What Is Advertising Frequency?

Advertising frequency describes how often a defined audience unit is exposed to an advertisement during a stated period. Average frequency is counted impressions divided by deduplicated reach, using compatible definitions. In newsletters, booking three issues or recording three deliveries does not establish that every subscriber saw the ad three times.

We sell newsletter sponsorships at Dupple, so we have a commercial interest in this channel. All numerical campaign examples below are fictional. We have not measured individual newsletter ad exposure for the scenarios in this article.

Start with the unit and the period

The calculation needs both a numerator and a denominator. If impressions count ad displays while reach estimates distinct people shown the ad, the result is an average per estimated person. If the denominator is browser cookies, email addresses or accounts, call it that. Relabeling an address as a person does not solve an identity gap.

Table: related measures with different meanings for a newsletter buyer.

The period is part of the metric. A monthly average cannot be compared directly with a weekly one. Even two seven-day reports can differ if one covers a fixed calendar week and the other calculates a rolling lookback for each day. Record the date range and time zone, rather than putting the word frequency above an unlabeled number.

Google Ads' frequency documentation, read 6 November 2026, distinguishes a selected-period average from seven-day and 30-day lookback measures for its video ads. Its single-day lookback calculation divides impressions by unique users within the same window. Those product definitions illustrate the importance of matching periods; they are not newsletter reporting capabilities.

Calculate an average without assigning it to everyone

Imagine a fictional campaign with 6,000 counted impressions and estimated reach of 2,000 unique users in seven days. The average frequency is 6,000 divided by 2,000, or 3. Label it as an average over that seven-day interval and retain the measurement system's definition of a user. It is not evidence that every person encountered the advertisement exactly three times.

Two very different distributions can produce that same average. In one fictional distribution, all 2,000 users receive three impressions. In another, 1,500 users receive one impression each and 500 receive nine each. The second adds to the same 6,000 impressions, but most users have only one exposure. The average alone hides this concentration.

If a platform provides a frequency distribution, inspect it alongside the average. If it does not, state that limitation rather than reconstructing individual exposure from campaign totals. Do not infer an exact distribution from opens, clicks or booked dates. Those events have their own definitions and do not supply the missing impression history.

Newsletter deliveries produce a different calculation

Take a separate fictional newsletter schedule with three sends, each producing 1,000 delivery events. Across those sends, the publisher can deduplicate 1,200 email addresses under a stated method. The total is 3,000 deliveries, giving 2.5 deliveries per distinct address. That is a delivery average, not 2.5 ad views per person.

The denominator also explains why you cannot divide by a newsletter's current subscriber total. That total may include addresses outside the selected sends or omit people who unsubscribed after receiving one. Use the cohort actually represented by the numerator. Preserve the cohort's definition if addresses are normalized, suppressed or combined across campaigns.

An email address can be shared, and a person can use more than one address. Without a justified identity method, address-level deduplication stays address-level. Our first-party data guide explains why directly collected data can still have gaps. A publisher's aggregate audience description does not establish a cross-publisher identity graph.

Opens and clicks cannot fill the exposure gap

An open event is not a verified observation of the advertising unit. Apple describes Mail Privacy Protection as downloading remote content privately in the background when a message is received, rather than when it is viewed. That primary description was checked on 6 November 2026. It is one reason an image-loading signal cannot establish that the person read the ad.

Our open-rate and MPP definition covers the reporting distinction. A primary ad can sit inside a delivered newsletter without receiving measurable attention. A dedicated send changes the message format; it does not convert delivery into proof of reading.

Similarly, unique clicks measure a filtered click event under the reporting system's rules. A click can be useful evidence of interaction, but people who do not click may still have seen an ad. Neither dividing clicks by subscriber counts nor counting repeat clickers estimates total ad exposure without additional evidence.

Reach across publishers needs real deduplication

Adding two newsletters' subscriber counts does not yield unique combined reach if their audiences overlap. Nor does buying one issue in each establish that their shared readers received exactly two ad exposures. The publisher can report its own delivery or interaction measures without having visibility into another publisher's identities or the reader's attention.

Google's reach and frequency explanation, also read 6 November, describes modeled deduplication across devices and other signals within its measurement system. That modeling is a specific platform method. It should not be borrowed as an explanation for a newsletter campaign whose reporting has no comparable capability.

When combined reach is unavailable, keep the publishers' figures separate and label the overlap unknown. You can still compare placements on their observed actions or defined commercial results. The honest report has fewer combined metrics, but its denominators remain interpretable and its limitations remain visible to the next person using it.

A booked schedule and a frequency cap solve different problems

A frequency cap constrains serving within the scope and rules of an advertising system. Booking a placement in three newsletter issues constrains the publication schedule. It does not necessarily limit how many times a reader reopens, forwards or encounters a message. Ask exactly which event and identity the cap controls before treating it as a personal exposure guarantee.

There is no single correct repetition count established by the definition. An urgent announcement, a long product evaluation and a resource download have different goals. Our guide to how many newsletter issues to test addresses test design and scheduling. Keep that buying decision separate from claiming an observed exposure frequency.

For a repeat schedule, give each issue a distinct identifier while keeping the campaign grouping consistent. UTMs and sub IDs help distinguish traffic sources and placements. They do not reveal every earlier exposure. Repeated landing visits or multiple tagged clicks should not be presented as a complete history of what someone saw.

Connect repetition to outcomes cautiously

Record the attribution window and the outcome before comparing issues. Someone may first click in one issue and register after another. A report assigning credit to the later click describes its attribution rule; it does not demonstrate that the earlier issue had no influence or that additional exposure caused the registration.

Changing creative, audience, offer and timing together makes it difficult to isolate the role of repetition. Keep a record of those changes. Our newsletter sponsorship ROI guide discusses commercial results, while the advertising glossary separates the underlying delivery and interaction terms. Neither turns a simple before/after comparison into a controlled experiment.

For a Dupple campaign, agree on placements, dates and available reporting before booking. This article supplies no guaranteed exposure count, universal cap or measured optimum. If the report can establish only delivered messages and filtered clicks, use those labels and make the individual ad-view limitation explicit.

FAQ