Reverse Engineering Amazon

The State of The Promo Address

The Landscape of Fear (and Loathing)

If you are an Amazon seller and partake in any of the many groups on Facebook, Whatsapp, Wechat and more for FBA and private label sellers, then you’ve undoubtedly heard people speculating about the effectiveness of promotions. Many are stating their efforts to rank keywords through promos using the storefront URL are no longer working. Some are saying their efforts to rank keywords using promos AT ALL aren’t working.

And of course there are just as many or more that come to the defense of the tactic with their own set of screenshots and successes. The counter argument to that is that the seeming swell of failures is growing, proving that Amazon is clearly taking some sort of action.

And on and on the debate goes, as newbies and experienced sellers alike scratch their heads wondering what to believe. Meanwhile, “gurus” from all corners weigh in with incomplete (or no) data, some with an agenda, and all confounding sellers even more.

I propose the best way to get to the bottom of it all is through a process of logical conclusions, and analysis of real data. Strap in folks, this is gonna be a bumpy ride.

Where’s All the Paranoia Coming From Anyway?

Amazon has actually been changing things. One of the first noticeable changes was with search page URLs and branded URLs. See, the first page of a search looks the same as usual from the perspective of the URL in the browser bar. However, if you navigate to the next page, you’ll see a shiny new QID timestamp added.

Here’s an ending URL string for page two of a search – &page=2&keywords=garlic+press&ie=UTF8&qid=1536823069

This means that, for those who were conducting a manual search for their product and then distributing the search page URL with their product on it, attempting to mimic organic search for their promos, NOW the aged timestamp proves the truth.

The same holds true for brand URLs. If you generate a URL filtering your brand (or any filter) from the left-hand side of the page on desktop, it will also have a QID timestamp as well as a new ID (RNID).

Here’s an ending URL string for a brand filtered page – &keywords=garlic+press&ie=UTF8&qid=1536823401&rnid=2528832011

As if that weren’t enough, some industrious thought leaders in the space ferreted out another, hidden QID timestamp IN THE STOREFRONT SEARCH PAGE!

All links to products also have QIDs, meaning the distribution of the ever popular keyword-embedded storefront URL would seemingly be as effective as the search and filter pages.

So, QID = BAD, Right?

Amazon sellers who use promotions as a means to help increase rank and visibility for their product listings, that have been around for more than three years, likely remember the days of the “SuperURL.” The original SuperURL wasn’t just a keyword embedded URL as many refer to it as today. Back then it was a URL that dynamically inserted a fake QID timestamp into the URL in an attempt to mimic an organic click.

The reason for this was because, if you simply searched your product and clicked your listing, then shared that URL in a promotion, it would have an old timestamp in it. The idea was to get the best rank juice from Amazon possible, and few would argue the most effective way to do that is through an organic search/find/buy. So any way to get as close to that as possible was the logical strategy.

That is when sellers started to be conditioned that the QID is a bad thing. The QID MUST be how Amazon tracks the authenticity of a listing click. At least, that was the general attitude.

This concept has perpetuated itself so much that even respected thought leaders and programmers in this space have subscribed to it today. Plain and simple, QID is apparently how Amazon KNOWS whether you are “gaming” their system.

Except…that just doesn’t make sense.

First of all, Amazon would have to store each and every QID timestamp (not a difficult feat for a service that provides ridiculous amounts of cloud storage). But that isn’t all. They’d also have to QUERY that data to match it to listing traffic results and purchases.

I can tell you from personal experience that it can take HOURS just to query data from 100,000 rows of information. Imagine how long it would take the billions of timestamps Amazon distributes. Now, I’m not saying Amazon isn’t capable of this. I’m simply saying that it would be grossly inefficient.

And I’m not alone in thinking that. My good friend and long time developer (15+ years of experience) agrees. He says:

“When the URLs are generated by Amazon, as you have noticed there is on several of them a QID at this time. This is literally a UNIX timestamp – Not a Globally Unique identifier. There is a fair possibility that more than a single user would hit the same keywords at even the same nanosecond. All of a sudden, we are looking at duplicate timestamped URLs.  This doesn’t bode well for me to believe the QID is an element that is used for tracking behavior. If this was a GUID or otherwise obscure encrypted character string like a HASH, then perhaps.

Furthermore, the consideration they could and would track these URLs, i.e. log them in a database, sounds feasible on the surface, but shear velocity of web requests coming in would jam pack a database so full of these URLs in a matter of days it would take hours to even query through them, and we are talking about years of data here, not days.”

So what DOES Amazon use these timestamps for?

To create a better experience for buyers.

