Google PLA (Shopping) vs. Amazon
When we launched our e-commerce business, one of our first questions was what to expect when selling on Google’s PLA vs. Amazon. As PPC experts and online marketers, we wanted to research ROI and funnel conversion metrics across the platforms to determine where to focus our efforts and what return could we expect from our advertising dollars – but found very little good data to inform our decision. So, we did it the old fashioned way, rolled up our sleeves and launched on both platforms. After 6 months of testing, we finally have enough to share our initial results and opinions on the Google PLA vs. Amazon debate.
Google PLA vs. Amazon
The Basics of the Selling Process
Before we get to the core metrics, it’s important to note that the platforms are fundamentally different. In Google, you market on a CPC network to drive traffic to your own website. All responsibility for landing page, cart/checkout, payment processing and fulfillment are yours. You will need an advertising budget from day 1 to generate traffic. You pay Google per click to your website and that’s the end of Google’s involvement.
Amazon on the contrary handles significantly more of the steps. They own the platform, amazon.com which already has significant search and click traffic. Clicks to your listing are free (although they do have an optional CPC model which we included in this test); they handle the cart process, checkout and payment processing. Amazon even encourages you use their Fulfillment by Amazon (FBA) process and they handle all shipping and return from their warehouses. Not only does this free you up from handling operations, but it makes your product Amazon Prime/Free Shipping eligible. Using Amazon FBA can offer a significant boost to your Amazon ranking and sales conversions since Prime shoppers are loyal and like their free 2-day shipping.
The Economics & Risk
Google PLA Economics
The economics of a transaction on Google PLA and Amazon are very different. Google charges a variable CPC based on market forces. Similar to standard CPC bidding, the price you pay is based on a few factors, mainly, your bid price and your CTR. The upside of PLA is that you only pay for the click. If your website has great conversion rates, you could be hugely profitable. But if you have low conversion rates, you may be spending advertising dollars for clicks that don’t generate any revenue. As an advertising model, there is more initial risk because you need to spend your marketing dollars upfront to get the sales process started – whether you generate sales or not. In our test, our website converted at almost 2%, which is within average ranges for our e-commerce category. (We are continuously A/B testing to increase that percentage.)
Amazon on the other hand charges not for the click, but for the sale.
Amazon Economics
In Amazon’s controlled ecosystem, you only pay when you sell your product (if you don’t use their PPC system). It’s very predictable and upfront. When you list your product, they tell you exactly how much of the sales price they will take in fees. It’s a lot more straightforward and in Amazon’s system, you can minimize your risk to ensure you don’t lose money on any transaction.
The tradeoff for not needing advertising dollars to launch is that Amazon charges a Referral Fee on all product sales. Marketing costs are fixed, but Amazon is not cheap. Consider this Amazon’s “tax” for having already generated search traffic and clicks on their platform for sellers to leverage. The referral is fee based on your product’s List Price. It differs by product category, but is usually about 15%. If you choose to use FBA, Amazon charges another 2-3 fees for processing and order, picking & packing items and for the shipping. Expect the FBA fees to run $2.50 – $6 depending on product size and weight. Bulky, awkward or heavy items could run significantly more than that. Knowing these costs for your specific product are key to choosing a price and estimating your margins.
|
E-Commerce Process |
Google PLA |
Amazon |
| Generate Paid Visits |
✓ |
Optional |
| Generate SEO Visits |
✗ |
✓ |
| Manage Landing Page / Site |
✗ |
✓ |
| Cart / Checkout |
✗ |
✓ |
| Process Payments |
✗ |
✓ |
| Fulfill Orders |
✗ |
Optional |
The Case Study
Introduction
Full disclosure, all data provided come directly from our own files. We launched this product as a self-funded side project. It fits into the apparel fashion category. We choose to sell a niche product that could be sold to a large audience. There is quite a bit of debate on how/why to choose a market. In our case we picked something that was a unique take on an everyday item and could potentially appeal to everyone. There is quite a bit of competition selling our exact product and private label (also called white label) versions of our product. We decided to avoid the brand-building route and sell the version most commonly available. This was definitely a double-edged decision that we will discuss in another post. [Next article]
The Data
This is real data from our 30-day test selling identical product on both platforms. The data can be tricky to pull and parse because Google optimizes to the click and CPC price while Amazon optimizes to the sale. I say tricky because if you advertise on Amazon you can get direct CPC data, but your PPC sales also help you generate Organic sales. So getting an accurate picture takes some knowledge of their reporting and optimization methods to back the data out correctly. In Google, you’ll need to make sure you are tracking clicks properly on the backend of your site so that you can properly attribute conversions back to the click.
