Business Intelligence Analysis

Business Dilemma: Marketing

This is the second post in the business dilemma study. The first post can be found here.

Assume that a small portion of the marketing sampling data was gathered on Under Armour customers and their app users, they plan to run a marketing analysis of a promotional event. What promotional event do you recommend? Why?

Given that Under Armour is working towards getting back to their roots (Chen, C., 2017, October 31), a contest that involves making athletes better would make sense. Under Armour’s mission is to help athletes perform better, one way they can measure this is by looking at the data from their apps. Half the battle of becoming a better athlete is just showing up to the gym, track, or where ever an athlete might train or exercise.

By athlete, it should be assumed to be any person who is seeking to improve their physical performance, not just sports athletes.

The suggested promotional event should be hosted on Under Armour’s fitness apps and marketed through social media. The event would be one where fitness app users would log a workout and take a “selfie” to post on a social media outlet such as Instagram, Twitter, Facebook, or a short clip to post on YouTube with a unique hashtag for Under Armour’s promotional event in their favorite Under Armour gear. For the unique hashtag, #uatitan might be good for the contestants to use on social media. Users would be scored by the number of workouts logged. The contest would have multiple prizes:

  • Lowest scoring prizes: A discount code for Under Armour products.
  • Medium scoring prizes: A choice of one piece of Under Armour gear out of a small selection.
  • High scoring prizes: A choice of one piece of Under Armour gear out of a large selection.

A random contest participant will be chosen at the end of the contest to get some Under Armour gear with an all-expenses-paid trip with their plus-one to watch Dwayne “The Rock” Johnson’s “The Titan Games” (The Titan Games., n.d.) where the winners get to do a meet-and-greet with Dwayne Johnson, Under Armour’s brand ambassador.

Prizes would be awarded weekly for a one-month period. Contestants can climb the prize brackets as the contest works towards the finale where a single, random contest winner is chosen.

Once a prize bracket has been reached and that bracket’s prize has been awarded the contestant can’t win that prize again and has to make it to the next bracket to win another prize.

The reasons for this promotional concept are:

  • Additional free advertising with customers and users of the fitness apps promoting Under Armour gear.
  • Collecting more data through the fitness apps that might be used to learn how to better serve customers through understanding how they exercise to design better products.
  • Help promote Dwayne “The Rock” Johnson as their brand ambassador.
  • Generate sales by offering discount codes at the first prize bracket.
  • Reinforcing their image and commitment to making athletes better.

What is your target customer pool and how do you plan to reach them?

Under Armour has over 22,600,000 downloads through their mobile apps just through the Android IOS app store alone[1]. More downloads are on the iOS app store. These potential customers are already using these apps in which the promotion will be marketed, the mobile fitness apps. Under Armour could also run a social media campaign to reinforce the contest as well as attract more non-app users to join the contest by downloading one of their fitness apps.

The fitness apps Under Armour currently have are:

  • Under Armour Record (Under Armour Record — Apps on Google Play., n.d.).
  • Under Armour (Under Armour — Apps on Google Play., n.d.).
  • Run with Map My Run (Run with Map My Run — Apps on Google Play., n.d.).
  • Walk with Map My Walk (Walk with Map My Walk — Apps on Google Play., n.d.).
  • Map My Hike GPS Hiking (Map My Hike GPS Hiking — Apps on Google Play., n.d.).
  • Map My Ride GPS Cycling Riding (Map My Ride GPS Cycling Riding — Apps on Google Play., n.d.).
  • Endomondo (Endomondo — Running & Walking — Apps on Google Play., n.d.).

The following table shows how the data on app users can be categorized to determine the likelihood of choosing a random ideal grand-prize winner. This table can also help determine disproportionate outcome probabilities. By understanding the ratio of males to females it can help Under Armour estimate how much of the prize products should be allocated towards females and males as well as age range appropriate products for this promotional. Also, understanding the ratio between Under Armour customers versus non-customers can help Under Armour estimate new potential customer conversions. The total data sample size is 100,231 app user records. Records were taken from the same fabricated dataset from the business dilemma presented in here, the first in this short series of business dilemmas.

( fabricated data for study)

What is your projected cost for this promotional event?

