Question A:

Multichannel retailers are often interested in targeting customers with the right deal at the right time. Read the attached article titled, "Know What Your Customers Want Before They Do" for a background.

Assume you are the data scientist working at a multichannel retailer (who operates via physical stores, Internet and mobile channels) and that the retailer is interested in offering the "right deal" to its customers. Based on the article you read, answer the following questions.

1) What data (think ROWS and COLUMNS) do you need that will help you model what customers want so that you can target them with the right deal? Be very specific with respect to description of the data.

2) How can multichannel retailers target their customers with the right deal at the right time? Be specific. Explain with model formulation, if necessary.

Question B:

The data set for this exercise is contained in the file elasticity_brands.xls. Each file contains information on unit sales and price for four brands that are named as brand1, brand2, brand3 and brand4. The unit sales have been summed up across several stores, whereas marketing activity variables are weighted averages across those stores.

The data set contains the following variables:

Week = Week number (1 to 75)
Unitsj = Total units of brand "j" sold (j=1 to 4) across stores
Pricej = Shelf price of brand "j" (weighted average across stores)
Featj = Newspaper feature, (1 = brand "j" is featured, 0 = otherwise)
Dispj = Weighted average of in In-store display, (1 = there is an in-store display for brand "j" in a store, 0 = otherwise)
Brands
1 Tide
2 Wisk
3 Era
4 Surf

A) Formulate (Be very specific) and estimate models (log-log model or hybrid models only) to compute and compare own price elasticities for all the brands. Comment on the results.

B) Do the results have face validity (hint: look at summary statistics of the four brands)? Why? What are the managerial implications that one can infer from your results? Be specific.

C) Consider the best models (for Tide and Wisk only) from Part A. For these two focal brands, how can you improve the models to account for the cross price effect (which is the effect of price of a competing brand on the sales of a focal brand)? Formulate models (by extending the best fitting models from Question A) for Tide and Wisk that would allow you to estimate the "cross-price" effect of a competing brand.Explain how you will compute the cross-price elasticities for the two brands, Tide and Wisk. What are the managerial implications? Suggestion: Even if you cannot calculate cross-price elasticities, formulate models and explain the steps that you will undertake.

D) The above models are based on analysis of aggregate data. What are the advantages and limitations of working with aggregate data? From a marketing manager's perspective, elaborate on how disaggregate data can be useful in understanding the effect of a price cut on sales. Be specific. What are the advantages and limitations of working with disaggregate data?

Question C:

You are the CRM manager of a multi-channel retailer (also a direct marketer) of women's apparel, shoes and accessories. You are interested in understanding if customers who regularly shop via online or offline channel are more likely to continue shopping with the retailer. You are also interested in understanding if seasonality plays a role on sales. Provide a snapshot (in rows and columns) of how the data should look like for you to estimate a model that can help you solve this problem. Formulate a model and explain how you will estimate it.

Attachment:- elasticity_brands.xlsx