By using machine learning algorithms, Amazon can pattern match these timestamps to better analyze and predict customer behavior. This seems a MUCH more feasible scenario.

My friend’s final thoughts on the subject:

“I honestly can’t think of a scenario where QID alone would be an effective means of authentication for real time product level rank manipulation. If, in fact, you could prove it was, we could just generate a QID from a UNIX timestamp command and append it to the URL. Without a super computer churning an ungodly amount of data I don’t see them validating the URL.”

So What IS Amazon’s Aim?

First off, I want to point out that the premise for all of this concern over rank manipulation and URLs stems from seller paranoia rooted in the idea that Amazon doesn’t want you to use promotions to rank your listing for keywords.

However, I’d like to present exhibit A; an excerpt from an email directly from Seller Support:

Couple things I’d like to point out…

“…you can contact your idealized customer and provide then the details about the promotion.”

Hmm. Sounds a lot like blasts services, huh?

“…which will popularize your product and improve your sales.”

How do you figure it will popularize the product?

Let’s unpack this, logically. If all a promo code does is allows someone to buy a product for less than its worth, why would Amazon push the feature so heavily? I mean, it would make more sense if ranking was given to the velocity of promo sales too (which is consistent with what we’ve seen over the last…oh…say…four years). That way, it gives a product the ability to become popular and therefore sell more, making Amazon more money.

Now I’d like to share an excerpt from another email (thank’s Stetson):

See here that by Amazon’s own admission, this ties marketing to specific keywords and their subsequent rank increase (que mic drop).

But let’s not stop there. We’re gonna take the logic train ALL THE WAY!

Over 90% of the click through and purchase action happens on page one of a search. This is the reason why increasing rank from nowhere-to-be-found to the bottom of page one is significantly easier than moving farther up page one (in terms of sales velocity).

Next, lets take a look at an often unexplored realm of Amazon; the Amazon affiliate program.

Amazon has made clear they have love for their affiliates. It is evident in the special newsletter they get, the special programs (such as Amazon influencers) and their API access. As a matter of fact, let’s dive into what affiliates get with the Amazon advertising API (which you can only gain access to with an affiliate account in good standing).

  • The ability to build out a search mechanism that will allow a user to search for items on Amazon from a third party site.
  • The ability to cache the information on a product and add it to their shopping cart in Amazon.

Essentially Amazon gives affiliates the ability to mimic the search experience on Amazon. This means they want to supply affiliates with fresh and updated information (which is possible through the API). As such, it stands to reason that they would lend rank juice through sales made when directed from external affiliate URLs.

Otherwise, affiliate sites would be less effective, and why give them all that API access if that’s the case?

At the heart of it, the affiliate (referred to within Amazon as the associates) program is a literal cash incentive for an end user to provide an external URL to persuade a sale. To me, this means that at the very core of Amazon’s platform is URLs that lend rank juice to sales velocity.

In the simplest terms, Amazon likes outside traffic. Amazon rewards it too.

Which brings me to the storefront URL.

Think about this; Amazon loves affiliates, and allows them to create external links to drive traffic back to Amazon. Wouldn’t it make sense for Amazon to offer the same type of shareable, rewardable link to brand owners?

I believe this is what the storefront URL was intended for. It gave brand owners the ability to drive external traffic, along with the benefit of ranking if they did a good job at promotion. The effect of that ranking would be that a popular product would sell more, generating more happy customers and more revenue for Amazon.

(For those of you arguing that the reason this type of marketing taints Amazon’s otherwise “pure” ecosystem is because of the crappy products that rank, Amazon has plenty of other mechanisms in place to keep crappy products from dominating their platform. Namely, the review system.)

If this is true, then that would mean Amazon has the sole discretion and ability to control how much rank effect it will allow to flow through the storefront URL (or any URL for that matter).

The QID MUST be how Amazon tracks the authenticity of a listing click. At least, that was the general attitude….Except, that just doesn’t make sense

What’s the REAL Reason Why So Many People’s Promos Are Failing?

Well, right now there are a couple of reasons. Chief among them is Amazon’s changing of layouts.

To understand why this would negatively impact promos, it’s important to understand Amazon’s code base (a little). See, Amazon is a HUGE company with an equally HUGE platform that requires HUGE amounts of code to run.

Imagine, millions upon millions of lines of code.

Also consider that Amazon is a couple decades old. So, over time, coders from different backgrounds, with different coding styles, and different language proficiencies have all contributed to the behemoth that is Amazon today.

PLUS, Amazon operates worldwide, so it has servers all over the world. With that kind of widespread service area, it would be impossible for Amazon to flip a single switch that affected the entire marketplace.

