Study : Coca-Cola
Coca-Cola realizes a lot of investments in its communication with its target audiences and on all the available media. Everyday, many mentions on social networks can be linked to the brand.
The problem : How to get a statistical indicator which allows to analyze the correlation between the media campaigns, financial investments and impacts on social networks, according to different classifications ?
To resolve this issue, we have analyzed the media investments' database as well as all social networks mentions made on the Internet.
Temporal consistency and data cleaning:
For this work, we have made the data consistent, considering a single time scale for all data analysis. Second, we have cleaned the data, correcting the values that have not properly been collected by their internal software.
Creation of an algorithm which has allowed us to analyze the global mentions in relation to the media campaigns.
Based on widely justified statistical assumptions, we have developed an algorithm that has associated media investments data with all social mentions linked to Coca-Cola made on the Internet.
Correlation analysis between global mentions and advertising investments:
Thanks to some correlation statistical tests, we have been able to highlight the trend of the campaigns investment and social mentions on the Internet in relation to the advertising campaigns lunched by Coca-Cola, according to their status in each week.
The initial hypotheses:
All this quality statistical analysis is based on assumptions that must have been justified. For this work, we have considered two conditions that allowed us to have valid results and at the same time no loss of vital information, namely :
- an uniform time scale for all datasets
- a suppression of the least relevant variables for the analysis
From now on Coca-Cola can get a new ROI indicator of its media campaigns, namely : which campaign generates the best ratio investment / social mentions. The brand can also analyze the seasonality or the existing correlations between media campaigns, their subjects and social mentions on the Internet.
In addition to the work made on the data, 10h11 has proposed a visualization that has been greeted with enthusiasm. Here is an example (above) of the graphic and interactive return we have realized in order to present our study and visually understand the goals that come from.