$100 million to $150 million is a conservative range for this promotion, based on half of the contest participants reaching the top tier prize bracket. Contest participant numbers coming from the mobile fitness apps (fabricated data for study). Under Armour’s marketing budget was over $560 million last year (2017) and has been rising (Under Armour: Marketing spending 2017 | Statistic., n.d.), so a promotional contest costing $150 million would seem to be within reason.

The random grand-prize winner and their plus-one costs are negligible in relation to the costs of the Under Armour gear giveaways as medium and high range bracket prizes. Giving discount codes as the lowest tier bracket prize will help offset a portion of the giveaway prizes.

Promotional material and social media marketing to build awareness of the contest may not need to be extensive since Under Armour’s fitness apps have such strong numbers but should still be part of the marketing campaign to reach a wider range of potential customers.

You are a strong advocate of the promotional event. How will you justify the cost benefits?

Based on the sample ratio of customers to non-customer in relation to Under Armour’s fitness app users, there is a good amount of potential new customers. There is also the probability that existing customers will become more active in their buying habits for this promotion and help build brand loyalty.

According to the data sample:

  • Customers app users: 50.26%
  • Non-customer app users: 49.74%

Suppose some legal concern has been expressed that a disproportionate number of sampled customers are men. If a customer is randomly selected from the data pool, what is the probability that the customer is a woman?

According to the categorical data counts from the section above, the likelihood of a female customer being selected would be:

Total female customers in data sample = 25,191

Total customers in data sample = 50,375

The probability of a female customer being chosen as the grand-prize winner is:

As unlikely as it happens to be, the likelihood of the grand prize winner is a male or female is about 50%.

If a potential customer is randomly selected, what is the probability that the person is a woman?

Again, taking from the categorical data counts from the section above, the likelihood of a potential female customer getting chosen would be:

Total female non-customers = 25,077

Total non-customers = 49,856

Again, the data for a female in this scenario has a relatively equal chance for either a male or female being selected as the grand-prize winner. In this case, a female has a slightly higher probability of being selected, being about a 50.3% chance.

What factors might enter into the apparent discrepancy between probabilities? Please provide some mathematics and/or a visualization.

Under Armour has been working on building up their female clientele, as the company is seen predominately masculine (Pasquarelli, A., 2017, July 19). The reason the discrepancy between male and female ratios could be to the sample size is that perhaps over 100K is not enough to determine if there is a discrepancy related to gender. It could also be that females and males use these fitness apps in about even numbers among the app users.

Figure 2.1: Kernel Density Estimation (KDE) of all app users. Users’ age (x-axis) referenced against user fitness log entries (y-axis) over the last 365 days.

Seen in figure 2.1, the center graph is a topical KDE similar to elevation lines on a topical map. The top KDE line is in reference to the user age density and the KDE line to the right is in reference to the total amount of log entries by users. The Pearson correlation (-0.0017) is very low, indicating that there is no correlation to the data.

The KDE covers the entire graph indicating that the data is uniformly distributed. Even though the data is uniformly distributed, density areas are still noticeable.

Figure 2.2: Kernel Density Estimation (KDE) of male app users (Left) and female app users (Right).

Again, the data is uniformly distributed across the board for both males and females. In general, it would appear that male app users have more density areas while females appear to be more uniformed. Either way, it seems that both male and female potential contestants are fairly evenly distributed.

In this case, a legal concern related to a sexist agenda in either direction seems to be an unlikely event.

Considering one person at random wins the grand-prize of the contest, what is the chance of each of your categories in the table provided in the “Target Customers” section?

( fabricated data for study)

Probabilities of the table above are shown in the table below between males and females:

In the table of probabilities, the discrepancies between males and females are almost non-existent across all of the categories. It should be noted that females have a slightly higher probability than males.

What factors might enter into the awarding of the grand-prize other than random selection? How do you ensure your drawing is fair with no possibilities of discrimination? Please provide an analytical visualization of your choice.

Looking at the probabilities from the total data population, the “Youth” are significantly lower than the other categories and might be factors that influence the probability of some people having a less likelihood of being selected for the grand-prize.

The easiest and maybe even the responsible solution would be to require contestants to be 18 or older to enter the contest.

