Traditionally, when marketers want to know how people feel about something, they commission a survey. With enough responses, it is possible to get a pretty decent view of the target population’s beliefs or opinions. But surveys come with inherent limitations, among them respondent bias and recall issues. And without a robust response, it is difficult to know whether the results truly represent the views of the entire target population. New research has shed light on an alternative approach: using “share of search” as a more objective and dynamic indicator of brand health that could potentially eliminate some of the pitfalls of a survey-based approach.
What is Share of Search?
Share of search has been popularized in recent years by marketing effectiveness experts Les Binet and Peter Field. They have been studying new ways to track brands based on what people actually do instead of what they say; specifically, what people do online. Binet coined the term share of search to refer to the proportion of Google searches for a particular brand out of the total number of Google searches within that brand’s category:
Share of Search = Organic searches for Brand XSearches for all brands in category
Using Google Trends data, brands can get high frequency search data going back to 2004. By pulling the search data for competitors, it is possible to aggregate all of the searches to create the total category searches, and calculate the share of search metric. Not only can you get an idea of what your brand’s proportion of searches are right now, you can see how it has changed over time – providing a valuable look at the growth or decline of a brand’s relevancy in the eyes of the consumer. For more information on this metric from Les Binet himself, this video shares how they’ve been using this technique.
From Search Data to Brand Health Insights
So what would a typical implementation of share of search look like? To begin with, it’s important to understand how search metrics are related to other brand health and commercial outcome metrics for your business. For example, does share of search mirror the up and down movements of survey-based brand health metrics such as Unaided Awareness or Familiarity? What about sales, or Website Traffic? To answer these questions, it can be helpful to use two techniques: correlation analysis and regression modeling.
Correlation Studies: Do Metrics Move Together?
A correlation study helps to answer the question, “Do these two metrics move together over time?” In other words, is a spike in search volume in November reflected in sales data as well? Does growth in share of search over time mirror the movement in Aided Awareness?
If these metrics are highly correlated, the answer to both questions would usually be yes. We also want to know the timing of the movement. Perhaps a spike in search volume in November wouldn’t result in a spike in sales in November, but it would clearly be seen in December. Sometimes there’s a lag between two metrics, so testing lags in both directions for each correlation can help identify these timing effects.
Value of Regression Modeling
While correlation results help to determine which metrics move together and their timing, it doesn’t account for other factors which may skew the correlation strengths. The two most obvious are long-term trends and seasonality. If two metrics share the same seasonality, they’ll typically move together even if one is not directly related the other. If both metrics are showing a long-term trend upwards, they’ll both likely be moving together most of the time, because they’re both moving up. To complement the correlation results and control for these other factors, we recommend utilizing multiple regression modeling.
By building predictive models with and without search metrics as predictors and studying the resulting model diagnostics, we can determine whether the search metrics are in fact predictive of the other metrics. Including predictors for the period of the year and a time index for long-term trends help control for those other factors which might skew a correlation. By giving our model the number of branded searches in November as well as enough data to know the long-term movement of the dependent and independent variables and what November typically looks like compared to other months, it should be possible to predict the metric in question, assuming there is a real relationship.
Case Study: Applying Share of Search
Transparent Partners recently conducted these analyses while working with a Fortune 500 retailer, and we were encouraged to see results which supported Binet and Field’s research. By establishing these relationships between search and brand health metrics, we were able to recommend Share of Search as a proxy for brand health metrics, with the added bonus that search metrics are available at a much higher frequency and without the limitations of a survey-based approach. Furthermore, we found share of search to be a leading indicator of brand health metrics, meaning that our client can detect changes in consumer sentiment sooner than ever.
Some additional benefits can be harvested once the relationship between search metrics and brand health or commercial outcomes is established. It’s possible to include search volume or share of search in a marketing mix model to provide a more dynamic and responsive signal than traditional brand health trackers, allowing marketers to identify key drivers of search lift such as media investment and channel mix. Search volume can also be used as a demand signal, providing a strong indicator of consumer interest. It can also serve as a valuable early signal of brand momentum or campaign effectiveness.
Future Proofing Measurement with Share of Search
Finally, Share of Search and Search Volume can also help future-proof your business. We know that Google is not always the search platform of choice, especially in global markets, but these metrics could be tracked consistently through many other platforms such as Youtube, TikTok, Meta, and LLM’s such as ChatGPT or Gemini). This allows marketers to have standard global metric definitions while allowing for market-level flexibility and localization.
We’re just scratching the surface of how search metrics can be utilized, but it’s clear there’s value here to be harvested. So if you or your company are interested in exploring how share of search can bolster your marketing measurement, you know who to search for!
Dan Weinbeck, Sr. Principal
