Deep Dive Into Indexing & Ranking, As Defined by Amazon

Michael Lee
11 min readAug 13, 2018

By Anthony Lee, Sixleaf COO

Before we dig into the meat of this topic, it is important to define a few terms. Amazon, and particularly their terms, can be extremely confusing. Most definitions rest somewhere between explicitly explained and vaguely speculated. I’ll do my best to keep the facts straight and the logic intact.

What Is Indexing?

Indexing is when Amazon’s system (or any search algorithm for that matter) catalogs a listing as searchable for specific terms. This is referred to as being “crawled” by Amazon’s “spiders.” Essentially, when the spiders reach the words you’ve input in your listing, they will index them and your listing should then be available for search for that term.

While there is no doubt serious complexities in Amazon’s search algorithm, it appears that, as far as indexing is concerned, this is a binary process. Meaning, a keyword is either indexed, or it isn’t. This is by no means the end of the story for a specific keyword and how the system handles a listing in regards to its search terms. Far from it, this is just the first step.

It is also important to note that not all terms entered into a listing will be indexed (more on this later). Just know that, your listing will either be indexed for a keyword, in which case it most likely will be available for search, or it won’t, in which case, it most likely will NOT be available for search.

How Does Indexing Affect Ranking?

It seems in just about every instance (with rare and few exceptions) a product cannot be ranked for a keyword in search if it is not indexed for that keyword. This would mean that indexing is an integral step in the process of ranking. This is the reason why so many softwares are now offering services that check indexing.

If a listing suddenly loses indexing (more on that later) then it can negatively impact ranking, dramatically and without warning.

A Little About Browse Nodes

Once upon a time we sellers had much more control over the subcategories our listings went into. We could easily force irrelevant browse nodes into a listing and win bestseller badges for subcategories that got little to no volume (badge hack circa 2014).

However now, Amazon has changed the way they assign browse nodes. Browse nodes are now automatically assigned based on item_type_keyword. And every listing only gets one item_type_keyword.

In fact, here is an excerpt from an email conversation I had with a member of the catalog team:

“However, according to recent change in policies, our business and development and seller performance team has decided to assign only one browse path to one ASIN. You might see ASINs with multiple browse paths listed on Amazon but we have a dedicated team working on removing multiple browse paths and correcting the catalog. This might take some time to be amended considering the size and volume of our catalog.

Please understand that browse nodes are assigned based upon the item type keyword assigned to an ASIN. As per a new change in the Amazon Catalog teams policy, associates are no longer able to assign or remove browse nodes assigned to an ASIN. We can only assign a single item_type_keyword to an ASIN since our systems do not allow us to add browse nodes but we can replace an item type keyword to change the browse node.”

So, item_type_keyword has a single, or possibly multiple browse nodes assigned to it.

Here’s the kicker though; browse nodes dictate what keywords you can rank for. Sure, you can index for a keyword or key phrase. You may even seem highly relevant for it because it is in your title. However, if the browse node associated with your listing doesn’t allow it, index or not, you won’t rank.

That means that, not every term you index for will be rankable.

To give you an example, let’s say you sell a grill brush that has a bottle opener on the end of it. Your item_type_keyword is grill brush. Amazon assigns the browse node associated to grill brushes to your product.

However, in the title you mention that it is also a bottle opener. BUT, despite that, your product is NOT a bottle opener…at least, in Amazon’s eyes. So, even though you are indexed, and even if you are told you are highly relevant for the term (based on the fact that it is in your title) you will not be able to rank for the phrase “bottle opener” unless you change your item_type_keyword.

What Is Relevance?

Keyword relevance is basically how Amazon’s system determines how “rankable” for a term a listing is. Obviously it is more complex than that (as mentioned above, browse nodes can supersede or otherwise affect relevancy), but in simple terms, this is the only thing about relevance that is important for sellers.

The more relevant a listing is for a keyword, the more capable of ranking for that keyword the listing is. So, if all other ranking factors are equal, a listing will win out over another listing for a keyword it is deemed more relevant for.

How Does Amazon Determine Relevance?

This is where things get interesting.

