Auction System Theory
Precious little is known about the technical details of the auction system. In this section we will attempt to guess at some of these details.
Just to be clear, much of this is postulation. Amazon does not publish the details of their auction system. This is our best guess based on our experience and research.
How many Sponsored Product ads per search?
How does Amazon determine how many winners it will accept for each auction? We can't know for certain, but we can make some educated guesses. Let's start by listing some facts:
- The number of Sponsored Products ads per search is not fixed. It varies over time and by search term.
- We know that each winning bid pays just $.01 more than the next highest bid.
- Amazon will never have all the results be paid ads. They want to deliver a good shopping experience, and that means a mix of paid and organic results.
Given those pre-requisites, let's take a look at a few hypothetical scenarios where we have three advertisers bidding on the same keyword:
Scenario 1: Amazon accepts bids from all three advertisers. Bidder C would pay $1.00 per click, Bidder B would pay $1.01 per click, and Bidder A would pay $1.02 per click.
Result: Amazon would show three ads, with a potential ad revenue of $3.03 if a user clicked on all ads.
Scenario 2: Amazon accepts bids from Bidder A and Bidder B only. Bidder B would pay $2.00 per click, and Bidder A would pay $2.01 per click.
Result: Amazon would show two ads, with a potential ad revenue of $4.01 if a user clicked on all ads.
So comparing these two scenarios we can see that Scenario 2 will always be more advantageous for Amazon. They will deliver a better shopping experience by showing fewer ads, and they have a higher potential ad revenue.
Takeaway: Amazon will not always show as many ads as possible, and there is not a fixed number of ads per search term. They will show as many ads as they think will deliver a good shopping experience, and maximize their ad revenue. Furthermore those two goals are not always at odds with each other.
But it is actually a little more complicated than this, the bids are not the only factor that Amazon takes into account when selecting the winners of the auction.
Quality Score
According to Amazon, they also take into account the ads "relevance to the shopping query" while selecting the winners of the auction. This vague language is the only information we have from them on this matter.
It's enough however for us to assume that Amazon is assigning a type of quality score to each ad. This quality score is likely based on the the following factors:
- Search Term Relevance - Based on the product description, how relevant is the product to the search term?
- Click Through Rate (CTR) - The percentage of times an ad is clicked on, compared to the number of times it is shown (indicates search term relevance).
- Conversion Rate (CR) - How often the product sells, compared to the number of times it was clicked.
- Product Reviews - Both the quantity of reviews and the average rating.
We don't know for certain exactly how this quality score is calculated, or how that is used to weight the winners of the auction. But it is likely something similar to the following.
Let's create a table with some products going head to head in an auction. In this scenario they will all be bidding the same ammount.
| Product | Quality Score (QS) | Bid ($'s) | Auction Score (QS * Bid) |
|---|---|---|---|
| A | 1.0 | 2.00 | 2 |
| B | 0.8 | 2.00 | 1.6 |
| C | 0.6 | 2.00 | 1.2 |
| D | 0.4 | 2.00 | .8 |
| E | 0.2 | 2.00 | .4 |
We can assume that there has to be some quality score cutoff, below which Amazon will not show the ad no matter what the bid amount is. If an advertisement is so wildly irrelevant to the search term, it would be a terrible shopping experience to show it. So let's assume that the cuttoff is 0.5. Products D and E have been removed from the competition.
Then let's say that Producs A,B & C have all been accepted to show ads. Product C would pay $2.00 per click, Product B would pay $2.01 per click, and Product A would pay $2.02 per click.
Ad Placement
Let's introduce a slightly more complex scenario.
| Product | Quality Score (QS) | Bid ($'s) | Auction Score (QS * Bid) |
|---|---|---|---|
| A | 0.7 | 3.00 | 2.1 |
| B | 1.0 | 2.50 | 2.5 |
| C | 0.8 | 2.00 | 1.6 |
In this scenario let us assume that all three products have been accepted to show ads. Product C would pay $2.00 per click, Product B would pay $2.01 per click, and Product A would pay $2.02 per click. Does that mean that Product A will be shown first?
We don't actually know for certain that the highest bidder will be shown first. It is possible that Amazon will show the ad with the highest Quality Score or Auction Score first, regardless of the bid amount.
This would be a good way to ensure that the most relevant ads are shown first. So even if you won the auction and payed the highest ammount, you may not actually be shown first, you've essentially just won the privilege of having your ad shown, and then the Quality Score or Auction Score determines you position on the page.
If we combine these considerations with what we conjectured about the number of ads per search, we can begin to understand how complex this problem is.
Takeaways
The nebulous Quality Score assigned to your ads has an incredible impact on your success with Amazon Advertising. It is incredibly important to ensure that your ads are relevant to the targets you are bidding on, and also to keep your product description updated with the terms that are most relevant to your product.