Non-Amazon Search Volume Methodology

The methodology for how we use Amazon search data to assist in estimation search volume for other retailers.

For non-Amazon retailers (e.g., Walmart, Target, Kroger, etc.), our Search Volume Estimates are statistically-modeled using a variety of harvested data from the retailer sites, external data related to site-specific search, traffic estimates, and more. The data available directly from the retailer varies, and currently, the keyword rankings are derived from the Amazon search terms.

Each week, we derive the relevant search terms for each retailer and use search rankings provided by Amazon using the methodology below:

1. Look at the rank of the various search terms, and the categories of the top items clicked on for each term.

2. We apply the ranks of the term to each retailer, and weight terms to categories based on the retailer industry or specialty (Example: More weight will be put on terms whose top clicked category is Grocery, when applying search volumes to a grocery chain like Kroger, but no weighted change for a big box store like, Target).

3. Using a variety of internal and external sources, we establish the estimated traffic of each retailer, and apply the traffic to the rank of the search term.

Assumptions:

1. The overall search volume is directional because it is based on the assumption that search behavior by retailers will be similar, but not exactly the same.

2. Weighting is applied to specific retailers that are preliminarily in specific categories. Target/Walmart will have the same ranks for most terms, while Lowes/The Home Depot would have terms showing similar ranks, with extra weighting on terms that are home improvement related.

3.The overall assumptions were made so that we could provide an idea of movement of terms and scale. These assumptions can show limited differences between similar retailers (Walmart v. Target), but should show differences in scale between the two, as well as giving high-level understanding of how a keyword is trending. It also helps show how category specific retailers might have terms indexed differently than big box retailers.