Data and Retail a Case Study With Amazon

Measures

Targets

Initiatives

Profitable Growth

Return on Invested Capital

Return on Equity

Only accept strong NPV projects

15% ROIC

20% ROE

Simplify the organization structure

Provide an open environment for idea generation and brainstorming

Industry leading innovation

Update product upgrade cycle. Refresh or introduce a product at least once every two years.

Highest Quality products and services

Higher Gross Margin

Invest heavily in R and D with excess Free Cash Flow

Establish strong customer and brand loyalty

Adopt the net promoter score and customer satisfaction rating survey

4% Market Share growth per year

Rewards programs and partnerships with other service providers

Establish a well recognized brand

Strong brand recognition

Become the number 1 or number 2 rated brand in each product category

Invests heavily in marketing and advertising

Question 2

A major retailer such as Wal-Mart would be best served by using a transactional database. Wal-Mart unlike many other retailers has a strong competitive advantage relative to peers in the industry. It competes primarily based on price and value. As a low cost producer, Wal-Mart has enduring advantage that resonates with both middle and lower class consumers. Many competitors, particularly in retail do not have this advantage and are thus forced to compete based on other metrics such as brand, style, or luxury. Due in part to the fact that Wal-Mart is a low cost producer, it will sell large quantitates of merchandise. The last fiscal year alone saw Wal-Mart sell nearly $500 Billion worth of merchandise. Although it had strong sales, Wal-Mart had a Net Operating Profit after tax (NOPAT) of roughly $18 Billion. Over the last 10 years, Wal-Mart has average a Net Operating Profit Before Tax margin of roughly 6%.

Due to these low margins, the company must have a strong understanding of transactions that are occurring. A transactional database, combined with Wal-Mart's just in time inventory management system, the company is able to quickly shift product assortment. Due to the fact that Wal-Mart has such a large volume of transactions that occur in varying regions throughout the world, a transactional database is required.

If a retailer does decide to use both systems, it should do so under varying circumstances. A data warehouse would be best suited for a retailer who has many disparate parts. A classic example would be Amazon who must operate a vast online data system while attempting to establish is brick and mortar footprint. The decision as to where a particular warehouse or distribution center will go is critical. To arrive at the correct decision, management must use a wide variety of sources within the company. This will include aspects such as customer demographics, purchasing behavior and so forth. Taking this information and synthesizing it will allow management to make better-informed decisions regarding product placement and product assortment within a warehouse.

This contrasts starkly with Wal-Mart, who is already established and is simply looking to expand its entrenched position. Amazon is attempting to venture out and create new markets, processes and systems. It is attempting to fundamentally alter to retail landscape. Wal-Mart however is already established and instead is focusing on expanding an already strong presence. As a result a transactional model would be most appropriate for it.

Question 3

I would use linear regression as it allows a practitioner to see clusters of date scattered around a particular area. Although many problems can persist with linear regression, I believe it provides the best means of explaining the overall relationship between loans and default risk. The practitioner must first eliminate non-stationary variables in addition to co dependence. Variable that depend on the proceeding variable can cause problems and errors in the overall regression analysis. However, solutions such as use of the adjusted R squared metric, the Dickey-Fuller test and others can help eliminate these concerns. Regressions, through the use of the R squared metric can help an analyst better determine what percentage of the loan defaults can be explained by variables such as income, debt, or other variables. Regressions are also flexible allowing for multiple variables to be used in an explanatory fashion.