Deep Dive Into Amazon Indexing

search

I am sure most people that started selling awhile ago have already tested all the nuances of Amazon’s indexing spiders, but since things have changed recently I was prompted to run some tests again. With the ever evolving Amazon landscape it is important to revisit the basics to truly get an idea of what is different. Or at least give you some idea of aspects of the algorithm that may have changed.

With that said, I’ll run you through my tests and the results I got from it so that you can compare notes from months ago.

What Do the Spiders Crawl?

About two years ago conventional wisdom (tests run by sellers as well as Amazon’s own words) stated that the only portion of a listing that factored into ranking was the title and back end search terms. Since then the layout of listings as well as what appears in search results has changed quite a bit.

Run any search in a web browser and you’ll notice over to the right (in certain layouts), under the review stars, there is a section titled “Product Features” and below that some text with your search terms italicized. This shows the snippet of listing that had your search terms in it.

If you dive a little deeper, you’ll notice an interesting trend. It would appear that listings with at least one term in your search phrase in the bullets outranks those that don’t have any of the terms in their bullets but have them in their description.

For example, if you search for “exercise bands” you’ll see all of the listings on page one have either some form of the word “exercise” or “band” in their bullets until you get down to the last listing on page one, which has no bullets but has the term in the description:

Search results have snipped below stars
Search results have snipped below stars
This listing has no bullets. That snipped came from the description.
This listing has no bullets. That snipped came from the description.

And then search for “teak table” and you’ll see the same thing. Pages one and two are full of listings with the terms “teak” or “table” in the bullets, heavily outweighing the number of listings with the terms only in the description:

Page one search sample. Only TWO listings on page one were missing terms in the bullets and only had them in the description.
Page one search sample. Only TWO listings on page one were missing terms in the bullets and only had them in the description.

This would lend some credence to the observation by some experts that bullets hold more “weight” than back end search terms. While this may NOT be the case for PPC, it appears possible for organic searches.

What About Stemming?

Amazon has been saying forever that their algorithm accounts for stemming. What does this mean? It means it recognizes search terms that are misspelled, in past tense or plural. This has largely been contested by experts and experienced sellers, but I now come baring proof.

If you search for the term “baby sling” you’ll see this result in the search, right at the top:

Stemming clearly supported
Stemming clearly supported

Notice the italicized search terms. “Babies” …..plural. If you run a search for “baby carrier” you’ll see this one with the term “baby’s” as well:

Stemming also accounts for possession.
Stemming also accounts for possession.

This would show that Amazon’s algorithm does, in fact, account for stemming. However, when you search for one term and then search for its plural, you’ll see a different number of search results. That is odd.

For example, “water bottle” has 1,860,446 results. However, “water bottles” turns up over 5 million. Curiouser and curiouser. Also, the sponsored products results are different.

There appears to be an obvious answer to this. The algorithm DOES account for stemming BUT it prioritizes listings that specifically include the plural term in the copy or search terms.

And What About Those Search Terms?

I did a little research on my own listings, since I don’t currently have access to the back end search terms of anyone else’s. I discovered that my products do, in fact, rank for specific long tail search terms that I have in the back end, regardless of whether the whole phrase is somewhere else in my listing (but at least one word always appears to be in there somewhere).

However, when I search for phrases comprised of individual keywords that are NOT listed as whole phrases in the back end, my listings still rank, but not extremely well (at least for the terms I checked).

It would appear that when Amazon changed the search term capabilities and updated their optimization suggestions to state that keywords should be listed in a logical order, they meant that the spiders would be more likely to index search PHRASES. This would mean that the old advice we got to never repeat keywords in the search terms is no longer valid.

The Takeaway

Use keyword phrases in your search terms. Put those terms in your bullets where you can, and in your description otherwise. Use plurals and other tenses as well to jump ahead of automatic stemming. And of course, test, test and test.

  • Popo

    Thanks for the article, Anthony. Really informative.

    You mentioned that we can repeat keywords in the backend search terms, and spiders would be more likely to index search phrases.

    One question – if we’re selling those exercise bands for example, we can put in the search terms the term “exercise bands” as many as we want?

    For example, exercise bands for men, black exercise bands, exercises bands, etc.

    It won’t harm the search results?

    Thanks.

    • anthonynlee

      Yes. It would appear this is the case.