Using Share of Search as an indicator for market share in the FMCG industry
Search data is a strong indicator to understand trends in sales, market share and social, economical or cultural shifts.
Today, 81% of us search the web when we are looking to buy something. It could be ingredients for a recipe, a new underground drill, that car we dream of. or a certain dress or jacket. Research by My Telescope, Les Binet, James Hankins and Google shows that there is a clear correlation between the number of searches and actual market share.
To illustrate these findings you can look at the graph comparing Share of Search and market share data for a well known FMCG brand in Sweden.
Use Search data to understand market drivers
In figure 1 the starting point of Share of Search was 8% and the market share 8%.
In figure 2 one can see that the search volume has decreased and the trend for market share data shows the same trend.
The benefit of using search data is that it's easy to follow and spot positive or negative trends. The downside is that the trend Is the same but not the exact percentage numbers.
The study shows that search data trends have a very strong correlation with market share trends.
So, in other words, by following how many searches that are done on a brand, product or topic we get a strong indicator of where the sales, market share and trends are going.
Methodology:
Using purely the brand name to establish the Share of Search data, gave correlations on about a quarter of all brands. For the other brands we had to generate a combination of the brand name with another search term to reach correlation levels.
The lag between the share of search and the market share was either 0 or 1 week difference.
Using different search terms could dramatically increase the correlation between market share and Share of search. One brand moved from a correlation of 0.69 to a correlation coefficient of 0.75 just by adding the words “ice cream” to it.
We took the sales numbers of all ice-cream brands in the Nordics from beginning of 2018 until start of 2021. With slightly over 3 years of data of about 190 different overall brands or particular products in Sweden we started investigating the correlation between Share of Search and their market share.
The market share data was created per brand on the bases of the total market of all 190 brands. The weekly sales numbers were taken on a 12 month rolling average in order to flatten any exceptional peaks that were driven by promotional activities, which are important for the ice-cream market.
For each brand we started by gathering the search volumes and trends on the direct brand name. That share of search data was then correlated with the market share data looking at different lags to see if there was any delay in between the search and the sales. Later on, for particular brands, we investigated the use of other search terms to see whether it would improve the correlation.
The investigation for the optimal search terms in terms of Share of Search continues, but it is clear from the initial research that Share of Search is valid metric as a proxy for movements in market share. This even for a category where search volumes generally are low.