QM 601 Research Methods · Walsh College · Spring 2025 · team project

Do customer loyalty programs lift retail sales?

A research design, built with Anthony Callan and Joy Zembe, for testing whether a loyalty program raises store sales. The analysis plan was checked end to end on simulated store data, and every test reruns here.

Research designPaired t-testCochran sample sizeLinear regressionPearson correlationPython
SituationA retailer cannot casually A/B test a loyalty program on real stores.
TaskDesign a defensible test of whether membership lifts sales before spending anything.
ActionSpecified the survey (n = 384 via Cochran's formula), paired t-test and regression, then validated the plan on simulated store data.
ResultThe design detected a +20.7% lift at p < 0.01 in simulation; every test reruns here.

Live demo

Edit any store value and the paired t-test reruns. The sample-size and regression tools work the same way.

1 · Paired t-test: monthly sales before vs after the program

2 · Survey sample size (Cochran's formula)

3 · Regression: monthly sales on loyalty enrollment

Monthly sales per enrollment level ($K, simulated). Edit a value to refit the line.

Results

Computed from the project's simulated inputs, not from your edits above.

Sales change per store

Satisfaction vs spending

How it works

  1. Framed the question as testable hypotheses. H0: mean monthly store sales are the same before and after the loyalty program. H1: sales increase after it.
  2. Planned the customer survey sample with Cochran's formula (95% confidence, ±5% margin, p = 0.5), with a finite population correction for smaller customer bases.
  3. Simulated before and after monthly sales for 10 stores and ran a paired t-test, since each store is measured twice.
  4. Added a regression of sales on enrollment rate and a Pearson correlation of satisfaction with spending to cover the secondary questions.
  5. Delivered the hypotheses, sampling plan and test choices as a research design, with the simulated run showing the analysis plan works end to end.

Data

  • Detailed_Quantitative_Analysis_Calculations.csv:
  • All values are simulated: 10 stores' monthly sales before and after the program ($K), 5 enrollment levels with monthly sales, and 6 satisfaction and spending pairs.

    Store data parsed from the file: