Three Views Is a 1972 Rule. Measured CTV Campaigns Run at 7 or More.

Ask almost any marketer how often an ad needs to be seen and you will hear the same number: 3. Why?

Because one view introduces, one explains, and one closes. After that you're wasting money. It sounds like science, and in a way it is. But the study behind it ran in 1972, in a world of three TV networks. And, while the research stopped in 1972. The quoting never has.

So here is the problem with running your streaming TV budget on that rule: the market stopped believing it years ago. And real campaign data proves it. Today, the average streaming TV campaign reaches a person nearly 7 times, with the biggest spenders targeting 10 or more. If you plan to 3 while measured campaigns run at 7+, you are not just being inefficient, you're spreading yourself too thin.

Where "3" actually came from

Herbert Krugman ran public opinion research at GE. In 1972 he published a short paper in the Journal of Advertising Research: "Why Three Exposures May Be Enough". His model goes like this. The first view asks "what is it?" The second asks "what does it mean to me?" The third is the nudge that makes people act.

But read the paper properly. Krugman wrote that from the third view on, an ad is "merely a repetition" in the viewer's mind. Three is where diminishing returns begin. It was never a target number. It was a caution.

The industry flipped the number anyway. In 1979 Michael Naples wrote the ANA's study on effective frequency. It settled on about three views per person. By most accounts, 90% of big US brands planned that way for 20 years. Think about what that fit: soap and cereal. Three TV channels. Simple ads. A quick buy. Three views for toothpaste in 1979 made sense. Three views for a streaming service on TV in 2026 is antiquated.

What measurement shows now

Once you could actually measure delivered frequency at scale, the numbers moved. And they did not move back to 3.

The Trade Desk published frequency research from its own campaign data. The finding: more views kept earning conversions well past three. Extra views only stop paying for themselves at about 6 per week on streaming TV, 10 a week on online video, and 20 or more on display. Before that line, every extra view is still earning its money.

Innovid's 2025 CTV report measured real campaigns. It used hundreds of billions of delivered impressions from 2024. The average campaign ran at a frequency of 7.09. Campaigns over 200 million impressions averaged 10 or more. Those are the biggest budgets, with the most to lose from waste.

The research moved the same way. A 2025 study in the Journal of Marketing Analytics found two things. A person has to notice an ad before it can teach them anything. And normal ads do better with more views that are shorter, not fewer views that run long.

And the honest summary: the Advertising Research Association said in January 2026 that no single number fits every campaign. People quote that line like it ends the debate. It does the opposite. If nobody knows the exact number, you plan to the one with the most evidence behind it. Three has none.

One more thing worth saying plainly. The biggest claims in this space fail a source check. Open almost any frequency guide and you will find it: "Nielsen says six exposures." We looked. No Nielsen document says it. Not a report, not a slide, not a press release.

What Should Your Ad Frequency Be? At Least 10 Views a Month

We plan every campaign so each person sees it at least 10 times a month. Longer flights stack on top: two months at that pace is 20 views per person.

Run that number against the proof and it is not aggressive. Ten views a month is about 2.3 a week. That is under half The Trade Desk's measured cliff of 6 a week. It is where Innovid's biggest campaigns land. And it clears the first hurdle from the research: inside a buying window, people who see you twice never formed an opinion to act on.

Our reach and frequency calculator has three tiers: light at 10 views a month, medium at 20, heavy at 30. A $10,000 budget at a $25 CPM buys 400,000 impressions. At light frequency that reaches 40,000 people properly. Spread the same money over 400,000 people and each sees you once. That is not a campaign. It is a rumor. Need the CPM first? The CPM calculator gets it from budget and impressions.

So what for you: frequency decides what your money buys: opinions, or just activity. You already know which one books meetings.

We sell frequency, so test this without us

That framing has an obvious conflict. We build campaigns at these frequencies. We profit when you buy more views. So use the sources yourself before you trust our math.

Pull up any campaign report, yours or a competitor's. Look for delivered frequency, not promised reach. Say a plan promises to "reach 1 million people." Now check the budget. If it works out to under four views per person over the flight, you found it yourself. The ads were too thin to stick. This test works on us too. Run it against our own proposals. We would rather you catch drift from a formula than from trust.

What this post does not prove

These benchmarks cover many categories, ads and markets. They are not your category's number. Ten a month is our planning floor, not a measured optimum for your funnel. A famous brand with great creative may work at less. A new name in a crowded field may need the heavy tier. What the proof does support, firmly: buying to the folklore number of 3 leaves money on the table.

Want that floor applied to your budget? The calculator does it in ten seconds. Our first CTV campaign guide covers what else a plan needs. For the budget side, see what CTV actually costs.

Know your number? Put it to work. Tell the Campaign Wizard your budget and audience, and it builds a streaming TV, video and display plan around a frequency that earns the impressions.