Instructions

1.   This is an open book exam. You are allowed to use the class material (slides, handouts, your notes, etc.). However, you are NOT allowed to use the Internet except for accessing Canvas.

2.   When applicable, please include your R code as well as the final results in your answer.

3.   You are NOT allowed to communicate with anyone regarding the exam.

4.   Submission is online through Canvas. After submitting your answer document, please make sure that you can see it on Canvas, and that the file is correctly uploaded.

5.   You have 2 hours to complete the exam. Your exam starts from when you download the file, and the 2 hours must be an unbroken (consecutive) period of time – you cannot spend one hour on Friday and one hour on Sunday. Please pick a 2 hour window of time when you know you will be able to work on this exam and won't be busy with other things.

6.   The exam has two parts for a total of 40 points.

7.   By taking this exam, you're agreeing to Foster School's code of conduct: you

acknowledge that you are a part of a learning community at the Foster School of Business that is committed to the academic standards of honesty, respect, and integrity, and that      you adhered to these standards while completing this exam.

Part I: Conceptual Questions

Nordstrom wants to change the design of its landing page for some of its search advertising campaigns. Before rolling out the new design, they want to run an experiment to make sure that  the new design increases the conversion rate by at least 10%. The current landing page has a conversion rate of 6%, so they want to know if the new design increases it to at least 6.6% (i.e., a 10% improvement in the conversion rate).

1.  Nordstrom wants 80% power and a significance level of 0.05 in its experiment. They also want an equal number of people seeing the old landing page vs. seeing the new landing page. How many users do they need in each group?

2.   How would you set up this experiment? Be clear about how you would assign people to the different conditions, what information you would track, and how you would determine an outcome at the end.

3.   Rather than running an A/B test, another approach would be to change the website temporarily to the new landing page for everyone, and then to compare the results for the next week (with the new landing page) vs. the previous week (when we had the old landing page). What are the pros & cons of this approach?

4.   In general, Nordstrom does not bid on its own brand name for search ads; i.e., they do not buy search ads for the keyword “Nordstrom” . Part of the justification for this is that their brand name is unique, so people searching for “Nordstrom” can just click on the Nordstrom.com organic search result link instead. What are the pros & cons of this approach?

5.   A manager at Nordstrom thinks that the new landing page might be more successful with customers on the West Coast, while the old landing page would be more successful with customers on the East Coast. Do you think this is plausible, and how would you test it?

6.   A manager at Nordstrom thinks that the new landing page might be more successful with customers who see it more. People might react poorly at first because it’s different than   what they’re used to, but as they use it more, it will yield better outcomes than the old landing page did. Do you think this is plausible, and how would you test it?