Reach and Frequency: How Many Times Should One Person See Your Ad?

What are reach, frequency and impressions?

Think of one loyalty card per person. Reach counts the cards, impressions count the punches, and frequency is the average number of punches per card.

Reach

The number of different people who saw your ad at least once. In practice a platform counts accounts: Meta's reach is "the number of Meta Accounts that saw your ads at least once" (Meta).

Impressions

The number of times the ad was shown. If an ad is on screen for one person 2 different times in a day, that is 2 impressions (Meta).

Frequency

The average number of times each person reached saw the ad. It is "impressions divided by reach" (Meta), so it is never below 1.

Frequency cap

A limit you set on how often an ad appears to the same person in a day, week or month (Google Ads). Frequency is what happened; a cap is the ceiling you asked for.

Gross rating points (GRPs)

Reach as a percentage of your audience, times average frequency (Washington State University). It sums up how heavy a schedule is, not how many people it touches.

How do you calculate reach, frequency and GRPs?

Five lines cover almost every media plan. The examples use a $5,000 budget, a $25 CPM and an audience of 400,000 people.

• Impressionsbudget / CPM x 1,0005,000 / 25 x 1,000 = 200,000
• Frequencyimpressions / reach200,000 / 40,000 = 5.0
• Reach %reach / audience x 10040,000 / 400,000 x 100 = 10%
• GRPsreach % x frequency10 x 5 = 50
• Two channels1 - (1 - a) x (1 - b)40% and 30% give 58%, not 70%

The last line estimates combined reach when seeing one channel tells you nothing about seeing the other (Mediaocean). Real audiences overlap in less tidy ways, so treat it as a first estimate.

Here is the trade-off on that budget. Every row buys the same 200,000 impressions and the same 50 GRPs, which is why GRPs alone cannot tell you which plan you bought.

Why does reach stop growing when you spend more?

Each new impression is more likely to land on someone you already reached, so a bigger budget buys repeats rather than new people. Nielsen's predicted curve for digital campaigns aimed at Canadian adults aged 25-54 shows the shape: going from 100 to 200 GRPs lifts reach from about 20% to 34%, a 70% rise, while going from 400 to 500 adds only 2%.

Delivery is also far from random. If 100 GRPs landed evenly at random, a Poisson model (which a Nielsen patent calls an industry standard) would put reach near 63%, or 1 minus e to the power of -1. That figure is our calculation. Nielsen's curve predicts about 20%, which means about 5 impressions per person reached. The gap is the repeat exposure a plan pays for without choosing it.

Once the curve goes flat, stop adding budget to that audience. Add a new audience or channel instead, or keep the money.

What does an average frequency hide?

It hides who took the repeats. These two campaigns report exactly the same numbers: reach 5, impressions 15, frequency 3.0.

Real delivery can look a lot like B. Two published examples, both from vendors:

• 77xAverage frequency for the 20% of households that took 81% of connected TV impressions in political ads, while the other 80% averaged 4.5x. Source: MiQ data, 2024, 9.2 million connected TV and linear devices.
• 60%+Share of linear TV ad impressions going to the 20% of households that are heavy viewers, across four markets. Light viewers are 45% of households and get 5%. Source: Samba TV, Q4 2021.

So read the spread, not the average. Google Ads has a frequency distribution view for video campaigns, and TikTok's Reach & Frequency campaigns have a Frequency Breakdown Report.

Do not panic if a report shows more than your cap. Check the date range first: Google notes that a 3-a-week cap read over 2 weeks "may find up to 6 impressions per person". The same page says caps are "applied by cookies", so one person on two browsers can pass it (Google Ads). Display & Video 360 counts only viewable display impressions towards a cap, while its reports may count unviewable ones too (Google).

Where do you set frequency on each platform?

Each platform below reports estimated or modelled reach in its normal reports, and each offers different controls. Checked on 2026-10-30.

How do you find the right frequency for your campaign?

Start low, then let your own results raise the number. The 2-a-week start and the 6-a-week alarm are our judgment from the studies above, not a published standard.

1. Set the window to the buying cycle. Frequency only means something over a period. Use a week for things people buy often and a month for slower purchases, as Naples's "within a purchase cycle" suggests.
2. Start at 2 a week for an established brand. Meta says brands with high awareness can limit ads to "one or two times each week", and 2 is the lowest weekly Target frequency Google supports. For a new brand or a short campaign, start higher: Meta says "a high-frequency strategy may be beneficial". Treat 6 a week as an alarm.
3. Spread delivery across the flight. Ephron's case for more weeks and the restaurant-search result both favour spacing over bursts. Avoid a schedule that spends the whole budget in a few days.
4. Bring enough creative. Google says to add at least as many creatives as your Target frequency number, to avoid fatigue.
5. Read the distribution every week. Look for people far above the average, and heed Meta's warning: if performance drops as frequency rises, change the creative or the targeting.
6. Test before you raise it. Split an audience between two caps, keep a group that sees nothing, and compare the outcome that matters. Small tests stay noisy: in 25 experiments at large US retailers and brokerages, measuring sales return reliably took a sample "typically in excess of ten million person-weeks", so read a small result as a direction, not a verdict.

Reach and frequency questions