First, if you dig into the Amazon Advertising API, under the ItemSearch Values, you’ll see the “call” for “relevance” actually defined (partially) by Amazon themselves. It states the following are factored in relevance for a search term:

  • How often the keyword appears in the description (keyword density).
  • Where in the listing it is found (title keywords are more relevant).
  • How closely the keywords occur (do they create key phrases within the listing).
  • How many purchases are made through the searched keyword.

So, according to Amazon, keyword density, phrase-ology, and location all matter in considering how rankable for a keyword a listing is (relevance). BUT THAT’S NOT ALL!

Recently I had a….situation, we’ll call it (more on that later).

I contacted Seller Central and communicated back and forth with knowledgeable members of their catalog team. In one of my communications I caught this snippet:

“…keywords that are indexed for search through sponsored ads will be changed to the keywords updated for the ASIN as soon as the sponsored ads is closed.”

Maybe that doesn’t mean much to you just yet. But it gets better.

Another communication on a separate ticket revealed this:

“Factors such as the history of buyer viewings, the sales history of an item, price, and current availability all influence the location which a specific product achieves in search results….”

Now let’s piece that together, shall we?

BUYER VIEWINGS affect rank, according to Amazon. Now, anyone who has tested driving traffic to a listing knows that just traffic does not appear to have much of an impact. The better explanation would be that buyer viewings affect relevance. Meaning, page views (sessions) from actual, logged in, Amazon customers, through a specific search phrase, will assign greater relevance for that phrase.

Ok, if we accept this as true, let’s revisit that little snippet about sponsored product ads. Keywords indexed for search through ads will update for the ASIN when the ads are closed.

I take that to mean when Sponsored Product Ads drive buyers to a listing, then that session counts toward greater relevance.

Ok, this isn’t surprising. I mean, it is a general rule of the internet, right? Greater click through rate (CTR) usually helps ranking while lowering click costs. It is an indicator for performance. DUH!

So how do we USE this information?

…having a well-written, keyword dense bullet and description section is a must!

What Happened to Me (an Anecdote)

Before we go deeper, I’d like to share how I came across all this info in the first place.

I recently launched a new product and brand. I actually created the listing at the end of March (in order to get the FNSKUs for my factory).

However, like a rookie, I forgot to close the listing while waiting for my stock.

Anyway, fast forward four months and I am trucking along, things are definitely going in the right direction. I am ranked decently for some big, important keywords. I have a small handful of reviews. I’m slowwwwwwwly growing in organic sales every day.

Then, the worst thing happened.

Sales. Stopped.

I looked at my rankings and for EVERY ONE of the 31 key phrases I was tracking, I lost ALL RANK. Overnight.

Panicked, I start to freak out. What did I do? Am I being punished? Is Amazon throttling my listing?

I contact some friends with much greater inside knowledge than me and they explained that this is very unlikely. Amazon has a specific set of actions it takes, against an account or against a product, and rank throttling isn’t one of them.

So I decided to open a case with Seller Support and allow them to help me get to the bottom of things. And they actually did. For starters, I found out that I wasn’t indexed for ANY of my back end search terms. They were ALL considered irrelevant.

And the reason this was odd is because when I created the listing I checked both indexing and relevance for each and every term in the back end and subject matter keywords. I indexed for them all and I was pretty relevant for most of them.

So what happened?

It seems in just about every instance (with rare and few exceptions) a product cannot be ranked for a keyword in search if it is not indexed for that keyword. This would mean that indexing is an integral step in the process of ranking.

The Honeymoon Period

While I was awaiting a response from Amazon I contacted a trusted colleague who runs several Amazon accounts for a number of different brands. He also launches new products and runs promotions quite regularly. I asked if he had any experience like this.

He informed me that this has happened a few times to him and that he believed my “honeymoon period” or “grace period” had run out.

See, due to the fact that it appears infinitely easier to rank a newer listing, with far fewer units than other listings on page one for highly competitive terms, it has been speculated that new listings are granted a period of greater exposure (with no negative consequences).

This is referred to as the honeymoon period.

Some believe it is purposefully granted by Amazon to give new listings a greater chance to compete. Others believe it comes down to the math of their algorithms. If sales history is a major factor in ranking, and benchmark periods are smoothed out into averages to figure ranking, then before hitting certain benchmarks a low sales history would have no negative impact.

