Measuring reach and frequency

By analyzing reach and frequency data, you can estimate how many people saw your ads and how many times they saw them over a certain period of time.

As a basis for estimating reach, we start with cookies. Cookies enable us to estimate reach because they can be used to distinguish individual browsers. When we count reach, we're really counting the number of browsers that have been served an ad or recorded a click on an ad. This measure is an estimate, because some people use more than one browser or computer, and some people share browsers, but it enables us to get a consistent picture of how widely an advertisement has been seen.

See reach and frequency data

You can view reach and frequency data by adding the "Unique cookies" and "Avg. impr. freq. per cookie" columns to your statistics table.

To add these columns to your statistics table:

  1. Sign in to your AdWords account at https://adwords.google.com.
  2. On the Campaigns tab, click the Columns drop-down, then select Modify columns.
  3. Click Reach metrics.
  4. Click Add all columns.
  5. Click Apply.

How Google calculates reach and frequency

Adjusting for cookie deletion

We want to make sure that each cookie that's counted represents a unique browser. However, sometimes users delete their cookies. When they do, a new cookie is created the next time they're served an ad. In such cases, the same browser might be represented by two different cookies: the old cookie that got deleted, and the new cookie that was just created.

Avoiding double-counting

We can make sure that each browser we count is unique by only counting mature cookies. Mature cookies are cookies that were created before the date range for which reach is being estimated. For example, if you run a report from March 15 to March 31, cookies created through March 14 are counted as mature cookies. Cookies created on March 15 or after are discarded.

If we only count mature cookies, then we know that each cookie represents a unique browser, because all of them existed when the reporting period began, and two different cookies can't represent the same browser at the same time.

Correct the data

To avoid double-counting, all new cookies are discarded, but some new cookies are legitimate. To estimate reach more accurately, we have to estimate how many of the new cookies were actually legitimate.

To come up with an estimate, we look at the number of impressions that are associated with mature cookies, and compare the number of impressions to the number of cookies. Then, based on the assumption that the ratio of mature cookies to mature impressions is the same as the ratio of all cookies to all impressions, we calculate a value for the total number of cookies. That value is what we display as reach.

The exact reach correction formula is displayed below:

    (Mature Cookies ÷ Mature Impressions) = (Total Cookies ÷ Total Impressions)

 

In this formula:

  • Total Cookies is the unknown value. It's the same thing as reach, and it's the value that is displayed in reports.
  • Mature Impressions are impressions that are associated with mature cookies.

As you can see, the formula is comparing two ratios: the ratio of mature cookies to mature impressions, and the ratio of total cookies to total impressions. With the data we collect, we can accurately calculate the first ratio. We then assume that the ratio of cookies to impressions is the same overall as it is for mature cookies and impressions.

Solving for Total Cookies, we can restate the formula as follows:

    Total Cookies = Mature Cookies × (Total Impressions ÷ Mature Impressions)

 

Did you know...

You might be surprised to see decimals in your Unique cookies columns, like this: 1,809,601.18

Because the estimated frequency is based on sample group data, the estimation of unique cookies, and the subsequent clicks associated with those cookies, isn't always a whole number.

Next steps

How helpful is this article:

Feedback recorded. Thanks!
  • Not at all helpful
  • Not very helpful
  • Somewhat helpful
  • Very helpful
  • Extremely helpful