That is why they have to roll out testing and final products region by region, server by server, and sometimes category by category. Understanding this, now you get a sense for what the layout switches have been like. First, they test different layouts in different categories.

For example: “Tennis Shoes for men” gets a ridiculous layout, forcing 48 listings onto page one.

“Garlic Press” gets 35 listings to page one, but 10 of them are sponsored products, and there are two new sections as well; Amazon Choice by price range, and “Expert Recommendations.”

After testing these layouts and ad placements, Amazon slowly began switching over the layouts across the platform. This layout change has increased dramatically the number of listings on page one.

Over 90% of the click through and purchase action happens on page one of a search. This is the reason why increasing rank from nowhere-to-be-found to the bottom of page one is significantly easier than moving farther up page one (in terms of sales velocity).

So, with the old 16-organic-positions-per-page layouts, moving from position 35 to 12 was not that hard. Now, however, page one is stuffed and position 35 may even be at the bottom of page one. Those positions are in higher competitive demand and require much more sales velocity than before.

This is the primary reason why so many have shared anecdotes about how promos are either not as effective or require much more cost in terms of inventory than before.

In fact, I consulted my developer friend on this too. He says:

“I believe that Amazon has seen unprecedented growth in the last several years and we have not hit that cap. For a long time suppliers appreciated their private label customers, but as with anything it was only a matter of time before they figured out how to compete with the middle man. This created a super surplus in most Amazon categories. As the categories expand and grow we find more products. This pushes the category into search pages with more real estate. So, you’ll see the shift into the tiled layouts. I think we are seeing this sort of layout start to take precedence in more and more areas of the site.

The real issue is, as more products become available and more sellers get involved, you have more actual physical products available in any given category that can be ranking on any given keyword. This is an interesting problem to have, because if your product is on page 3, but 24 products come back in stock from other suppliers that directly compete you could easily lose a whole page of rank in a few moments. Also, with more and more products taking clicks for your keywords I think we are seeing that the effect is being diluted or watered down. This is really a rule of numbers and would point at the fact that it is possible some of the figures currently in use that predict give away volumes are not doing a good enough job factoring in market expansion.”

Amazon shifting their layouts to accommodate more sellers, and unprecedented numbers of competitors entering the site (especially with Amazon opening the gates for actual suppliers to compete directly on the platform) may be only part of the problem.

Another issue appears to be Amazon testing custom search experiences.

Take a look.

This is a screenshot of the first organic results for the search term “fitness bands” under my account, logged into my IP address in Taiwan. I was using Chrome.

  1. Fit Simplify
  2. Limm
  3. PPwer Band
  4. Fitness Master
  5. Fitness Insanity
  6. Panathletic

Here is a screenshot of the same search under my wife’s account logged into a Houston IP address. This one was in Internet Explorer.

  1. Fit Simplify
  2. Odoland
  3. Limm
  4. PPwer Band
  5. Fitness Insanity
  6. Fitness Master

Odoland wasn’t even in those options under my account.

Now here is the same search logged into private viewing (no account) on Firefox from the same Houston IP.

  1. Fit Simplify
  2. Odoland
  3. Limm
  4. Fitness Insanity
  5. Panathletic
  6. Fitness Master
  7. Suncaya
  8. Chicmoda

Almost identical to my wife’s search (same IP) except the layout is different, so there were two more listings available above the fold.

And that’s just for the TOP results, which mostly remain somewhat consistent. You really want to see where this gets hairy, compare the differences further down page one (where many people end up after a promo).

And all that was on desktop. The results continue to get scrambled on mobile too. The point is, with Amazon shifting search results across multiple browser instances (logged in, not logged in, mobile, desktop, Chrome, Firefox, etc) it can be hard to see the desired results for a promotion.

We also hypothesize that there may be a lag between servers after a high velocity promotion. If you suddenly shoot from 20 sales a day to 250 one day, then gain significant rank, it may take days longer for that rank to reflect across most browser instances.

What Does the Data Say?

I took a very wide view of the data, so there are many nuances missing, but what I did find was interesting. I simply looked at about 21,000 blasts and analyzed the average keyword movement. This did NOT account for anomalies or failed promotions or number of codes redeemed or any of the other details that would go into a proper analysis. This is just a surface level view of whether it appears promotions still perform. The data was also only pulled from recent blasts, and the past three months were broken down separately to ensure any recent changes would be seen. Here’s what I discovered:

The total average keyword movement for over 18,000 Heatseeker blasts is approximately 24 positions. That means, the average number of spaces each keyword increased was 24 spots. Some were significant, with over 150 spaces moved, some not so much, with only 1 or 2 spots increased. But I take this to mean that yes, despite the difficulties with Amazon, promotions are still causing rank increase.

