Forecasting Advertising Reach and Frequency:
Tips from the Field
Advertising is an industry which pours over data. With the average marketing spend being 10% to 12% of companies’ budgets, ensuring return on investment is crucial. Forecasting key performance metrics before campaigns begin can help choose the correct approach. Reach and frequency are two measures that should be on every advertisers list of performance metrics. Before diving into some tips I’ve learned while measuring these metrics, let us discuss some definitions for reach and frequency.
Reach is the measure of unique users that are served impressions for a given campaign. Understanding the total number of customers for a given campaign, and the reach toward this group identifies how effective the advertising campaign will be at getting the attention of the desired customer group.
Frequency is the average number of ads (i.e. impressions) that each unique customer in the target group will see over the course of the campaign. With the average internet user seeing 5,000 advertisements a day[1], connecting to a customer with your message is key. To do that, message repetition may be needed, but not at the cost of customer exhaustion. Frequency is a measure to get that ideal number. Advertisers talk about the “Rule of 7”[2] which states a customer needs to be exposed at least seven times to a company’s message before they’ll take an action. Is your campaign extensive enough to make these repeated connections with customers and trigger a positive response from them? Reach and frequency calculations require an existing forecast of impressions and an understanding of the total customer base for a product. Now let’s talk about some important features when forecasting reach and frequency.
I’ve found that the best target variable to forecast is the percentage of an audience reached by a campaign. The predicted percentage of reach can then be multiplied by the users in the customer group to get the actual number reached. Within our machine learning model, there were a few key features that lead to improved prediction results. First, our model predicts percentage of reach at the campaign level. Each row of our dataset is a completed advertising campaign. Second, having a way to estimate impressions as an input feature is very important. A separate machine learning model should be built for these forecasts and is outside the scope of this article to discuss. Next, including the number of different advertisements within the campaign gives another metric of exposure for the campaign message. Finally, consider the types of channel the campaign includes to get customer’s attention. Getting the message out in a wider array of channels allows access to the customer base across the different platforms they use day-to-day. These are a few of the informative features we identified when testing models to predict reach percentage.
Reach and frequency are two important metrics every advertiser should measure and consider when building campaigns. Both metrics provide insight into the extent of a customer base reached by an advertising campaign and can help advertisers make adjustments. Adjustment considerations may include:
· Should the budget be increased?
· Are there more channels a campaign should include?
· Can the demographics of a campaign be expanded to attain better performance?
Measuring the percentage reached using a machine learning model was the ideal way we’ve identified to forecast these metrics for advertising campaigns. Hopefully, this article has provided a good case to include reach and frequency as important performance metrics and provided useful tips when generating forecasts for both measures.
Alex Knorr is a Consultant in Slalom’s Data and Analytics practice. He is a data scientist focused on both the advertising and telecommunications industries with a passion for delivering automated machine learning solutions.
Slalom is a modern consulting firm focused on strategy, technology, and business transformation.
[1] https://clario.co/blog/live-secured/ads-seen-daily/
[2] https://www.digitaldealer.com/latest-news/rule-7-social-media-crushes-old-school-marketing/#:~:text=The%20Marketing%20Rule%20of%207,movie%20industry%20in%20the%201930s.
