Amazon Releases Dynamic Bidding & Bid Adjustments by Placement
Amazon made some big moves in 2018, and as we step into 2019, it appears they have no intention of slowing down. We’re less than two weeks into the new year, and Amazon has already launched a new feature for its Sponsored Products: AI-incorporated bidding strategies.
Machine learning has been a hot topic across the marketing industry for some time now, and we aren’t surprised by this latest roll-out from Amazon. That said, we are still proceeding cautiously with these new tools. As history *cough* Google *cough* has shown us, good results might only come in time.
To help you build your own testing strategy, we are taking a deep dive into these new features and what they might mean for your business and how you can optimize your advertising strategy on Amazon’s Marketplace.
IMPORTANT NOTE FOR ADVERTISERS: All campaigns created before January 2019 have automatically switched over to the Dynamic Bid: Down Only strategy.
What is Changing for Sponsored Product Bidding?
Amazon goes dynamic! Marketers are getting new access to machine learning in Amazon’s platform. Known as “dynamic bidding,” these new bidding strategies will automatically increase/decrease bids based upon whether a sale is likely to convert.
This new AI bidding can look at something the advertiser can’t: the person behind the search query. Amazon could potentially use millions of hidden signals, such as immediate previous searches and past purchase history, to get your ad in front of not only the right searches but the right searchers. This opens up new avenues for marketers to see into the unique buying patterns of Amazon shoppers.
So what are my new bidding options?
Sponsored Products now give marketers the option of using either Fixed Bids or Dynamic Bids:
- Fixed Bids – Amazon will use your exact bid and any manual adjustment you set.
- Dynamic Bids – Amazon will use your set bids as a base level but will adjust your bids either up or down based upon when your ad is likely to convert.
There are two different types of Dynamic Bids you can set:
- Dynamic Bids: Down Only – Amazon will only adjust your bid lower if your ad is less likely to convert.
- Dynamic Bids: Up and Down – Amazon will increase your bid (by a maximum of 100%) if your ad is likely to convert and decrease your bid when it is less likely to convert.
- Bids will not be raised by more than 100% for placements at the top of the first page of search results or by more than 50% for all other placements.
Which Bidding Strategy is Right For Me?
Amazon’s bidding strategies are only as effective as the data powering them. If Amazon sells a lot of your product/category, odds are the dynamic bidding strategy will be effective since Amazon has more data behind these automated decisions. The opposite is also true. If Amazon does not sell much of your product/category, the dynamic bids will have less data to make decisions with and will likely have worse results.
Marketers need to test these bidding strategies to determine which one drives the best results for their business and products. As you begin to experiment, keep in mind these overall best practices:
- 100 clicks should be the minimum amount of data collected before you analyze your data. The more data gathered, the better quality analysis.
- Keep your branded campaigns using the Fixed bidding strategy. Branded has inherently higher ROI, and our goal with a branded campaign is to achieve the maximum number of impressions.
- In general, we do not recommend testing out these new bidding strategies (or other new risky features in general) on campaigns spending more than 30% of your budget. But this spend needs to reflect the size of your business as well.
- For a smaller company, we recommend choosing a campaign that spends no more than 10% of your ad spend.
- For a mid-sized company, pick a campaign that spends no more than 20% of your budget
- For a large company, pick a campaign that spends no more than 30% of your budget.
What Happened to the Bid+ Tool?
Previously, the Bid+ Tool allowed advertisers to increase bids up to 50% if the search result was likely achieve top of search, AKA first page, placement. Amazon has since switched out this tool for a new feature: Adjust Bids by Placement.
Adjust Bids by Placement allows an advertiser to enter up to a 900% increase to your base bid for two placements: top of search (first page) and product pages. Your base bid will apply to rest of search. Unfortunately, we are unable to apply negative bid adjustments yet.
As a reminder, placements according to Amazon are “places across Amazon where your ads may appear,” and they are bucketed into three big categories:
- Top of search – Ads at the top row on the first page of the search results.
- Rest of Search – Ads in the middle or at the bottom of search results.
- Product Pages – Ads appearing on product detail pages or other off search results like the add-to-cart page.
What is My Final Bid? (How to Calculate It)
Knowing what you’re actually bidding can get a little murky when using the new Dynamic Bidding Strategies in conjunction with Adjust Bids by Placement. Amazon calculates your final bid by first adjusting your set bid by the Adjust Bids by Placement modifier before applying the dynamic bid increases or decreases.
Let’s assume you are bidding $1 for the keyword “soccer cleats” and you have a 10% bid adjustment for top of search (first page) with a 15% bid adjustment for product pages. Your spend may look something like this:
It’s important to keep in mind what your Adjust Bids by Placement modifier is when you’re using the Dynamic Bids – Up and Down, since Amazon could increase your bid by as much as 100% for specific searches.
The Bigger Picture
Are in for a bit of deja vu? Amazon’s Dynamic bidding launch may spark less than favorable memories for PPC marketers as they recall Google’s grim first attempt at automated bidding. From then till now, Google’s AI-bidding results have dramatically improved. Will Amazon’s dynamic bidding strategies follow Google’s path of poor to good, or does Amazon have better data powering its AI from the start? Testing alone will tell.