To illustrate what I mean, if product A has three solid weeks of NO sales, and then on that last week does 150 units a day, product A will likely rank page one. That is, until the 30 day average kicks in. See, before that time, when product A didn’t have a 30 day average, those low sales days weren’t a factor. But once the 30 day average came into play, product A didn’t look as competitive to Amazon as it did before.

It has also long been debated what those historical bench marks are. For a long time it appeared Amazon factored heavily a historical period at the six month mark (as evidenced by the large number of anecdotal accounts of sellers dropping in rank for several keywords right around 180 days). Some claim it is now the three month mark, and some say six weeks.

Back to my anecdote, this didn’t make sense. Of all the recent theories about benchmark periods and sales history averages, nothing has ever alluded to a four month grace period (remember four months ago was when I created the listing).

Something didn’t add up. Upon further investigation, I realized that while I created the listing four months ago, I hadn’t populated it with full text (full title, bullets, description) until THREE months ago.

Ok, that makes a LITTLE more sense. A 90 day honeymoon period seems feasible.

Then I got that info from Amazon. Honeymoon or not, why did my listing suddenly de-index for all of the search terms?

Was it a special character in the wrong place? No.

Brand names in the back end? No.

A glitch? If it had been, support would have hinted at as much.

Then I remembered a little hack that I had learned through my network.

Someone explained to me that, when trying to revive older listings with promotions, she had experienced a renewal of rank ease if she deleted the listing and waited for 24 hours to re-upload.

I wanted to try it, but I had promotions scheduled and that would knock those out. However, some others had claimed that it was just as effective to delete all the text from the listing, wait awhile, and then repopulate the text. This apparently forced a re-index.

This was a better option, so I tried it.

And lo… search terms were indexed once again.

But….something was bothering me. Why would forcing your listing to re-index somehow wipe away a poor sales history (referencing the people who gave me this advice)?

Indexing would have no bearing on those sales history averages.


The NEW Honeymoon Period

Suddenly it occurred to me; What if this “honeymoon period” of easy ranking had nothing to do with sales history averages (or much less than we thought)?

…….Well that spun everything I thought I understood about the algorithm on its head.

So if it isn’t tied to sales history averages, then what?

What if the honeymoon period is actually tied to relevance?

As in, when you first populate a listing with keywords and search terms, they ALL index for a period. This gives the listing the benefit of the doubt that it should be relevant for those terms. And if you think about it, we’ve all experienced this. When we put keywords into the search terms and subject matter sections, we often see almost immediate indexing for most of the keywords and key phrases.

What if the listing stays indexed for those terms for a period and the relevance of those terms, affected by sessions, sales, and SEO, is what keeps the terms indexed or not?

This would mean that, for that grace period in the beginning, since all terms are indexed and possibly relevant, it would be much easier to rank for those terms. And, as sales pour in, those terms, and more, stay ranked, while search terms stay indexed.

However, if performance is poor, de-indexing occurs because relevance disappears, and therefore rankings suffer. This seems the most plausible explanation of what goes on with a new/old listing with regard to ranking (in my humble opinion).

How You Can Use This Information

If I am right about a relevance grace period, and Amazon’s catalog team is right about how you can affect relevance, this is a powerful combination of knowledge.

First, it means that you have the opportunity to rank for way more terms in the early stages of your listing. Now, ranking for a ton of long tails may not bring in significant sales for your niche, BUT the added visibility can help you to gain relevance for many more terms, including the important higher volume ones.

It also means that, having a well-written, keyword dense bullet and description section is a must! We’ve said it for years, but now it is clear that even if these sections aren’t often fully read, they are important to the algorithm.

In the beginning, while you are indexed for….basically everything you put in your listing….it is a good time to run Sponsored Product ads. As mentioned by the catalog team, you’ll get some indexing, page views, and relevance for search terms if you get clicks.

Some things to keep in mind:

  • Cast a wide net and hit the ground running when you launch.
  • Velocity is king.
  • Visibility is more important now than ever.
  • SEO will make or break you.
  • Know and understand your browse node.

As always, remember that ZonBlast offers just the service you need to affect this kind of launch strategy.