Ad frequency has no single right number, but the research points to a range. Meta’s brand-lift analysis found a cap of 2 impressions per person per week captured 95% of the potential purchase-intent lift. A CTV study found purchase intent fell as one viewer’s exposures in a session climbed toward six.
Start with a modest weekly cap and adjust when fatigue signals appear.
What ad frequency and frequency caps mean
Ad frequency is the average number of times one person sees your ad during a set period, usually a day, a week or the length of a campaign. If 1,000 people saw your ad 3,000 times in a week, your weekly frequency is 3. Because it is an average, some people saw the ad once and others saw it six or seven times.
A frequency cap is the ceiling: the most times the platform will show your ad to one person in that period. Before serving an ad, the platform checks whether that person has reached the cap and skips them if they have. Many platforms also offer a second control. Epom’s guide to ad frequency separates the cap, which limits how many times, from recency, which sets the minimum gap between views. Improvado gives a programmatic example of a recency cap: one impression per hour.
Frequency is counted in impressions, and an impression is not the same as a view. Improvado’s frequency capping glossary defines an impression as each time an ad is displayed on a screen, whether or not the person noticed it. Your reported frequency can therefore run higher than the number of times someone actually paid attention, and that is one reason the right number varies by format.
How frequency works on TV, CTV and social
How much control you have depends on how the channel is bought. Traditional linear TV is sold by time slot and program. Amazon Ads’ linear TV guide describes advertisers buying commercial time based on expected viewership, with ads running in scheduled program breaks. You choose the shows and dayparts, but you cannot limit how often one household sees your spot. Streaming TV runs on internet-connected devices, and Amazon Ads notes that these platforms add tools such as completion rates and frequency caps.
| Channel | How you control frequency | What to know |
|---|---|---|
| Linear TV (broadcast, cable, satellite) | Indirectly, through the programs, dayparts and number of spots you buy | No per-person cap; frequency is estimated from ratings |
| Streaming TV (CTV) | Per-viewer or per-household caps where the platform offers them | Caps apply within the platform or buying route you use |
| Facebook and Instagram | A target frequency (impressions per person per week) or a frequency cap, set at campaign or ad set level | Campaign-level and ad set-level controls can’t be combined |
| Google display and programmatic | Hard impression caps per day, week or month; some platforms add lifetime and recency caps | Granularity depends on the platform |
Every cap stops at the edge of one platform. Improvado gives an example: a cap of 3 per day on Google and 3 per day on Facebook can still mean 6 impressions a day for the same person. Neither platform knows about the other, so cross-channel frequency is something you have to estimate yourself. For more detail on streaming TV specifically, see our CTV frequency guide.
What research says about effective frequency
Effective frequency is the number of exposures one person needs within a period before an ad produces the response you want. The Universal Marketing Dictionary’s definition rests on the assumption that the first few exposures may have little effect and the message starts to influence people only after more of them.
The idea has a long and somewhat shaky history. In a chapter on what effective frequency means today, advertising researcher John Philip Jones argues that it is “rooted not so much in research as in common sense.” He notes that Michael Naples’s 1979 book gave the practice academic respectability, but its strongest support came from Colin McDonald’s work, which Jones calls “merely a small-scale experiment” done years earlier in another country. Be skeptical of any single magic number, including the popular rule that people must see an ad three times.
Newer platform research is more useful because it measures specific outcomes. Meta’s effective frequency study found that lift in both ad recall and purchase intent rose with frequency, but at different rates:
- A cap of 1 per week captured up to 80% of the total potential lift in ad recall.
- Recall lift slowed sharply after a cap of 1 per week, while purchase-intent lift didn’t start to slow until close to 1.5 per week.
- A cap of 2 per week captured 95% of the total potential purchase-intent lift.
Meta reads this as support for the idea that bigger behavior changes, such as intent to buy, need more exposures than simple recall.
Streaming TV research shows what happens when frequency goes too high. An MNTN CTV study, reported by Epom, found recall rose from about 64% after one view to 92% after six in a single session, while purchase intent fell 16% over the same range and brand affinity slipped. Taken together, these findings suggest recall is a poor guide to frequency: it keeps climbing after the measures that lead to sales have started to decline.
