Is there a correlation between Google Store Visits and Wizely's Simulated Sales?
TLDR : Yes, absolutely!
One of Wizely's many valued clients has recently been making great use of the Store Visits function inside Google Adwords. Store Visits is Google's proprietary mechanism for connecting advertising impressions with visits to a company's physical location. This is still a relatively new capability that isn't yet heavily utilized in the automotive space.
We wanted to see whether there was a correlation between Store Visits reported by Google, and Simulated Sales reported by Wizely's Dollars and Sense sales attribution model, so we did a quick analysis project.
We started by downloading some raw data: Store Visits at the Keyword Level and Simulated Sales (Assisted Conversions) at the Keyword Level.
We then used "R" to merge the datasets together and plot each individual keyword as a point on a graph, to show the relationship between store visits and simulated sales by keyword.
Next, we computed a linear regression model to see if we could create a function that predicted Store Visits (the harder to determine variable) based on WDNS Simulated Sales, and added that line to the graph.
We then checked the quality of the regression model to see how closely related these two variables are. The R2 (R-Squared) of the model was 0.89 - meaning that 89% of the variability of the data set is explained by the model. In oversimplified language, there is an 89% relationship between these two variables. The P-value is 2.2e-16 (a decimal point then 16 zeros before the 2!) We'll further oversimplify here by saying that P-values less than 0.05 indicate a high degree of correlation, and the P-value here is basically zero.
At this point we've determined that there is an incredibly high correlation between Google's Store Visits and Wizely's Simulated Sales.
After showing that Wizely's Simulated Sales are an excellent proxy for Store Visits, we wanted to further investigate the sales attribution model. Specifically, we wanted to look at keywords that included store visits but no Simulated Sales. (The table below is obscured to conceal the identity of this dealership...)
With the exception of the 2019 Ford Fusion keyword, all of these keywords are either location related or service related. This explains why we would see store visits without simulated sales.
Finally, the two outlying data points were bugging us, so we decided to investigate...
Both of these outlying data points contain the brand ("Ford") and the word "dealership." To a certain extent it makes sense that these keywords would lead to a higher percentage of sales visitors vs. service visitors, however this is something we would like to look into in the future.