Improving customer experience for Mall of America’s gift stores

Client:

Mall of America is the largest mall in the United States with 2.5 million square feet of retail space attracting over 40 million customers annually.

Business Problem:

Improving the customer experience with reorganizing inventory for three Mall of America’s gift stores to increase product sales, reduce employee attrition, and gain more clarity on customer buying patterns

Solution delivered:

Reduction in 40% inventory with exploratory data analysis mining for popular products across different categories to redesign the product placement across stores helping customers with finding products that are frequently bought.

Tools used:

1. Python

2. Tableau

3. MySQL

Mall of America, an international attraction is located in Bloomington, Minnesota a.k.a the Antarctica of the United States. (In fact, when this article was being written, Minnesota made the international news for being hit by a polar vortex with temperatures falling below -40 in both Celsius and Fahrenheit. )

I worked with Mall of America as an Analytics Consultant where the business problem was to increase the overall customer experience with:

1. understanding customer buying patterns to reorganize product shelf

2. looking for seasonal and cyclical trends that could be used for a competitive advantage

3. reducing staffing issues with balanced workloads and appropriate training.

Clearly, there are many elements that influence the overall experience of the customer. Also, seasonal products, sporadic promotional offers, competitor stores, data availability, and other such factors add more complexities to the challenge.

This was a perfect time to fall back on the legendary pyramid principle by Barbara Pinto (a go-to option for consultants) to limit scope creep and define deliverables that are feasible to deliver in a 6-week consulting project.

Considering these factors, I decided to define the project deliverables as:

1. Increased clarity on customer purchase patterns and with an interactive Tableau dashboard to track customer segments by season, store, category, product clusters, price bucket and transactions by time of the day and month

2. Mining popular products across stores to either:

a. Reduce the inventory of infrequently sold items

b. Increase the visibility of infrequently sold items with product rearrangement

Using Pareto Principle (20-80 rule), I was able to mine for popular products and found that 95% of the sales are generated by 60% of the products in the store.

Things that I learned:

1. A data science project provides business value by augmenting business acumen. This is the primary reason why stakeholders prefer to use techniques that are white-box (interpretable) and not black-box (hard to understand and interpret). No analytical model can tell you the best business decision unless it is combined with business acumen, domain knowledge, and subject matter expertise.

2. An informed decision is the result of a stimulating discussion of stakeholders from diverse teams within the company to brainstorm and account for finer details that could be affected with a decision. (reminds me of a great podcast that I recently heard on mastersofscale featuring Airbnb’s CEO Brian Chesky)

3. Requirements gathering is the most crucial part of the project plan and every detail including the way you deliver your solution can affect your analysis as a success or failure in the minds of the stakeholders.

I hope you enjoyed reading this article as much as I enjoyed writing it. Stay warm and visit Mall of America especially if you have kids. They have a great amusement park.