Amazon Keyword Analysis for Sponsored Product Ads

How Statistics Can Help Optimize Your Search Marketing Campaigns

Diego Sanchez

Table of Contents

As a search marketer, it’s critical that you understand how each individual keyword is performing, and one key indicator of performance is keyword distribution. By graphing your keyword distribution, you can discern if your bank of keywords is too narrow or too broad, helping you refine your Amazon marketing strategy. Two statistics that are particularly important for Sponsored Product and Sponsored Brand keyword analysis are Skewness and Kurtosis.


Skewness is a measure of symmetry or the lack thereof. A symmetrical distribution will give you that much sought-after bell curve. A perfect bell curve shows that your keyword pool includes a good balance of investment keywords – keywords that are more expensive than desired now, but are important to your strategy and are likely to return a profit in the long-term – and highly efficient keywords that keep your campaigns profitable (enabling you to sustain the investment keywords). If your keyword distribution shows a bell curve and your target advertising cost of sale (ACoS) is equal to the campaign’s average ACoS, you’ve nailed it.

A deviation from this bell curve symmetry indicates that your campaign is under-optimized, and should drive revisions in your strategy. The keyword distribution – skewed to the left or skewed to the right – can be used to understand how exactly your strategy should be adjusted.


Left Skewed Distribution

A distribution skewed to the left (negative skewed distribution) indicates a majority of keywords are performing above the average ACoS; your campaign is spending too much per conversion. When your target ACoS is below the campaign’s average ACoS, you can make optimizations to improve the efficiency of ad spend, but it may come at the cost of total sales. This occurs because the optimizations will bid down the poor performing keywords (those driving up spend) until the bids are no longer competitive. It restores your target ACoS, but leaves you with a small pool of keywords.

If you can justify keeping the high ACoS keywords, leverage inbound marketing services like paid social and influencer marketing to drive traffic to those keywords. A social ad may refer to your product with a specific, high-ACoS term, for example. As more consumers see the ad, more will use that specific term to search for it. In this way, your inbound marketing efforts can help build relevancy and lower the keyword cost.



A distribution skewed to the right (positive skewed distribution) indicates a majority of keywords are performing below the average ACoS. In other words, the majority of keywords are operating at an efficient ACoS, but are driving few conversions.

If you are consistently winning auctions on these keywords and not running into budget issues, you can use a similar tactic as mentioned above, leveraging off-Amazon marketing to drive additional traffic to these keywords to boost sales volume for those keywords. 


Kurtosis is a measure of heavy or lighttailed distribution within a data set. Data with heavy tails indicate the presence of outliers. Conversely, a lighttailed distribution indicates the lack of outliers. 

Light-tailed Distribution

Kurtosis light-tailed distribution

A lighttailed keyword distribution represents a very targeted keyword mix. This may indicate missed opportunities because your current keyword selection is too refined. To correct this issue, add a greater variety of keywords in your campaign. This widens your net, helping you capture more sales, although it may increase your ACoS. This strategy works best when in conjunction with off-Amazon marketing, for the aforementioned reasons.

Heavy-tailed Distribution


A heavytailed distribution represents a broad keyword mix with a significant number of outliersIn other words, you’ve cast too wide a net. These outliers spread your budget too thin and detract from campaign performance. 

To resolve this issue, you need to identify the most egregious outliersIf they happen to be terms that you consider highly relevant to your product, don’t cut them from your ad campaigns. Instead, use social media marketing and influencer marketing to drive external traffic to these terms and build up their relevancy. If the terms are not highly relevant to your product, cut em loose. 

Simplify the Process

Using these two measurements of keyword distribution will help you better understand and improve your Sponsored Product and Sponsored Brand campaigns. However, repeating this process for multiple campaigns quickly becomes time-intensive and tedious, especially as you repeat it on a monthly, weekly, or daily basis. 

To simplify the process, you have two options: 

Excel Plugin

Use the Data Analysis ToolPak in Excel. This plugin is a fantastic tool for users who spend some of their time managing ad campaigns in Excel. The plugin does enables you to effortlessly create summary statistics which includes skewness and kurtosis.

Instructions for installation can be found here.

Ad Management Software

Excel may not cut it for users who are managing many campaigns, dealing with large budgets, spending hours in Excel even with the plugin, or are generally looking to further professionalize their operation.

If that’s the case, it’s time to invest in a dedicated ad management tool. Kaspien began as a third-party seller in 2008, and we’ve gone through the full process of using increasingly massive and complicated Excel spreadsheets and macros to optimize our Sponsored Product campaigns. Eventually, we outgrew Excel. Literally. So, our engineers set to work  on creating a custom solution. Over three years in the making, we’ve built and refined Kaspien AdManager until it became a search marketer’s dream tool. With AdManager, we’ve achieved:

  • 10% avg. ACoS and 10:1 avg. ROAS on over $50MM in ad spend
  • Generated 30% sales growth on Amazon for retail and agency partners
  • Reduced ACoS by 50% for key partners 

AdManager is currently available as a managed service, where our experts use the software to manage your ad campaigns for you. A self-service model will be released in March 2020 for users who want to handle the process entirely themselves. 

Enjoy What You’re Learning?

If you found this content helpful, consider subscribing to our weekly newsletter! We regularly cover the latest Amazon news and strategies.

More About the Author

Diego Sanchez