SPAG (Amazon Ads)
Concerning Amazon Ads, there was a time, that measurements were aggregated at the AdGroup level, necessitating using SKAG - Single Keyword per AdGroup (meaning: Having a singular keyword and product per AdGroup). In 2024, Amazon improved reporting.
However, at the AdGroup level, results are not attributed to individual products. E.g.: You can see the orders and sales per keywords, but if there are multiple products per AdGroup, you wouldn't be able to tell to which products these orders/sales belong.
Solution: Use SPAG: Single Product per AdGroup.
Review: Why SKAG is still good for Google Ads
Concerning Google Ads, SKAG is almost always a good idea. In a AdGroup (using SKAG), the following things come together:
- One keyword
- One landing page
- Two or more advertisements.
There are two main reasons for using SKAG:
- Tight match: To make sure that there is a tight match between those three entities: The more keywords you throw in, the more it gets delituted. E.g.: All those keywords should come back at the landing page in a sufficient amount
- Measurement: The real reason: When there are multiple keywords in an AdGroup (and multiple advertisements), it becomes really difficult to measure the performance - Because now two factors are changing at the same time: A Cartesian product or combinatorial explosion of Keywords x advertisements. Performance of individual keyword-advertisement combination is possible, but just harder. By fixing the keywords, this becomes much easier: There is just a long list of AdGroups and you can much easier see how they are doing, without knowing if you are compairing apples with apples or something differently.
SKAG: Why this is differently for Amazon Ads
- Amazon Ads doesn't use Quality Score to the extend that Google Ads does → A tight match between the different entities in an AdGroup, is hardly relevant
- There are no advertisements. Just products, which are equal to the landing page in the Google Ad situation
- Keywords
So, when both products and keywords are more than one per AdGroup, there is a combinatorial explosion. Additionally, where in Google Ads it is still possible to evaluate such AdGroups, in Amazon Ads, you cannot: There is simply no way to attribute sales to keyword to product.
Use SKAG instead of SPAG?
Rather than having SPAG, could you equally well opt for SKAG? In either case, the combinatorial explosion is avoided.
The answer seems to be no, but I don't really understand why. The general problem is, that sales are not attributed to products and keywords at the same time, which would make it a symmetrical situation, I would think, but appearantly that's not the case:
| Adgroup | Remarks |
|---|---|
| One AdGroup with 20 keywords and 1 product | You know exactly for which keyword you have a sale for this product |
| One AdGroup with 1 keyword and 20 products | Ahh, it's indeed the same situation |
Yes: You can use SKAG instead of SPAG - In theory. But practice seems differently. Why is that?
In Google Ads, you control which ad shows, what it says, and where the user lands. So SKAG is about tightly matching keyword → ad copy → landing page.
In Amazon Ads, the "ad" is always the product detail page (you can’t customize it per keyword), so the creative is fixed. That shifts the optimization mindset from: "Which ad is best for this keyword?" to: "Which keywords work best for this product?"
That’s why SPAG feels like the natural unit of control for Amazon: You don’t write custom copy — you optimize listings.
So even though both SKAG and SPAG reduce the keyword × ad (or keyword × product) tangle by isolating one entity, SPAG fits the Amazon Ads mental model better, because the ad is inherently tied to the product:
- SKAG = better fit for Google Ads where you control the message
- SPAG = better fit for Amazon Ads where the product is the message.
Conceptually, they do the same thing: fix one variable to isolate performance — but SPAG reflects the reality of how Amazon ads work, so it’s the better, more intuitive term.