Amazon FBA Case Study: Get Data Directly From Amazon
Last episode we discussed how testing can provide valuable data directly from Amazon that will help you see how competitive or lucrative a keyword/product idea is.
We also showed you how to set up a test listing.
We discussed how the collection of data and setting up tests will help eliminate most risks when starting to sell a product on Amazon.
In today’s episode, we’ll go over how to verify if the test listing is indexing for the right keywords.
We’ll also show you how to set up a test pay per click campaign.
We’ll then show you how to analyze the data that you’ve compiled from your tests, and how to use it to determine which products to launch.
Let’s take a look at the images we managed to take for our test listing.
As you can see, we didn’t optimize the images, but our purpose here was only to have a listing that Amazon considered relevant for keywords we wished to gather data on.
After we verify indexing for relevant keywords, we need to create an Amazon Sponsored Product advertising campaign in order to further collect the necessary data we need to make an informed decision.
First, we use Helium 10’s index checker to see if our test listing is indexed for the important keywords we identified.
But what is indexing?
Indexing is how Amazon connects keywords and key phrases to a listing. When a listing is indexed for a key term, that means it is searchable for the term on Amazon. This is important because you cannot run effective advertising on keywords that you are not indexed for.
After we verify that we are indexed, this lets us know that we can expect to be able to successfully run advertising (and therefore collect data) on our chosen keywords.
Now we need to set up our test Sponsored Product ads. In order to ensure we are gathering all the data we need, we also must make sure we aren’t setting our daily budget too low. We’ll set our budget at $40 to $50, with the expectation that we will likely not spend anywhere near that much.
Wait. What? What does that even mean?
You spend money on Amazon’s ad platform only when someone clicks on your ad. The reason you set a budget is to let Amazon know how long to show your ads for each day. This means, if you don’t get enough clicks, you don’t use up all of your budget (and your ad runs all day). Since we don’t have a super optimized listing, we don’t expect to get too many clicks.
Aside from setting up the ad with the proper budget, we will also need to choose “manual targeting.” This means we will be entering the keywords we wish Amazon to show this ad for on search result pages.
Our bidding strategy, to keep things simple, will be “fixed.”
We will also set a fairly high “default bid.”
Ok, so Amazon’s ad platform is an auction, so a bid is what you are willing to pay for a click. The highest bidder tends to show up higher in rank for the targeted keyword (with other factors such as relevance and ad performance playing factors).
The reason we want to bid high is because if we bid too low we won’t get any “impressions” (this is how many times the ad is shown). Impressions are one of the important data metrics we are collecting for our test.
Next, we choose to target products or keywords. Product targeting allows you to place your ads on specific products or brand listings.
However, since our strategy rests entirely on keywords and keyword testing, we’ll choose keyword targeting.
In this test ad, we need to choose “broad match type.”
But what is a match type?
Match types dictate how Amazon treats a keyword when it shows your ad. Let’s define each of the three options to clarify:
Exact – This is EXACTLY the keyword you input. That means, if you tell Amazon you want to run an exact match type ad for the keyword “stainless steel garlic press” it will only show that ad on search pages for “stainless steel garlic press.” It will not show on search pages for “garlic press” or even “garlic press stainless steel.”
Phrase – This match type allows for additional keywords to be added so long as the phrase remains intact. So, using the previous example, if you run a phrase match ad on the term “stainless steel garlic press” you may also be shown for terms like “stainless steel garlic press crusher” or “blue stainless steel garlic press.” Phrase match allows for additional, relevant, supporting keywords to be added before or after the phrase you input.
Broad – Broad match ads target any combination of the input key terms, in any order, along with any relevant supporting keywords. So, the term “stainless steel garlic press” may also show for “garlic press stainless steel blue” or “stainless garlic press red steel.”
Since we hope to gather data on very specific keywords, this test requires exact match targeting.
A quick note on “negative keywords.”
Input negative keywords when running Broad or Phrase match ads. This will tell Amazon to avoid showing your ads for those specific terms when it is finding relevant additional keywords. Something like “plastic egg tray” may seem relevant to Amazon, but if your product is wood, it would be a waste of potential clicks to show your product for that term.
After setting up our test ads, the next step is to send in our test units.
A quick note on “suppression.”
A listing becomes active as soon as inventory is available for sale. However, when a listing is out of stock, it becomes suppressed. When Amazon receives a shipment of FBA inventory, it immediately disperses that shipment to multiple other fulfillment warehouses within its network. This is so inventory can be available as quickly as possible for anyone who might order it all over the country.