Starting caps by channel
Epom has published suggested starting caps per user per week. They are one practitioner’s starting points, not industry rules or measured results, and they are useful mainly as a first setting you then test.
| Channel | Suggested starting cap (per user per week) | Epom’s reasoning |
|---|---|---|
| CTV / streaming TV | 4 to 6 per viewer | Ads interrupt the show and few creatives are usually in rotation |
| Facebook and Instagram Feed | 3 to 5 | Fatigue commonly sets in around 6 to 7 on a single creative |
| Facebook and Instagram Stories and Reels | 2 to 4 | Higher-attention format with lower tolerance than Feed |
| Programmatic online video | About 10 | Holds attention, but skippable formats fatigue faster |
| Display, prospecting | 15 to 25 | Banner blindness means people tolerate more exposures |
| Display, retargeting | 10 to 15 (about 3 a day at most) | A smaller pool of people gets overexposed quickly |
The Meta Feed range sits above the 2 per week that Meta’s own brand-lift work found efficient for purchase intent, which is why Epom also notes Meta’s data pointing to about 2 per week for awareness. If your goal is a booked appointment or a sale, the lower end is a sensible place to begin. Our guides to Facebook advertising for small businesses and display ads for small businesses cover those channels beyond frequency.
Signs of ad fatigue
Ad fatigue happens when people see the same ad so often that it stops working. Criteo’s ad fatigue guide lists warning signs worth checking weekly:
- Click-through rate falls while impressions hold flat or rise.
- Cost per click, cost per thousand impressions or cost per acquisition rises with no clear change in the market.
- Frequency keeps climbing while engagement drops.
- One creative takes a large share of the spend but underperforms the campaign’s average click-through rate.
- The same people see the same message several times in a short window, which is common in retargeting.
Keen Decision Systems’ ad fatigue guide adds two more: unique reach that plateaus while impressions keep piling up, and more negative feedback such as skips or hides. Criteo stresses that combined signals matter more than any one metric. A dip in click-through rate on its own could be seasonal. Falling clicks plus rising costs plus rising frequency is a much clearer warning.
Caps alone won’t fix fatigue. Epom cites DoubleVerify’s Samir Karpe, who argued in January 2026 that fatigue is increasingly a quality problem: repetitive ads wear people down however clean the caps are. A cap slows wearout, and a new creative resets it. On TV, that means having more than one spot ready. Our CTV creative best practices cover what makes a streaming ad hold up over repeated views, and the same logic applies to frequency capping in political campaigns, where saturation is common.
How to set frequency on a small budget
- Size the budget to the audience. Frequency is largely budget divided by audience. When there aren’t enough unique people to reach, platforms keep showing ads to the same ones. A large daily budget on a tiny retargeting list will push frequency up fast. Widen the area or lower the budget before you reach for a cap.
- Set a weekly cap where the platform allows one. Start near the low end of the ranges above, especially for purchase or booking goals.
- Count exposures across channels. Add up what one person could see on TV, Facebook, Instagram and search in a week. Two modest caps on two platforms can stack into a heavy load for the same household.
- Rotate creative. Plan at least two versions per channel so you can swap one in when fatigue signs appear, instead of only tightening the cap.
- Test caps against each other. Run identical ad sets with different caps and compare cost per lead or per sale, not recall or clicks alone.
Here’s an example of how this could look in Adwave, offered as an illustration rather than a measured result. A neighborhood restaurant sets one daily budget starting at $30 a day on Adwave, targets a 10-mile radius or a list of ZIP codes, and runs TV, Google Search, Facebook and Instagram from a single plan, so total exposure across channels sits in one place instead of separate ad accounts. When click-through rates slip and costs rise together, the owner orders a fresh spot: extra creation costs $6 per 15-second TV spot, $12 per 30-second spot and $6 per social video. Reddit opens at about $35 a day, and TikTok, which you turn on yourself, starts at $91 a day. Each new channel adds reach but also more chances to see the same message.
If you’re deciding which channels a budget can support, see multichannel channels by budget and our guide to how much to spend on social ads. To run TV, search and social from one budget, see Adwave’s omnichannel advertising, and to see how a campaign plan comes together from your website, visit Adwave for small businesses.
