Two tactics top Toys advertisers use to improve growth on Amazon

By: Andrew Holsopple, Analytics and Media Manager

We studied over 1,400 brands in the Toys category in Amazon’s store to discover insights for year-over-year growth in detail page views and new-to-brand customers.

Story highlights:

In this study, we analyzed over 1,400 brands in the Toys category in the United States in 2020. The Toys category includes brands selling products such as construction, games, action figures and collectibles, arts and crafts, toys for infants and ride-ons. We created a composite score of detail page view year-over-year growth rate (DPVGR) and new-to-brand customer year-over-year growth rate (NTBGR), and then identified the top advertising and retail strategies to help increase the composite score with machine learning algorithms.

Advertisers looking to improve their DPGVR and NTBGR should consider:

For more on how we collected our data, see the Methodology section at the end of this article.

1. Top-performing toy advertisers run always-on Sponsored Display and Sponsored Products campaigns


In 2020, 74% of the top-performing toy advertisers ran always-on campaigns for both Sponsored Products and Sponsored Brands year-round.


When using always-on campaigns, we recommend:

  • Keyword coverage: Use category keywords to help reach new audiences higher in the funnel, and then use branded keywords to drive conversion.
  • Sponsored Brands seasonal budgets: Shoppers’ browse and purchase behaviours have peaks and dips throughout the year, and synchronizing budgets to reflect this helps maximize return on investment (ROI). Our analysis showed that top performers in Toys increased Sponsored Brands usage during Amazon shopping events.
  • Do not change promoted ASINs too frequently: To support discovery and relevance, allow sufficient time for support to take effect and do not change promoted ASINs too frequently, such as daily or weekly.

2. Top-performing toy advertisers increase Sponsored Display and Sponsored Brands support during Amazon shopping events


100% of top-performing toy advertisers delivered Sponsored Brands impressions during Amazon events, and 61% delivered Sponsored Display impressions during Amazon events. In addition to delivering impressions, our analysis shows that brands that use Amazon DSP campaigns anywhere customers spend their time show higher New-to-Brand growth.


When advertising during Amazon shopping events, there are a few things to consider:

  • Shoppers come to Amazon to research, consider and buy products. They typically increase engagement before, and remain engaged after, Amazon shopping events. During the holiday season, they often start their research as early as late October and early November, peak during Black Friday and Cyber Monday weekend, and remain engaged throughout the end of December. Advertisers should therefore engage early with shoppers with lead-up packages and consider using remarketing after events to maximize potential.
  • Advertisers can use audience segments to reach customers most likely to purchase during Amazon shopping events.
  • Advertisers can use Sponsored Display and Amazon DSP anywhere customers spend their time.


We first used a supervised model to identify a list of attributes that help improve the composite score among 30+ media and retail attributes. We then used this list of attributes and performed cluster analysis among advertisers/brands, so advertisers/brands in the same cluster are similar in ad and retail attributes, while advertisers/brands in different clusters are different in ad and retail attributes. These attributes are X1, X2 and Xn. (Attributes shown as bubbles on the Visual).
The machine learning algorithms returned four clusters. We ranked these four clusters by their success metrics, compared the top and bottom performing clusters, compared their differences, and identified the key attributes that differentiated their performance on NTB and GV growth.

How does the clustering work?
We created a binary composite score based on DPVR, and then applied an XGBoost classifier to identify which features and with which weights these features best predicted these labels. In doing so, we considered advertising or retail actions as features such as ad product usage intensity and mix, timing of advertising support, tactics of targeting, creatives and placements, customer review counts and ratings, percentage of products with quality product pages and the types of products promoted in ads, etc.

Using the identified features and weights mentioned above, we then applied a k-medoid clustering algorithm to classify advertisers into clusters. Note that we classified advertisers by their actions rather than by the components of their composite score. Next, we ranked the final clusters by their composite scores from high to low.