Launch announcement

We launched the ability for advertisers to select a optimization strategy with cost controls, which helps to simplify the campaign management process for advertisers using Sponsored Display. This launch allows advertisers to set performance based parameters that remove the guesswork out of selecting an exact bid or target. This helps advertisers of any size and with any budget achieve their goals with the help of Amazon’s 1P data and machine learning capabilities.

Before this launch, advertisers needed to translate their desired goals into efficient bids and targets across different cost types. In reality, this turns out to be the biggest issue driving lower scale and campaign effectiveness. By translating goals into bids and targets using machine learning with an optimization strategy, Amazon can help advertisers achieve their goals through continuous optimization.

For any campaign, advertisers can apply a metric-based rule such as cost per 1,000 ad views (vCPM) and cost per click (CPC). Optimization strategies will then utilize machine learning models to define the bids or targets, all while adjusting base bids up and down to increase results (clicks or view-able impressions) and working to adhere to the input of the cost control.

Advertisers specify an advertising objective in the form of their optimization strategy, the products they’d like to advertise, a daily budget, and an average cost they are willing to pay per click or 1k view-able impression. There are two important levers to influence marketing goals and costs today.

• Targets selected: Different targeting allows advertisers to reach different audiences, and ads may be more relevant to some than others. When selecting targets, they should be 1) relevant to the advertised products and 2) broad enough to help advertisers reach as many shoppers as possible. The machine learning models that serve ads will be adjusting bids in real time for every opportunity to ensure the cost stays at the desired value, wherever the ad is rendered, while only paying when it works.
• Bid values: Bids signal the most an advertiser is willing to pay for an action (view or click) on each of the targets. With how the Amazon Ads auction’s work, advertisers only pay the least bid amount needed in order to win, so they don’t need to worry about strategizing about how to set bids.

When advertisers use optimization strategies, our machine learning models help choose the right targets and bids that can drive a desired action at the specified (maximum) cost and budget. We want each campaign to achieve as many possible actions as long as the costs are under the specified target cost per action (CPA), without spending more than the daily budget. Bids and targets are optimized by models that look at the past campaign performance and similar campaigns: if costs are too high, bids are reduced, and if they’re below the target, they are increased to get more total scale. Bids and targets will be quickly adjusted based on changes in competition (for example, another brand is targeting similar audiences with higher willingness to pay, or traffic changes due to seasonal factors) to ensure advertisers stay within their desired range, all while getting as much scale as possible.

• Seller Brand Owners
• Vendors

• Advertising console
• Amazon Ads API

Optimization strategy with a cost control is now available for Sponsored Display through the Amazon Ads API. For full technical details, please see our updated Sponsored Display documentation.