The problem is, shipping times to these warehouses will vary. What that means is, if you have two test units become available and then you sell them both, your listing will become suppressed and this will cause your PPC ads to pause.
When the listing goes live again when inventory is available, the ads will start up again. However, if the listing gets suppressed again, the ads will pause again.
This may interfere with your impression data (since you only get a report of impressions over time, and the ads turning on and off will not show accurate time-frames that impressions were being gathered).
Our strategy, then, is to intentionally suppress the listings ourselves until all of the test inventory is available, just in case any of it sells. How you do this is, after shipping the FBA inventory, remove the primary image. Listings on Amazon cannot be live without the main image.
As soon as all test inventory is checked in, by returning the main image to the listing, the listings and ads should go live within minutes.
Sending test listings into FBA is simple. You just click send/replenish inventory in seller central and follow the prompts to send in your box(s) of test units.
Be sure to choose “case-packed” and input the proper number of units per case.
Now we can look at our test PPC campaign results.
The important data that we want to study is “impressions” and “CPC” which stands for “cost per click.”
Impressions will let us know how searched or in-demand a product/keyword is. Cost per click lets us know how competitive that keywords is. These are critical pieces of information (how many people are looking for a product and how many competitors are selling that product….tells us all we need to know pretty much).
To view our results on-screen, all you have to do is go to your Campaign Manager in seller central.
Then click into the adset and then the ad. The data necessary for the test is under the “Targeting” tab.
For our results, we see that the keyword “camp axe” got 21,000 impressions over our test period of about two weeks. “Hatchet” received 61,000 impressions. Interestingly, “camping axes” received little over 250 impressions, but also displays no clicks. This means no one clicked the ad from that key term.
This information actually tells us quite a bit. We can conclude which keywords were the most popular (based on a number of impressions) as well as which ones were highly competitive (based on the number of clicks, or lack thereof).
A quick note about impressions
An impression occurs any time the ad is shown. Many ad placements exist, on the search page, branded pages, and even listings. As such, a single search could yield multiple impressions. So while a number of impressions is an indicator of popularity, the number won’t be an accurate expression of the number of searches.
While analyzing this data, it is important to remember, we do not care about orders, sales or conversion rate. We are just trying to gather information, and at this stage, the only information that is important is impressions and CPC.
Another thing to note is that Amazon’s suggested bid should not be considered. Why? In our example, we see the suggested bid of a keyword to be between $0.43 and $1.12. However, our actual CPC was $1.52. That means the suggestion was WAAAAAYYYY off, making the suggested bid a completely inaccurate way to choose bids as well as to gauge competitiveness. You need the REAL data.
We were able to get meaningful data for our first two products; forest axes and wooden egg trays. However, we had to run deeper tests on them, especially the coffin shelves. Why? Because Amazon was lumping together the impressions of multiple keywords. Sometimes this happens with lower volume keywords. Sometimes it happens for no reason whatsoever.
Since we saw that of all of our keywords, only one was getting any impressions, we had to split up the campaigns and individually target some of the keywords to get Amazon to report the proper data.
The data in the screenshot above came from a “search term report.” How you run a search term report is to go to the advertising reports tab, and indicate a date range.
When we analyze the data for our “coffin shelf” we see that the display from Amazon is not as accurate as this report, so it is important to always run the search term report.
We see in our coffin shelf search term report that our broad and phrase match targeting showed our ad for multiple different key phrases. This is valuable for uncovering phrases that might have not been thought of initially.
While we said to only pay attention to impressions and clicks, there is quite a bit of data these reports can give us. Another thing you can pay attention to if you so choose is click-through rate. This lets you know the percentage of people that clicked onto your listing from the ad (ratio of clicks to impressions) which is a great way to illustrate how in-demand your specific product might be for a keyword.
Another thing to notice is the low costs per click for many keywords with decent impressions. This is a fantastic illustration that competition might be very low.
Our egg try results were a little less stellar. We see a lot of more competitive keywords at higher costs per click, indicating (as expected) more competition. This is ok though because we are still armed with enough data to make a sound decision moving forward.
Ultimately, while impressions and CPC are the critical pieces of data necessary for this stage in validation, there is much more data that can inform future decisions.
So far, the data we’ve collected seems to support a case for moving forward with all of the products we’ve tested. This will allow us to illustrate the different launch methods for low to high competitive terms as well as low to high search volume terms.
In the next episode, we’ll dive into how to source products. This involves reaching out to suppliers, negotiating, how we wish to differentiate our products, and so much more.
Until then, tell us, what is the most interesting thing you’ve seen in our data?