Marketing Web Trends: An Algorithm and Brand Equity Nowcasting Application: An Abstract

Abstract

Web search data are a valuable source of marketing information. Previous studies have utilized Google Trends web search data for economic forecasting. We expand this work by providing an algorithm to combine and aggregate search volume data, so that the resulting data are both consistent over time and consistent between data series. We give a brand equity example, where Google Trends is used to create several brand equity indices of 100 top ranked brands. Monthly Google Trends data are collected from 2008–2017 for each of the 100 brands. We utilize our algorithm to combine the data and ensure that they are consistent, both between the different brand series and over time. The indices are compared with the widely used Interbrand index and the results show good face validity. Google Trends data have been utilized for both economic forecasting and nowcasting. Nowcasting is the process of using current data to predict information that is released after the period for which it is collected. Information of this type include company financial results and economic indicators. We utilize the gathered Google Trends data both for forecasting revenue for the individual series components and for nowcasting a range of economic variables. These variables include the University of Michigan consumer sentiment index, the OECD Composite indicator of consumer sentiment, an index of personal expenditures on durable goods, and the civilian unemployment rate. We describe the importance of out of sample prediction when nowcasting and show how principal component analysis (PCA) can be used to improve the signal to noise ratio and prevent overfitting in nowcasting models. In summary, we have created an algorithm that can patch together different Google Trends searches to create a single multivariate dataset that is consistent both between items and over time. We have given an example where Google Trends is used to create a trends-based index of brand equity and this index is used in a range of forecasting and nowcasting applications. There is scope for future work in combining this index with other measures of consumer buzz, such as social media buzz and with sentiment-based measures of brand approval.