Decomposition & Regression Analysis

Decomposition

We can expand on our understanding of Flourish's conversion and brand image potentials with an AR store through decomposition. In decomposition, we are breaking down the uncertainty in our decision into smaller and more workable parts to help expand understanding of different variables.

In our decomposition, we looked at potential influences that could be impacting conversions for Flourish's AR store the most. Our objective for entering an AR store space is to enhance brand image and will be analyzed by our BI model which includes conversion rates as a variable.

Influence diagram of influencing factors and their relationships:

Product (Virtual or Physical): What product customers are interested in purchasing may be significant to the conversion rates. If the customer is seeking an innovative way to search, view, and purchase products for their real life home, they could use AR to visualize the product in their space and lead to a sale. However, if the store only offers virtual products, the AR storefront may not drive sales and therefore not increase overall conversions.
Repeat Buyer: Loyal and previous customers to Flourish may be more inclined to use and shop in the AR store knowing the brand and their quality product offering. Flourish may see higher conversions with customers who have previously made a purchase with the store.
Device Type (Used to Access AR Store): If a customer uses a certain device to access the AR device it could have an impact on their overall experience, ultimately influencing their decision to make a purchase or not. A desktop may not be as ideal for an AR store front and could lead to an unsatisfying experience. However, a customer using a mobile or VR device, like a headset, to access the store can experience a more immersive and interactive shopping experience, potentially resulting in higher sales. Device type that a customer chooses to use may be influenced by their age, as we predict younger generations will be more likely to use VR while older generations may resort to using their desktop.
Time Spent in Store: The time spent in the store is likely correlated with their likelihood to purchase. When a customer spends more time engaging with products in the AR store, there can be more opportunity for increasing conversions; although it may not be a direct linear relationship as longer visits do not guarantee higher sales or more purchases.
Customer Income: Given Flourish is a high-quality, expensive, luxury store customer income may impact the overall conversions. Flourish may attract high-income customers who are willing to make large purchases. However, as an AR store is more accessible to a wider audience and offers a unique shopping experience, customers with lower income but interest in experiencing the AR experience may visit the store with no intention to actually purchase providing no additional purchases for the store. We predict that customer income may also influence the time spent in the store.
Customer Age: Similar to customer income, age could influence if a customer will make a purchase in the AR space. Younger generations are more comfortable and willing to use technology like AR and may be more willing to engage in a digital shopping space. Whereas, older generations may not be willing to make a purchase in an AR store and prefer the traditional brick and mortar experience. We would expect customer age to be influencing additional variables like device type used to access the store and previous experience to AR technology/storefronts.
Previous AR Experiences: If the customer has used AR previously for shopping, or other experiences, they may be more willing and able to use and trust the technology to make relatively expensive purchases. Customers with AR exposure previously are likely to be tech-savvy and at ease with visiting a virtual world. The more familiar a customer is with the AR technology, the more likely they may be to spend time in the store exploring products and potentially making a purchase.

Regression Analysis

For the regression analysis, we can collect primary (Flourish's existing customers) and secondary data (market research, competitor insights, industry statistics, consumer surveys/focus groups).

Example Regression Data:

For our initial regression analysis we begin with the following variables:

Dependent

Conversion Rate

Independent

Product Type
Repeat Buyer
Device Type
Time Spent in Store
Customer Income
Customer Age
Previous AR Experience

When combined, we receive the below regression model to see the relationships among the dependent factor, conversion rates, and the various independent variables listed above.

Conversion Rates = α + β1ProductType + β2RepeatBuyer + β3Mobile Device + β4VRHeadsetDevice + β5TimeSpentInStore + β6CustomerIncome + β7CustomerAge + β8PreviousARExp

Regression Results

From our regression statistics we can make a few conclusions:

About 49.74% of the variance in customer conversions can be explained by the independent variables in the model (R square value).

Product Type, Mobile Device, Customer Age, Time Spent in Store, and Repeat Buyer were the independent variables found to have a statistically significant relationship with customer conversions. From the coefficient, we can see that product type (virtual or physical) has a significant impact on conversions with physical goods driving higher likelihood of purchasing. Flourish managers will want to take this into serious consideration when deciding the product mix in their AR store. Repeat buyers also tend to generate better conversions compared to first time buyers, suggesting that Flourish will want to focus marketing efforts for their AR store with loyal customers who have shopped with them previously.

Additionally, when customers access the store through a mobile device, there is a greater chance of conversion, telling Flourish managers they will want to make their AR storefront accessible and optimized for mobile devices, they may choose to use a VR app as well, but they are going to see greater conversions with a successful mobile app that allows customers to see products in their home using their cell phone.

And finally, time spent in store and customer age both are statistically significant to conversion success, suggesting Flourish will want to target their marketing to younger shoppers and develop features in the AR store app that keeps customers engaged for longer amounts of time, as focusing on both these variables will lead to greater conversions.

We checked the following assumptions to validate results:

Normality of residuals (no patterns in residual plots)
Linearity (linear relationship between dependent and independent variables)
No influential outliers (no large or small data points skewing results)

We had a few potential candidates for interaction terms, of where one variable might depend on another variable:

Time spent in store & customer income: We hypothesized that customers with higher income may spend more time in the store, but their purchasing behavior may include making fewer purchases, however the purchases may be more intentional and of a higher price. The time spent in the AR store depends on the customer's income level. We determined this interaction was not statistically significant when added to the regression analysis.
Customer age & previous AR experience: AR experience impacting conversion rates could be different depending on the customer's age. Younger people are more likely to engage with AR, while older individuals may not be interested. We determined this interaction was not statistically significant when added to the regression analysis.
Customer age & mobile device used to access store: We predicted that younger age customers were more likely to use a mobile device and make a purchase. We suspected older customers using mobile devices are less likely to convert. We determined this interaction was statistically significant when added to the regression analysis.

After running our initial regression with all the independent variables and then testing for interaction terms, we determined our final regression with statistically insignificant variables removed.

Dependent

Conversion Rate

Independent

Product Type
Mobile Device
Repeat Buyer
Customer Age * Mobile Device

When combined, we receive the below regression model to see the relationships among the dependent factor, conversion rate, and the various independent variables listed above.

Conversion Rate = 0.31365 + 0.337(ProductType) + 1.029(MobileDevice) + 0.468(RepeatBuyer) - 0.016(CustomerAge*MobileDevice)

Regression Results

with Interaction Terms

From this analysis we can make some conclusions about business strategy for Flourish and their AR store. We concluded that physical goods drive significantly higher conversion rates than virtual goods and Flourish should consider prioritizing physical goods in the AR storefront, this aligns well with the decision analysis that also indicates a higher mix physical goods as driving overall brand image. We also know that repeat buyers are more likely to make purchases compared to first-time buyers, so Flourish can consider offering loyalty deals, rewards, or personalized promotions for existing customers to encourage them to shop in the new AR space. They can also focus AR storefront marketing efforts on existing customers. And finally we can conclude that using a mobile device to shop is significant for Flourish's conversions. Customers who use a mobile device are more likely to convert, however the interaction term between customer age and mobile device tell us this is true particularly of their young customers. If they want to see conversions with their older customers, Flourish will want to make the store accessible on various devices (desktop, tablets, etc.) to ensure they are not isolating their older customers by only being available as a cell phone app.

Page updated

Report abuse