Given the p-values from figures 2.1 and 2.2, there is strong evidence for the “null hypotheses”. What this means is that there are no significant trends to be found within the data. This is why the KDE graphs in figures 2.1 and 2.2 are completely covered. That being said, it would be difficult to find a discrimination factor when choosing a grand-prize winner. To further prove this point here are histograms of ages of app users (Left) and fitness logs (Right) show uniformed distributions:

Figure 2.3: Age histogram (Left) and fitness log histogram (Right)

Under Armour is thinking about implementing Bayesian Paradigm (for this study) in their marketing. Please provide some mathematics and/or a visualization. What do you recommend?

As shown in the probability table above, under “Contest Analysis”, the probability distribution is remarkably uniformed, besides the youth category. Normally, implementing Bayesian Paradigm would be a good idea, but with this data, it is intuitive even to non-technical readers that the probability of the outcome based off of any of the qualifying categories is fair and even.

The uniformed distribution shown in other categories, again above in figure 2.3 further illustrate the equal probabilities all qualified contestants have at winning the grand-prize.

The lowest qualifying probability on the category table is a non-customer male, as expressed in the equation below:

Seeing that the lowest qualifying probability still has a pretty fair chance of being selected, the need for a Bayesian Paradigm may not be a need in this instance.

Based on your analyses above and given the test data, working company, and industry type, do you think Under Armour should expand to Allegheny County, Pennsylvania? Why or why not? Your recommendation may not change the executive decision, but you should take your stand and argue your case.

Based on the geographical information discussed in this post and the information discussed above, Under Armour should expand to Allegheny County, Pa. Given that Allegheny County, Pa. has a large amount on fitness app users that are not active customers combined with the potential of increasing the company’s customer base through this promotional event. Under Armour could have a good chance of gaining a strong foothold in that area.

Knowing that around 50% of app users are not active customers gives Under Armour a large pool of reachable potential customers. If Under Armour is going to make a successful expansion to another part of the country, then it should be where the community already fits the culture Under Armour is trying to build. Allegheny County, Pa. is the most primed location that fits that description. If Under Armour is going to expand anywhere, that location is the best chance for success.

[1]: (Under Armour Record — Apps on Google Play., n.d.), (Run with Map My Run — Apps on Google Play., n.d.), (Walk with Map My Walk — Apps on Google Play., n.d.), (Map My Hike GPS Hiking — Apps on Google Play., n.d.), (Map My Ride GPS Cycling Riding — Apps on Google Play., n.d.), (Endomondo — Running & Walking — Apps on Google Play., n.d.).


Chen, C. (2017, October 31). To Save Itself, Under Armour Is Going Back to Its ‘Scrappy’ Roots. Retrieved from

Endomondo — Running & Walking — Apps on Google Play. (n.d.). Retrieved from

Map My Hike GPS Hiking — Apps on Google Play. (n.d.). Retrieved from

Map My Ride GPS Cycling Riding — Apps on Google Play. (n.d.). Retrieved from

Pasquarelli, A. (2017, July 19). Under Armour Chases Women’s Business in New Campaign. Retrieved from

Run with Map My Run — Apps on Google Play. (n.d.). Retrieved from

The Titan Games. (n.d.). Retrieved from

Under Armour — Apps on Google Play. (n.d.). Retrieved from

Under Armour: Marketing spending 2017 | Statistic. (n.d.). Retrieved from

Under Armour Record — Apps on Google Play. (n.d.). Retrieved from

Walk with Map My Walk — Apps on Google Play. (n.d.). Retrieved from




IT Professional, Technical Support, Software Developer, and Business Intelligence Professional

Love podcasts or audiobooks? Learn on the go with our new app.

Packaging the New Normal

Selling on eBay Tools and Tips from Official eBay Sources — Selling Tools

Social media are wonderful Personal Branding tools. Which one to choose?

How to Use the Green Screen for Live Video Recordings

How fear-based marketing invented bad breath.

How to Drive a Successful Affiliate Marketing Campaign

Week 15 — bitter goodbyes, sweet hellos and a hell lot of work to boot

9 terrible mistakes to avoid when creating web content

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Liu Maumasi

Liu Maumasi

IT Professional, Technical Support, Software Developer, and Business Intelligence Professional

More from Medium

Google developers course: Android basics in Kotlin (Part 6 Icons)

Sending data to another activity is that easy — Android?

What is Android and Why to use it?

Android 13: Notification Runtime Permission