GoodBelly Case Study: Using Regression to Justify Marketing Expense

This case study provides a regression analysis of GoodBelly sales data to determine which factors predict the number of units sold per store per week. The regression model explains approximately 67.26% of the variation in unit sales, with a significant p-value indicating the model's validity. The analysis reveals that average retail price negatively impacts sales, while having a sales representative, participating in Endcap promotions, and recent demos (within 1-5 weeks) positively influence sales volume. However, the 'natural' and 'fitness' variables were found to be insignificant. Specifically, a unit increase in average retail price decreases unit sales, while stores with sales reps sell significantly more than those without. The study highlights the importance of Endcap promotions and recent demos in driving sales for GoodBelly products. Desklib provides access to similar case studies and solved assignments for students.

Computer Sciences and Information Technology
Student Name:
Instructor Name:
Course Number:
27 October 2018

Regression analysis
We ran a regression analysis to determine which of the factors predict the number of units sold
per store per week. As can be seen below, the value of R-squared is 0.6726; this implies that
approximately 67.26% of the variation in the number of units sold per store per week is
explained by the factors in the model.
SUMMARY OUTPUT
Regression Statistics
Multiple R
0.8201
43
R Square
0.6726
35
Adjusted R
Square
0.6707
33
Standard Error
63.693
03
Observations 1386
In the ANOVA table below the p-value is 0.000 (a value greater than 5% level of significance),
we therefore reject the null hypothesis and conclude that the model is different from zero hence
fit to predict the dependent variable.
ANOVA
df SS MS F
Significance
F
Regression 8 11477979 1434747 353.6647 0
Residual 1377 5586216 4056.801
Total 1385 17064195
Coefficients
Standard
Error t Stat P-value
Lower
95%
Upper
95%
Intercept 298.4881 16.18309 18.44444 4.62E-68 266.7419 330.2343
Average Retail Price -28.5354 3.952153 -7.22021 8.56E-13 -36.2883 -20.7825
Sales Rep 77.43691 3.864453 20.03826 1.39E-78 69.85606 85.01777

Endcap 305.1021 9.055737 33.69158 5.4E-182 287.3376 322.8667
Demo 111.1328 7.403698 15.01045 2.78E-47 96.6091 125.6566
Demo1-3 73.51717 4.895384 15.01765 2.53E-47 63.91395 83.12039
Demo4-5 67.56981 6.541973 10.32866 3.87E-24 54.7365 80.40312
Natural -1.59417 1.776401 -0.89741 0.369655 -5.07891 1.890576
Fitness -1.01967 1.084023 -0.94064 0.347056 -3.14619 1.106844
In terms of significance of individual variables, it was established that only two variables were
insignificant in the model. The two insignificant variables are the number of units sold per store
per week (natural) and the number of units sold per store per week (fitness). The rest of the
variables were significant.
The coefficient of the number of units sold per store per week (Average Retail Price) was found
to be -28.54; this implies that a unit increase in the average Retail Price would result to a
decrease in the number of units of sold per store per week. Similarly, a unit decrease in the
average Retail Price would result to an increase in the number of units of sold per store per week.
The coefficient of the sales rep was found to be 77.44; this implies that stores with sales rep have
a higher volume of sales by 77.44 as compared to the stores without sales rep.
The coefficient of the Endcap was found to be 305.10; this implies that stores that participated in
an Endcap promotion have a higher volume of sales by 305.10 as compared to the stores that did
not participate in Endcap promotions.
The coefficient of the Demo was found to be 111.13; this implies that stores that participated in a
demo in the corresponding one week have a higher volume of sales by 111.13 as compared to the
stores that participated in a demo in the demo in more than 5 weeks ago.

The coefficient of the Demo1-3 was found to be 73.52; this implies that stores that participated
in a demo in the last 1 to 3 weeks have a higher volume of sales by 73.52 as compared to the
stores that participated in a demo in the demo in more than 5 weeks ago.
The coefficient of the Demo4-5 was found to be 67.57; this implies that stores that participated
in a demo in the last 4 to 5 weeks have a higher volume of sales by 67.57 as compared to the
stores that participated in a demo in the demo in more than 5 weeks ago.