Resources - Blog

Rules-Based vs. AI-Driven Bidding: What Amazon Advertisers Need to Know

Discover the differences and benefits of bidding automation technologies for Amazon Advertising to determine which solution is best for your business. By Natalie Taylor April 30, 2020
Rules-Based vs. AI-Driven Bidding: What Amazon Advertisers Need to Know

Machine learning. Algorithmic technology. Artificial intelligence. 

When it comes to automation solutions, there are a number of terms used loosely and interchangeably across a variety of industries, but what do they really mean? Is a rules-based algorithm the same as a machine-learning algorithm? More important, how can you determine which type of solution is best for your business needs?

For sellers and brands on Amazon, an effective advertising strategy has become imperative to maintaining visibility and driving conversions on the increasingly competitive marketplace. 

Amazon over the last few years has expanded and optimized its slate of advertising features and offerings, and will certainly continue to do so to encourage increased investment and adoption. As Amazon Advertising grows progressively complex, so too will the technology you use to automate and optimize your advertising strategy and performance.

To determine which type of bidding automation solution is best for your business, you first must understand the nuances and complexity of Amazon pay-per-click (PPC) advertising.

Understanding Amazon PPC

PPC is a keyword-based auction where sellers bid to place an ad for a specific keyword. Although simple from a user’s perspective, Amazon PPC is driven by extremely complex algorithms. These algorithms decide which ad to display based on the keyword bids.

On Amazon, the auction is a second-price auction, meaning that you will never have to pay what you actually bid on that keyword but rather you pay 1 cent more than the second-highest bid on that keyword. For example, if you bid $1.50 on a keyword and your competitor bids $1, you will pay $1.01 to beat your competitor’s bid so that your product appears for that keyword or search term.

However, the actual bids of other advertisers vying for the same keyword are not visible to you. Therefore, you must decide on a bid using your own metrics, including impressions, clicks, and conversions. 

Impressions indicate how many times your ad has been displayed, either within the search results or on a product detail page, essentially revealing how many times you won the bid.

Clicks indicate how many times shoppers click on your ad. While for Sponsored Brands, the creative may influence your click-through-rate (CTR), Sponsored Products ads do not allow advertisers to customize the creative. Therefore, the CTR is determined by the product’s relevancy to the search term, your market share, and the price of the product compared to the prices of other similar products appearing in the same search results page. The more relevant and attractive the product, the higher your CTR.

Both impressions and the CTR are critical metrics to determine the bid. If the search term is rising in popularity and bids increase, you might experience a decline in impressions if your bid remains the same. Conversely, if the CTR is low, that could indicate the search term may not be right for your product and, essentially, you are wasting ad spend.

Also important, conversions reveal the effectiveness of your product detail page. If the conversion rate is low, your page may require content optimizations, such as improving your product title, description, and bullets for SEO, or generating more, positive customer reviews. A poorly optimized product detail page can result in wasted ad spend if a shopper clicks on your ad, but leaves your listing and does not make a purchase because of an unoptimized product detail page and subpar shopping experience.

Impressions, CTR, and conversions are the key metrics for Amazon’s bidding algorithm. Furthering the complexity is the fact that, for each product, the advertiser typically bids on multiple keywords — each with a different history and conversion metrics. A bid price decision is not only based on individual keywords but also requires aggregate consideration around how to allocate your budget across various keywords to achieve optimal performance. This is a complex algorithmic task that often requires extensive data-crunching and analysis beyond the scope of human capability.

Although manual bidding will typically lose to a computer-driven bidding system, most computerized systems do not fare well either due to their reliance on simple historical performance — the bidding and conversion history of a single keyword. 

This approach may work well for a small set of keywords that generate a lot of traffic, but it will not function well for a much larger number of keywords that each generate few impressions. The fewer the number of impressions, the less data the computer system has to make accurate bidding decisions.

Most advertisers have a small number of keywords that experience high levels of traffic and a much larger number of keywords that each generate few impressions but, in the aggregate, can generate incremental sales.

Long-tail keywords consist of those that, albeit generating fewer impressions individually, often contain the highest-performing keywords on a per-click basis. Failing to adequately bid on these long-tail keywords results in a missed opportunity to optimize your campaign performance.

Indeed, manually managing your Amazon PPC strategy is a timely and complex undertaking. Below, we explain the types of algorithmic solutions you can leverage to automate the process and help enhance your performance.

Natalie Taylor
About the Author

Natalie Taylor is the content manager at Feedvisor, where she oversees and executes on the company's content marketing strategy. Prior to her work at Feedvisor, she wrote for a B2B supermarket magazine, focusing on merchandising and marketing trends in the grocery industry.

What Is Rules-Based Bidding? 

A rule-based algorithm is designed to take action based on pre-set rules and conditions, ranging from simple to complex, that you enter into the system. In regard to Amazon PPC, a rules-based bidding system requires the user to make manual decisions for changes made to campaigns by defining “if/then” rules that would trigger bid and keyword changes based on your pre-set bid and keyword parameters. 

Changes and optimizations only occur once your specific conditions have been met. For instance, you could set a rule that, if your advertising cost of sale (ACoS) falls below your pre-set target after a certain number of clicks, then the system would automatically increase your bid by a pre-set percentage. 

Rules-based bidding can be beneficial for advertisers who want more control over their PPC strategy. These solutions are often easy to set up and integrate, and can save you time compared to manual campaign management.

However, because a rules-based solution is restricted to respond solely within the scope of your defined criteria, the algorithm will be unable to respond to any new conditions or market changes, which can negatively affect your performance. 

What Is AI-Driven Bidding? 

AI is a subset of computer science that focuses on machine-driven intelligence. Essentially, in AI-powered systems, the machines mine, interpret, label, and react to data. Unlike rules-based systems which are static, AI-driven systems are continuously updated and grow more powerful as it learns and responds to new inputs and data. 

An AI-driven bidding solution continuously analyzes search data to identify trends and keywords that generate the most conversions. The technology then uses that data to continuously optimize your bids in real time, maximize your ad spend efficiency, and drive ROI. 

Feedvisor’s optimization solution for Amazon Advertising employs a complex set of AI-driven algorithms that optimize bids, even for keywords with sparse impressions and conversion data. The AI technology leverages deep-learning models that consider the actual semantic relevance of the keyword to the product that is being advertised, the attributes of the advertised product, and the performance of the overall campaign, among other factors.

This includes the novelty of the keyword in order to ensure the technology gives new keywords a chance to prove their worth while maintaining tight cost control, thereby allowing for rapid exploration of long-tail keywords.

These deep-learning models are created by training the AI technology on a myriad of data points from various campaigns, products, and keywords. Feedvisor’s access to hundreds of thousands of campaigns, millions of products, and tens of millions of keywords gives the solution an unprecedented ability to create the most precise and reliable models that are used by AI in real time to produce the most optimal campaign performance.

To learn more about Feedvisor’s AI-driven solution for Amazon Advertising, get in touch with us at feedvisor.com/connect.

Feedvisor’s End-to-End Platform Helps Sellers and Brands Drive Growth on Amazon

Request Demo

This site uses functional cookies and external scripts to improve your experience. You may change your settings at any time. Your choices will not impact your visit.

I agree to receive cookies

Click here to read our Cookie Policy.