Of those promotions, the average keyword rank increase for promotions discounted at 90% or HIGHER was 32.5, which is significantly higher than those with UNDER 90% discounts. This tells us that velocity is still EXTREMELY important to the ranking process (since higher discount deals move much faster).

Another interesting finding was that the movement from listings that were not on page one (average 31.6) was significantly higher than those that already were on page one (average 3.2). This has always been assumed, but it is still good to get the validation. It appears that movement within page one will take significantly more aggressive velocity, with that increasing the farther up the page you try to land.

Now for the most interesting findings. I also analyzed 3,500 Sidewinder (storefront URL) blasts. For awhile (early Summer) the average keyword movement was slightly less than Heatseeker but still within the predicable range. However, as of recently (sometime in August) Sidewinder results became significantly less powerful. As of the most recent blast data, Sidewinder rank movement capability has decreased over 100% (yes…we were shocked too). Rank movement went from an average of in the low 20’s to less than 1!

What does that mean?

Has Amazon manipulated the rank-ability of the storefront URL?

Is it just market saturation (everybody uses the URL now)?

That, we don’t know. We likely can’t know. All we can do is review the data and try to react as quickly as possible.

For that reason, we are switching out the Sidewinder URL to a new URL that will NOT be the storefront.

What Can You Do?

Here’s the real question on everybody’s minds. What can be done about all of this?

Currently our developers are working on a super hush-hush project to identify exactly what consistencies exist programmatically for the browser instance changes. When they nail that down, we’ll be able to solve for the problem within ZonBlast itself (hopefully).

In the meantime, I suggest first that educating yourself on your product’s situation, as well as the landscape that is Amazon, is key.

Important things to note, these are the browsers supported by Amazon (according to Amazon themselves):

  • Apple Safari
  • Google Chrome
  • Microsoft Internet Explorer
  • Microsoft Edge
  • Mozilla Firefox
  • Opera

Also important to note is that mobile is driving 56% of all traffic to major sites, with that number increasing to 68% in the instance of Amazon. And they aren’t just browsing either; that’s the amount of people of all Amazon.com’s total traffic that are using mobile to actually shop.

Lastly, Google Chrome accounts for 56.6% of all desktop browser use and iOS accounts for 77% of mobile use. So, that means that the two most important browser instances for your listing on Amazon is Chrome and Apple Safari.

Whenever you are tracking your keywords, it is wise to double check, manually, these browser instances. And with that in mind, also be sure that when you are planning a promotion, the tracking software of your preferred launch service (*cough* ZonBlast *cough*) is able to find your product.

That means that, if your product is suffering from volatile keyword rank movement in your tracking software of choice, as well as when you look for it manually, you may want to WAIT to run any promotions.

Most of the browser instance issues are impacting products by category. It typically causes a couple of weeks of volatility, then things settle down. When they do, the next thing you’ll need to account for is the greater competition.

Right now, with the insane amount of new sellers, insane amount of products featured on page one, and the insane increase in paid ad placements, gaining top ranking spots is more challenging than ever. To add insult to injury, the least volatile positions, with regard to rank position changes per browser, are the top six.

So if you want to ensure you get the most out of promotion launches, you’ll want to:

  1. Start fast. Run your promotions as quickly as you can after your listing goes live.
  2. Go hard. Promote with at least 70% units of the CURRENT average sales velocity of the top five spots on page one of your target keyword.
  3. Aim high. Your ultimate goal will need to be landing within the top six spots on page one.

In the end, yes, getting things going fast and profitably on Amazon is becoming more challenging, but that is an evolution just about every industry faces. The numbers show that promotions are still a viable strategy to HELP the process. But, just like a healthy diet, you should supplement your promotions with a healthy mix of paid advertising, customer outreach and off-Amazon brand building.

Wait a minute. We’re over 3500 words in and you said this is only part ONE? So there is MORE?

We want to ensure we fully understand what is happening in the space right now, from every angle and accounting for every nuance. To that end, we are having a data scientist analyze hundreds of thousands of blasts to ensure the information we supply you with is as accurate as possible.

It may take several weeks, but part two of this series promises to be most enlightening.

By |2018-09-17T17:11:13+00:00September 17th, 2018|Amazon, Analysis, Case Studies, Mythbuster, Paid Advertising, promotions, Ranking Factors|

About the Author:

Share This

Share This

Share this post with your friends!