Cheap CPM doesn’t mean efficient advertising

What CPM won’t tell you

CPM is the price of a thousand ad impressions. It tells you what you pay for a banner loading on a page. It doesn’t tell you whether anyone could see it. A banner at the bottom of an article the reader never scrolls to counts towards CPM exactly the same as a banner in the first screen of the page.

That’s why viewability is measured: the share of impressions where at least half of the banner was on screen for at least one second (two seconds for video). It’s the industry standard that buying platforms also use.

But viewability isn’t enough either. A banner seen for one second and a banner that stayed on screen for half a minute count the same.

Three numbers to ask for next to CPM

1. Viewability: what percentage of impressions had a chance to be seen.
2. Average viewable time: how long the banner stayed on screen once it was visible.
3. Cost per viewable impression: CPM divided by viewability. What you really pay for a thousand impressions that could be seen.

For a quick comparison of offers, one more calculation helps. Divide the cost per viewable impressions by the average viewable time, and you get the cost per thousand viewable seconds. It isn’t an official metric and it doesn’t measure attention, but you can work it out in a spreadsheet from a standard report. And suddenly the differences between offers look very different.

Example 1: Same price, different value

A baby nutrition brand ran an awareness campaign in the autumn, targeting women aged 25 – 44. In the Czech Republic it bought banners in two ways: through a deal with a large Czech advertising network (a pre-negotiated purchase of premium inventory) and on a list of selected women’s websites that we put together ourselves.

A standard mobile banner cost exactly the same in both cases: CZK 56 per thousand impressions. That’s where the similarity ends.

Czech Republic, September 2026. Same CPM, very different value.

But beware of jumping to the conclusion that deals don’t pay off. With the same supplier, desktop branding (a large ad format framing the page content) came out the other way round.

Czech Republic, September 2026. Here the premium deal wins.

Here the deal is roughly half the price per viewable second, and it was the most cost-effective purchase in the entire Czech campaign. So even the most expensive format isn’t necessarily expensive.

Premium deals also bring value that viewability doesn’t capture: a quality environment and trustworthy content. In this campaign, visits from the deals were more engaged on the website. According to Google Analytics, 63% of them were engaged sessions, compared with 49% from the selected websites, and they spent an average of 8 seconds longer on the site. These are dozens of visits, though, so we treat this as a signal, not a rule.

Example 2: The cheapest market isn’t the cheapest

The same campaign ran in the Czech Republic, Poland, Hungary and Slovakia. The viewability target was 80%, and every market hit it. According to the report, all was well. But when we recalculate the numbers, this is what we get:

Banner campaign, September 2026. All costs in CZK.

The Czech Republic has the highest CPM: 37% higher than Poland and 78% higher than Hungary. Yet per viewable second it comes out 13% cheaper than Poland and 7% cheaper than Hungary. That’s because the banner stays on screen for over 25 seconds on average here, compared with just 12 in Hungary.

Hungary is the opposite case. The highest viewability and a low CPM look like a great offer, but people only had the ad in front of them briefly. Slovakia is cheap by both measures; the gap is just smaller than CPM suggests.

A note on comparing countries: CPM differences between markets are expected. Each market has its own price level, advertiser demand, publisher landscape and purchasing power, so a higher Czech CPM isn’t a sign of a bad buy on its own. We’re not trying to crown the cheapest market. The point is the method: once you add viewability and viewable time, the ranking can change.

Be careful when comparing averages, though. The markets were bought differently: in the Czech Republic a large share of the budget went to the more expensive branding format and to desktop, while in Slovakia it was almost exclusively standard mobile banners. A market’s average CPM is therefore the result of its mix of formats, devices and websites, not the “price of the market”. The table above is a good first look. But decisions need to be made on comparable rows: same format, same device, similar websites.

Example 3: What we did about it

Numbers only matter if you change something based on them. In Poland, in mid-September, we ran into two problems at once. The campaign was underspending its budget because most of the audience had already seen the ad as many times as the frequency cap allowed. And viewability was sitting just below the 80% target.

First, we raised the frequency cap, increased our bids in the auction and tightened the viewability requirement. Viewability moved above target, but at a price: the cost of a viewable impression nearly doubled.

So we added open auction, targeted at relevant website categories, alongside the list of selected websites. Open auction is cheaper but less controlled. We therefore kept checking where the ads were showing, excluded unsuitable sites, and removed sites that were already on our list from the auction, so the same inventory wasn’t bought twice.

Poland, September 2026.

This isn’t a controlled experiment: the open auction ran on different websites, and several changes happened at once. But it shows the principle. When you can see the cost per viewable impression, you know where there’s room to save and where it pays to spend more.

So why not buy everything in open auction? In Poland it came out at roughly half the CPM of the selected websites (CZK 46 vs CZK 87), with higher viewability (88% vs 84%). On these numbers alone, it looks like an easy call. We still keep premium deals and selected websites in the mix, for three reasons:

Formats and placements. Large formats such as branding, homepage takeovers and other premium placements are sold almost exclusively through deals. Open auction gives you mostly standard banners on whatever inventory publishers decide to release there.

Websites. Many of the strongest publishers sell their best inventory only through deals, or don’t offer it in open auction at all. If you want your brand there, a deal is the only way in.

Viewability isn’t attention. A viewable impression only means the ad could be seen. Attention measurements typically show that premium formats in uncluttered, quality environments hold people’s attention better than open-auction inventory with comparable viewability. A cheap viewable second isn’t worth much if nobody looks.

Open auction is a tool for adding cheaper reach under close supervision, not a replacement for premium buying. Deals carry the formats and environments that do the brand work; open auction adds volume around them.

What to check and compare

• Ask for three more columns. Next to CPM, request viewability, average viewable time and cost per viewable impression in every report. Buying platforms such as DV360 report them as standard.
• Compare. Put the same format, device and type of website side by side, not campaign or market averages. In the Czech campaign, overall viewability of 81% hid a range of roughly 58% to 92%.
• Move budget by cost per viewable second. Within each format, shift spend towards the suppliers and websites that deliver viewable time more cheaply, and away from those that only look cheap on CPM.
• Use open auction for volume, with oversight. Add it where you need cheaper reach, but review placements regularly, exclude unsuitable sites and make sure you’re not buying the same inventory twice.
• Look past the screen. Viewable time tells you the ad was on screen, not that anyone looked. For key formats and suppliers, add attention measurement, and check visit quality in Google Analytics before cutting a premium buy that only looks expensive.

Summary

CPM is a convenient shorthand, but it won’t tell you whether anyone saw your ad, or for how long. As soon as you add viewability and viewable time, the ranking of markets, formats and suppliers can flip. A more expensive purchase is often the better deal, and a cheap one can turn out to be expensive.

Want to know how much of your budget ends up in viewable impressions?

Send us your campaign report and we’ll show you, free of charge, how it looks once recalculated for viewability and viewable time.