Effectively Measuring the Success of Marketing and Advertising

The Client

Goodwill Industries International Inc. provides job training, employment placement services, and other community-based programs for people who have barriers preventing them from otherwise obtaining a job. In addition, Goodwill Industries may hire veterans and individuals who lack education or job experience or face employment challenges. Goodwill is funded by a massive network of franchised retail thrift stores that also operate as nonprofits.

The Problem

Our partner, Goodwill of Greater Grand Rapids, came to Pro Bono Analytics seeking volunteer help to get a better understanding of the effectiveness of their media advertising campaigns and the impact their campaigns had on driving foot traffic to their stores. As with most other retailers, store traffic for franchisee Goodwill of Greater Grand Rapids had been declining and there was no clear direction as to which media campaigns were more effective than others.

Goodwill of the Grand Rapids, Michigan Region had used a variety of both paid and unpaid advertising channels (radio, print, and television), and it was important to understand the difference across the ad channels and whether or not they were a paid service. They wanted volunteer help identifying which channels/payment status would yield a statistically significant effect on the transactions. Additionally, they wanted help identifying the effectiveness of their various coupon and discount offerings included in their advertising.

The Approach

Our volunteer team got together and began working through Goodwill’s advertising and sales data from 2014 to 2017 to identify relationship trends between advertising channel and spend and in-store foot traffic and sales. They determined that building linear regression models in R would be the best, most powerful approach for showing relationship trends and recommendations and analysis. R is a common programming language used for custom data analysis and is a common skill in the data science toolkit. Often, custom code can be used to yield some of the same insights as a costly software package with one-size-fits-all analytical tools.

The Solution

Having Pro Bono Analytics volunteers review their data proved successful for Goodwill of Greater Grand Rapids. Our volunteers were able to determine that paying for printed advertising (magazines, newspapers, etc.) had the most positive impact on increasing in-store foot traffic and recommended that Goodwill consider investing more advertising budget into this channel in general. They also proposed decreasing some advertising spend on underperforming channels in addition to creating a survey to inquire with in-store customers to see which media channels they utilize. This project served as an excellent first step toward making data driven decisions that promote optimal effectiveness of marketing and advertising efforts.