An optimization question we still get pretty regularly is “what should I put in the ‘Subject Matter’ keywords section?”

And honestly, up until recently, I wasn’t 100% sure how to answer that question. So, I finally conducted an experiment. In this video you can see exactly how many characters you can put in subject matter keywords, what indexes, what doesn’t, whether it stitches complete phrases together, whether repetition is necessary, and all the good details you always wanted to know.

Check it out…

Almost every keyword and phrase combination put into the Subject Matter keywords field indexed within 20 minutes.

Even though the video is only five minutes long, I’ll go over some of the finer points here.

First, let’s take a look at the raw data. The Subject Matter keywords that were in the test listing before were:

Mothers Day
Baby Registry
Hudson Baby
SwaddleMe

The listing was indexed for all but the last term. I erased them all and decided to start our test from scratch. I started with a single keyword on the first line. Then added a second keyword to the same line. Once I could see that additional keywords on the same line were indexing, I added keywords in all available spaces.

These are the keywords, followed by the number of characters per line:

baby saxophone quilt liner powder pacifier binky
49
bassoon monkey fidget spinner chandelier guacamole
52
winter salad dressing tongs toy car illustrations
51
autumn carnivore remote control butterfinger shoes
51
kitchenaid nike adidas mrs butterworth xfhghghghgh
51

255 total characters and the system stopped anymore from being added. It appears Subject Matter keywords have the same limitations as Search Terms. Here are the results of the experiment. Each keyword either indexed (Y) or did not index (N) on the listing:

liner – Y
powder – Y
pacifier -Y
binky – Y
bassoon – Y
monkey – Y
fidget spinner – Y
chandelier – Y
guacamole – Y
winter – Y
salad dressing – Y
tongs – Y
toy car – Y
illustrations – Y
autumn – Y
carnivore – Y
remote control – Y
butterfinger – Y
shoes (and shoe) – Y
kitchenaid – N
nike – N
adidas – N
mrs butterworth – Y
xfhghghghgh – Y

 

Almost All Characters Indexed

Regardless of whether the term was relevant to the listing or not, the Subject Matter keywords seemed to index. Here’s an example of this baby blanket listing indexing for the term “guacamole” which is obviously not remotely relevant:

What’s interesting as even keyword phrases that weren’t intact indexed. What I mean is, when putting words together that were NOT in order in the subject matter field, and don’t have any logical correlation together, the phrase still indexed:

 

Stemming Worked….Sort Of

Stemming is a process that essentially means a search algorithm has the ability to reduce a search term to its stem word and index based on that. So, past tense, ing, plurals or other modifications shouldn’t affect the indexing. Amazon claims their algorithm does this. However, this experiment illustrates that it only works for plurals:

I tested both shoe and shoes, even though the term shoes is the only shoe-related term in the fields. I also tested this with terms that were singular in the fields but plural in the search. So, Amazon’s system definitely accounts for the addition of an “s”. However, it does NOT account for “ing”.

 

What Does NOT Index

Almost every keyword and phrase combination put into the Subject Matter keywords field indexed within 20 minutes. However, some big brand names (like Adidas for example) did NOT index. This was not the case for all brand names though. The terms “ButterFinger” and “Mrs Butterworth” both indexed despite being large brands, but “Kitchenaid,” “Nike,” and “Adidas” did not:

And in case you were curious….Mrs Butterworth products ARE sold on Amazon.

So what does this all mean?

It means you get an extra 255 characters to use to index for your product.

Regardless of whether the term was relevant to the listing or not, the Subject Matter keywords seemed to index.

But the next question you may be asking is why that is important.

We will definitely be diving into this topic more deeply in the future, but for now it is important to understand that keyword relevance appears to be the other half of the whole ranking puzzle (sales history/velocity being the first half). The more relevant a listing is for a keyword or phrase, the easier and more likely they will rank for it. However, relevance cannot occur if indexing doesn’t happen first.