channel, source and campaign again overwrites the row, so you can re-upload a corrected month without duplicates. - Read cost per click and cost per acquisition. On the ROI tab, the page divides spend by sessions (cost per click) and by purchases on the order ledger (cost per acquisition), and divides revenue by spend (return on ad spend). For channel A that is $0.72 per session and $60.00 per purchase, the numbers that overturn the CPM ranking.
- Group by campaign. Switch the grouping from channel to source or campaign to see which social campaign drags the average down. The quadrant plots sessions against conversion rate, so a campaign with lots of cheap traffic and few purchases stands out.
- Trust the denominator. Bots and AI crawlers are counted separately in TapCub, so they do not inflate the session count that cost per click is divided by.
TapCub does not change bids or move budget; it only measures. Check the decision with the marketing ROI calculator, and see CAC when you need the fully loaded cost of a customer rather than one channel's media cost.
See also: What is real-time bidding (RTB)? How the auction works and what it buys.
See also: What is an ad network? Types, revenue and publisher trade-offs.
See also: What is a banner ad? Sizes and how to measure it.
Sources
- Google Ads Help, Cost-per-thousand impressions (CPM): Definition.
- Google Ads Help, Understanding viewability and Active View reporting metrics.
- Google Ads Help, Impressions: Definition.
- Meta Business Help Center, Impressions.
- Media Rating Council, Viewable Ad Impression Measurement Guidelines (PDF).
Was this explanation clear?
We use your answers to fix and expand this page.
Couldn’t send. Please try again in a minute.
Thanks for the feedback. We’ll use it to improve this page.You’ve already sent feedback on this page. Thanks.

Product Experience Specialist
Explores usability, user journeys, and feature adoption through practical examples of onboarding and everyday product use.