Website visitors make choices with every click and conversation. How does a business anticipate what they are going to do next or inspire visitors to select their product or offer?
Our client needed insights to help them more effectively engage visitors and target their web ads. Currently, predicting conversion for a particular website visitor is not well understood and credit for conversion is often attributed to the last ad viewed or clicked on by the consumer. However, this fails to account for the consumer’s decision stages and other factors which drive conversion.
Fractal Analytics was approached to build a predictive model exploring which factors influence visitor conversion relative to the timing and frequency of ads, page placement and other advertisement attributes.
We worked closely with our Client to identify all of the data sources and elements that were available for modeling and establish the sampling methodology. Our data scientists then applied a systematic approach to analyzing the data, beginning with principle components analysis (PCA) to identify which variables/factors have the most influence on conversion.
Next we constructed a logistic regression model using the variables identified in PCA to determine the relative influence of various attributes or components on conversion.
The results of the modeling process demonstrated that other variables besides “last ad” viewed do indeed influence website visitors’ conversion probability. These predictive attributes provided our client with key insights which allowed them to formulate more objective and profitable marketing strategies. Our client was able to more effectively target web ads (e.g. creatives, ad parameters (flash, non-flash), dimensions, timing, duration and mix) across websites to maximize online revenues and improve marketing ROI.
One of the world’s largest media communications services companies (this client has requested anonymity).
More effectively target and maximize returns from web advertisements.
Principle components analysis and custom logistic regression model to identify factors influencing conversion.
Actionable insights lead to effectively targeting web ads and improved marketing ROI.