Here's what we learned from 200k Sponsored Products campaigns

By: Qianyun Ye, Media and Analytics Manager, and Ashton Brown, Technical Writer

In this 2020–2021 study of over 220,000 Sponsored Products campaigns of China-based global apparel sellers on Amazon, we compare advertising strategies of top- and lower-performers. We then use this comparison to derive actionable insights advertisers can use to explain the value proposition for using Sponsored Products, combined with recommendations China-based sellers in the apparel vertical can use for improving Return-on-ad-Spend, conversion rates and Click-Through-Rates.

Story highlights:

In this study, we analysed over 220,000 Sponsored Products campaigns of China-based apparel sellers selling globally (12 locales, including: United States, United Kingdom, Germany, France, Japan, Canada, Italy, Spain, India, Australia, United Arab Emirates and Mexico) between August 2020 and July 2021. To perform our analysis, we grouped Sponsored Products campaigns into five clusters, with Cluster One being the most successful and Cluster Five being the least successful in terms of Return-on-ad-Spend (ROAS), conversion rate (CVR) and Click-Through-Rate (CTR).

Our analysis finds that top-performing China-based sellers (Cluster One) had a 4.6X higher ROAS, a 3.7X higher CVR, and, a 3.2X higher CTR.



Higher Return-on-Ad-Spend


Higher conversion rate


Higher click-through-rate

To provide advertisers with actionable insights, we used machine learning to analyse advertising and media attributes that contribute more, or less, to clicks and conversion rate. We then identified which attributes have the largest positive impact on ROAS, CVR and CTR.

This article provides insights/best practices on the key attributes or strategies by quantifying the degree to which top-performing China-based sellers (Cluster One) and lower-performing China-based sellers (Cluster Five) have adopted each key attribute or strategy.

For more on how we performed this study, see the Methodology section at the end of this article.

Top performers run Sponsored Products campaigns for nearly 90 more days than lower performers

In the period observed, top performers ran Sponsored Products campaigns for a median of 237 days, while lower performers only ran them for a median of 149 days. For best campaign performance, we recommend that apparel sellers use Sponsored Products year-round. Doing so helps to ensure that the brand is always present and always more time for campaign optimisation.

Number of Sponsored Products always-on campaign days





Top performers were nearly 20% more likely to use dynamic bidding in Sponsored Products campaigns

Our analysis found that dynamic bidding is an effective tool for engaging customers, specifically as it pertains to ROAS, CVR and CTR. Over the 12-months studied, top performers used dynamic bidding in 86% of campaigns, compared to lower-performing campaigns where dynamic bidding was used just 67% of the time.

Percentage of Sponsored Products campaigns that incorporated dynamic bidding





Getting started with dynamic bidding

Advertisers should begin with Sponsored Products campaigns. There are also three considerations for using dynamic bidding:

  • Consider testing dynamic bidding strategy on a stable campaign: When testing bidding strategy, it is best to choose a campaign that is stable. Meaning, campaigns that have been running for at least two weeks and shows a positive relationship with conversions.
  • Consider limiting changes while testing: We recommend that advertisers limit changes while testing strategies, so that they can attribute the difference in performance to a specific change.
  • Consider comparing strategies on different campaigns: When creating new campaigns, be sure to compare the new bidding strategy against an existing, stable, strategy (when possible).

Prime delivery is available for 10% more products in top-performing Sponsored Products campaigns

The third insight-driven recommendation is to specifically increase Click-Through-Rates by increasing the number of ASINs, or products, that offer Prime delivery. We use machine-learning models to identify products with a high likelihood of being clicked on if advertised. We’ve also found that having the Prime badge alongside advertised products helps increase the chance of a customer clicking on advertisements.

Our analysis found that of the available ASINs (unique serial numbers), in top-performing Sponsored Products campaigns, 59% were available for Prime delivery, compared to 49% availability for lower-performing campaigns.

Percentage of total ASINs offering Amazon Prime delivery





Things to consider when offering Prime delivery

  • Consider enrolling Sponsored Products ASINs into the Fulfilment by Amazon (FBA) programme.
  • Consider enrolling Sponsored Products ASINs into the Seller Fulfilled Prime (SFP) programme.
  • Consider combining both FBA and SFP programmes to increase Prime delivery offerings.

Top performers maintained a 2:1 ratio of targeted keywords to negative keywords in Sponsored Products campaigns, while lower performers maintained a 1:1 ratio

Finally, in previous articles we’ve found that including negative keywords on product ASINs or serial numbers was an effective tool, in the case of Chinese apparel sellers (selling to a global audience), we have found that targeted keywords are also an effective tool for ensuring that ads are relevant to the right customers.

In the 12-months studied, top performers used targeted keywords in 68% of Sponsored Products campaigns and negative keywords in 32% of campaigns, compared to 49% and 51% in lower-performing campaigns respectively.

Targeted keywords vs. negative keyword targeting





Things to consider when using keywords

  • Check your existing campaigns’ reports to find terms that you should use negative targeting on. Lower CTRs, higher spends and lower conversion rates are good indicators of targeting that is underperforming and could become excluding keywords.
  • Check the performance of your excluding keywords frequently to learn and optimise your campaigns with the ones that actually work for your brand.
  • Check for exceptions. For example, non-branded keywords used to increase awareness of new products launched could have poor performance, but should not be used as excluding keywords because they are in fact the correct audience for products in the apparel category.


As seen in our analysis, in combination with our supervised machine learning model, we identified four tactics China-based sellers in the apparel vertical can use to improve ROAS, CVRs and CTRs: (1) Run Sponsored Products ads year-round; (2) Opting into dynamic bidding; (3) Consider enrolling Sponsored Products ASINs (unique serial numbers) into fulfilment programmes (FBA, SFP) so that they display the Prime badge when advertised; (4) increasing target keyword and negative keywords per unique ASIN.


We first used a supervised model to identify a list of attributes that help improve the composite score among 20+ media and retail attributes. Specifically, we followed a five-step process to create a suite of success metrics including: Return-on-Ad-Spend (ROAS), conversion rate (CVR), and Click-Through-Rate (CTR), and then identified the top advertising and retail strategies to help increase the success metrics with machine learning algorithms.

  • Select brands: In this study, we analysed 222,853 apparel’s Sponsored Products campaigns delivered in 12 worldwide locales between August 2020 and July 2021. Each campaign was launched by endemic Chinese advertisers i.e., Chinese advertisers that sell products on 12 locales include United States, United Kingdom, Germany, France, Japan, Canada, Italy, Spain, India, Australia, United Arab Emirates and Mexico.
  • Create success metric: Calculated based on three metrics: Return-on-Ad-Spend, conversion rate and Click-Through-Rate.
  • Identify effective ad or retail actions: Once the success metrics were defined, we identified the top campaign actions to help increase the success metrics with machine learning algorithms. We leveraged a gradient-boosted decision trees model to identify the most important campaign actions that contribute to the success metrics. This method helps us understand which advertising actions are the most important to drive strong success metrics.
  • Group brands: Campaigns by their success metrics and assigned them into five clusters. Leveraging more than 20 advertising actions, we then analysed their impact on the success metrics. Advertising actions include SP campaign duration in days, dynamic bids, negative keyword tactics, etc.
  • Compare brand groups: identified what strategies top-performing brands (Cluster One) use to increase clicks and conversions, compared to strategies that lower-performing brands (Cluster Five) are or are not using.