Aug 062011
 

Have you ever calculated the right or Optimal Frequency for your marketing campaign? You can save some marketing dollars by using Optimal Frequency (number of times customer see an ad before making a purchase).

Methodology of Digital Campaign using user level data

Before we delve into detail, let me mention that outcome can be a Click/Visit/Purchase. This analysis assumes all creative concepts are similar so that environment is stable for right frequency calculation. Refer to my previous post Effective Creative Testing for sanctity of the test environment.

  1. Keep single outcome for each user.
    1. In case of multiple outcome for each user, consider the first outcome (first click/first visit/first purchase). This is to eliminate any effect of product experience on future marketing. Some Analysts like to keep the last outcome which is fine, as I have not seen too different results at least in Optimal Frequency Range.
    2. Secondly, if we are determining Optimal Frequency for a shorter time period, there might not be many cases for multiple outcomes to have any significant impact on results.
  2. Determine all impressions that occurred prior to each outcome for each respective user
    1. All Impressions where Impression time < Outcome time (Click Time / Visit Time / Purchase Time) for respective users.
  3. Determine frequency distribution of users using SQL or other data mining tools illustrated as below:
  4. Determine for each frequency bucket:
    • Likelihood to Click = # of Clicks / # of Distinct Users
    • Likelihood to Visit = # of Visits / # of Distinct Users
    • Likelihood to Purchase = # of Purchases / # of Distinct Users
  5. Plot data in excel using one horizontal and two vertical axes as illustrated below:
    • Optimal Frequency is 4 or above. Purchase behavior is significantly different for consumers who received 4 or more impressions compared to 3 or less impressions.
    • A closer look reveals 72% of users received 3 or fewer impressions and 28% received 3 and more.

Advantages of Optimal Frequency Analysis

  1. It helps us determine overdelivery and underdelivery of Ad.
  2. A frequency distribution tells us what percentage of customers is in each frequency bucket.
  3. This analysis will help determine right frequency and adjust the reach accordingly. In case of limited budget, you can reach either a large number of audience with low frequency or a small number of audience with high frequency.
  4. It can help do Site Frequency Analysis. Determine frequency by each site. Site A with higher frequency or Optimal frequency might be doing better than Site B with lower frequency. If AD on Site A is bought on a high CPM and Ad on Site B is bought on low CPM then increasing frequency on site B might save some marketing dollars.

Other Facts

  1. Optimal Frequency can be one number or a range of numbers.
  2. Optimal Frequency will change for different time periods. Optimal frequency for a 3 month time period might be different from Optimal Frequency for 6 months. This is because the frequency distribution and conversions will change over different time periods.
  3. Optimal frequency will vary by brand.
  4. Determining Optimal Frequency based on Ad Size or different Creative Concepts can get a little tricky.
    1. A customer receives ad of different sizes in different order. For example, a customer sees a 728×90 and then see 300X250. Other customer might have an entirely different experience. We have to bucket customers in different ad unit sizes, for example customers who just received 728×90 vs. 300X250 and then find out the conversion rate based on impression count.
    2. Also, if many ad sizes are running on one page then it could get a little tricky.

Have you ever calculated Optimal frequency for your campaign? Do you use a different methodology? Please share your experience via comments.

  2 Responses to “Optimal Ad Frequency”

  1. Excellent article and very well explained Shilpa. It is also interesting to see several times how Optimal Frequency change as customer moves in different stages of purchase funnel. We can segment customer data by their purchase funnel stage and Optimal Frequency might be different for each stage.Customers in the top of funnel might need a much lower frequency compared to customers in bottom of funnel. Again nice article. Thanks Mike.

  2. Thanks for your thoughts Mike. I agree it might be useful to calculate Optimal frequency for different segments of Purchase funnel. Shilpa.

 Leave a Reply

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

(required)

(required)

CommentLuv badge