Marketing Funnel Results
As you can see the results here are drastically different. The CPC was almost 5x higher for Google than Amazon. This is primarily a result of the fact that we were generating organic clicks on Amazon that are included here. Our page conversion rate was also 2x on Amazon what our own website did. We attribute this to 2 factors: (1) The amazon platform attracts extremely qualified shoppers who are already low in the sales funnel. They are on Amazon to buy. (2) Over the months, we ran a number of tests on our website to optimize site conversion rate but Amazon has been optimizing their pages for a decade and with an astronomically higher sample size. The number that surprised us the most was the marketing cost per order: Nearly $26 for Google and only $8 for Amazon.
Further Optimization
The biggest factor we found that mattered in bidding on PLAs was adjusting our mobile vs. desktop bids. With equivalent bids ($.49 v. $.50) we were able to but 7x more impressions on mobile vs. desktop. Our site is optimized for mobile and we saw similar conversion rates of around 2%. Moving forward, we would focus on profitability. Having seen that lower bids significantly reduce volume, we’d likely test raising our price to maintain margins while being able to increase our maximum bid.
Unit Economics
The significantly higher marketing cost of generating an order on Google may immediately make you think the channel was a bust for this product. But there was a significant difference in the selling price we could command on each platform. We could sell an identical product for $11 more on our site via Google PLA than we could on Amazon. This was a result of the competitive market dynamics.
The average price point for competing listings on PLAs was $50 – $125, giving us room to price at the bottom of the range at $50. We had previously tested Google at a $40 and found similar click and conversion rates, showing that within our page competition, consumers weren’t very price sensitive. Our Amazon listing page was competing in a much different marketplace and that significantly influenced our pricing. With many sellers offering close or identical items, we were competing in a much more pure capitalist environment. Without the ability to differentiate our product, shipping or service, we had little choice but to compete on price alone.
Unit Economics (continued)
Our COGs and other product costs were very similar. There were slight variations between them because we used Amazon FBA vs. USPS to ship the product. Our packaging costs were different only because we changed our packaging design when we launched on Amazon. And the dollar difference is small enough we think it’s immaterial to our overall conclusions.
When we put the Marketing Costs & Unit Economics together you can see that the Amazon platform was about 3x more profitable per unit sold. When you then compare the ROI on marketing dollars, Amazon’s platform showed a 10x efficiency over Google PLAs.
|
Unit Economics |
Google PLA |
Amazon |
|
Avg. Sale Price |
$48.73 |
$37.54 |
|
Unit COG |
$17.00 |
$17.00 |
|
Shipping |
$2.60 |
$3.22 |
|
Packaging |
$0.50 |
$1.00 |
|
Total Product Cost |
$20.10 |
$21.22 |
|
Gross Margin |
$28.63 |
$16.32 |
|
Net Profit (Less MKT Cost) |
$25.89 |
$8.26 |
|
Marketing ROI |
10.6% |
97.6% |
The Conclusion
Looking back at our test, the most valuable takeaway is the knowledge we gained from the test. Including our time invested which was approximately 120 hours, we earned approximately $3.16 per hour of time invested. That is certainly not a sustainable hourly rate. We did however create lasting value. We now have a Amazon listing that without additional significant investment from time or PPC will generate approximately 3 – 4 sales a day. We also have a model for running Google PPC profitably moving forward. We will probably test a higher per item price to help us maintain our current spend level in Google PLA but will hopefully raise our profitability without sacrificing sales. Additional time and money will need to be invested in Google PLA in the future to improve profitability and sustainability.
Amazon Conclusion
Both platforms have pros and cons. For Amazon, we like the self contained selling environment. No need for a website, cart etc. We also appreciate that PPC sales help generate organic sales and that it’s possible to start with little to no advertising budget to minimize your upfront risk. And while we appreciate that this is a market designed for people ready to purchase, it’s a double edged sword. Conversion rates are high, but this makes it intensely competitive and pricing pressure can quickly erode margins. We like that the overall ROI and profit margins were significantly better, but we also believe we have room to improve both those metrics with more testing on PLA.
Google PLA Conclusion
Google advertising poses more initial risk because the advertising must be paid for regardless of the return. In our market, our COA was very high and overall ROI was not as good as we would want. However, the benefits to Google are the ability to launch faster, and we believe the potential to create superior margins over time by building a brand and optimizing the website and price.
Summary
If you are launching an e-commerce business and have the resources, we would definitely recommend launching on both platforms. This will give you broader exposure, 2 marketing channels and the ability to gather more data. If you have limited resources, we would recommend launching on Amazon. Not only will it require a smaller advertising budget, but it will also limit your downside risk – and frankly, you can launch without needing a website or operations process. After you are launched on Amazon, then we would recommend getting a site built and advertising setup. You can leverage learnings about keywords and selling factors from Amazon to launch PLAs effectively.
It’s critical to note that our test included only 1 product in a very specific market and we would expect other products to behave differently, but we hope the platform overviews will help guide your e-commerce ventures.