|
The owner of a smoothie shop noticed a sizeable increase in
sales after placing a full-page ad in a local magazine.
|
|
|
To determine the percent of her budget she should allocate towards advertising in
similar medium, she wanted to know the return on investment (ROI) for the money
she spent on the original ad.
|
|
|
The last month's weather data was used as input for
a model we had built a month earlier, prior to the ad placement. The resulting
graph indicated the number of sales we would have expected based on past
sales trends and the month's actual weather.
|
|
|
After uploading her actual sales data, the two graphs were combined, comparing estimated and actual
sales. She noted that the estimates closely matched actual sales. There was a good
possibility that her advertisement had not generated the increase in sales.
|
|
|
Upon request, we analyzed her data and noted an important interaction: warmer weather had an extraordinarily strong influence
on sales every September. Communicating with her directly, we discovered that college
students returned from summer vacation in September. The interaction between warm
weather, fewer days of rain, and the return of college students had led to more
sales, and we were unable to conclude with any certainty that the effect of her
advertisement had increased sales.
|
|
|
We supplied her with a confidence interval, in order to provide a range for the
ad's actual effect on sales. With this information, she was able to perform a cost
analysis and calculated an accurate ROI for the ad she had placed.
|
|
(click to enlarge any image)
|