Can the presence or absence of different structural elements within a landing page hero predict company revenue?

• Data was assembled based on the top 10 Fortune 500 companies
• Revenue data was based on published data located here: https://www.50pros.com/fortune500
• Hero component data was compiled by looking at each company’s main landing page hero section

##   ID Company_Name Revenue Employees               Website Animation CTA_Count
## 1  1       Amazon $716.9B 1,576,000            amazon.com         1        14
## 2  2      Walmart $713.2B 2,100,000           walmart.com         1         5
## 3  3 UnitedHealth $447.6B   390,000 unitedhealthgroup.com         1         0
## 4  4        Apple $416.2B   166,000             apple.com         0         1
## 5  5     McKesson $403.4B    43,000          mckesson.com         1         1
## 6  6     Alphabet $402.8B   190,820               abc.xyz         0         0
##   Title Subtitle Body_Text Hero_Height_As_Percent
## 1     0        1         1                    100
## 2     1        1         0                     25
## 3     1        0         0                    100
## 4     0        1         0                     75
## 5     1        0         0                     50
## 6     1        1         1                    100

\[ \begin{aligned} \text{Revenue} = {}& \beta_0 + \beta_1(\text{Animation}) + \beta_2(\text{Title}) \\ &+ \beta_3(\text{Subtitle}) + \beta_4(\text{Body Text}) \\ &+ \beta_5(\text{CTA Count}) + \epsilon \end{aligned} \]

We clean the data like this using R:

component_research <- component_research %>%
  mutate (
    formated_revenue = parse_number(Revenue)
  )

\[ \text{Revenue Multiplier} = \log_{10}(\text{Revenue}) \] We can do that in R this way:

final_revenue <- log10(component_research$formated_revenue)

• We do this because otherwise we would accidentally conclude that a Hero components should always look like the component of the company with the highest revenue.

After that we can perform the multiple linear regression to find out which structural elements are associated with company revenue.

regression_model <- component_research %>% lm(
  final_revenue ~ Animation + CTA_Count + Title + Subtitle + Body_Text,
  data = .
)

## 
## Call:
## lm(formula = final_revenue ~ Animation + CTA_Count + Title + 
##     Subtitle + Body_Text, data = .)
## 
## Residuals:
##        1        2        3        4        5        6        7        8 
##  0.07173  0.07419  0.01061 -0.05922 -0.02559 -0.05552  0.12725 -0.01251 
##        9       10 
## -0.11228 -0.01867 
## 
## Coefficients:
##              Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  2.503954   0.133077  18.816  4.7e-05 ***
## Animation    0.192416   0.097563   1.972    0.120    
## CTA_Count   -0.008956   0.005193  -1.725    0.160    
## Title       -0.056093   0.092500  -0.606    0.577    
## Subtitle     0.183523   0.102408   1.792    0.148    
## Body_Text    0.029223   0.091275   0.320    0.765    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.1088 on 4 degrees of freedom
## Multiple R-squared:  0.634,  Adjusted R-squared:  0.1765 
## F-statistic: 1.386 on 5 and 4 DF,  p-value: 0.3871

Note: The revenue multiplier is how many more times having that element from the combination in the Hero increases revenue.

• Having an element both present and absent can positively affect revenue depending on the element
• Having no elements increases revenue 2.43x vs just having a title which increases revenue 2.38x.
• The best combination is having animation with a subtitle and body text and no title which increases revenue 2.84x
• Revenue is negatively affected both by a higher CTA count and the presence of a title

• The data set is incredibly small and correlation is NOT causation. This was done only as a proof of concept as a small school assignment.
• Use results and numbers in this study at your own risk.