CPM is what you paid for a thousand impressions. It says nothing about what those impressions did. What you are actually buying is people moved — people who are now more likely to pick your brand the next time they buy in the category. Here are two buys with the same budget.
The low-CPM buy reached about 7× as many people. The high-CPM buy moved 2.1× as many, for less than half the cost per person moved ($0.39 against $0.80). It cost 4.5× more per impression and was still the better buy, because its lift was 14× higher.
This doesn’t mean expensive always wins. At this price and frequency, the cheap buy would need about 1.0 points of lift to break even. It got 0.5. You can’t see that from the CPM.
A moved person doesn’t buy today. They are now more likely to pick you when they next shop the category. In the “What a brand is worth” model, moving someone from aware to considering adds about 0.55 purchases over the next year. So the low-CPM buy adds about 69K extra purchases, and the high-CPM buy about 142K. The ratio is the same as people moved — the extra purchases follow the lift, not the impressions. And the lift fades unless it’s topped up, so the size of the gain depends on what you do next.
Spend ÷ CPM = impressions. Impressions ÷ frequency = people reached. People reached × lift = people moved. So the dearer buy wins only when its lift is higher by a bigger multiple than its cost per person reached. That holds for any numbers.
The numbers are illustrative. Both buys reach the same kind of people, all of them category buyers. Lift is the same for everyone reached. The extra purchases come from a simple model of how people move between groups, not from a real brand.
The real lift of each buy, against a group that didn’t see it, with enough people to tell half a point from zero — a zero from a small study isn’t proof of zero. Whether the people moved actually buy in your category. And how fast the lift fades.
Cost per person moved, not CPM. CPM is still worth negotiating down — it is the price. But it only tells you which buy is better when the lifts are the same, and they rarely are. A plan that looks efficient on CPM can be the least efficient plan you could have bought.
Numbers illustrative