This is a new audio-only analysis series which is a more relaxed format discussing what we've learned from our internal primary research over the last week. Please add our feed to your podcast player to keep up to date with future podcasts.


Visible Alpha Logo_cmyk.jpg
Visible Alpha Logo_cmyk.jpg

Visible Alpha built a platform to analyze consensus data for financial and operating metrics on over 6,400 publicly traded companies. Rather than digging through models one by one, Visible Alpha creates consensus data for every line item included in sell-side models so institutional investors can better understand expectations on metrics beyond just revenue and earnings. Try Visible Alpha for free here.


Welcome, everyone. This episode focuses on Amazon's retail business. Over the past few years, we've been trying to understand the unit economics of Amazon's business, as the consolidated numbers don't reveal much. We'll discuss our approach to primary research and the insights we've gained over the past 12 to 18 months. We've also recently published work on our new enterprise tier, which outlines what we believe are the potential gross contribution and EBIT margins of Amazon's retail business over the past seven to eight years. As always, we welcome your feedback and remind you to conduct your own research. Nothing shared here should be considered investment advice.

Is ecommerce a poor business?

Shipping $10 parcels for free is quite challenging.

How has your perspective on this business evolved over time? For context, you've been conducting extensive research on Amazon. You've been studying this business for as long as I've known you, which is roughly seven to nine years. How do you view the retail business today?

I still believe it's superior to traditional brick-and-mortar retail, primarily due to the foundational structure of the business. It's changed significantly over time. If you consider the early days of Amazon or ecommerce in general, and compare it to the infrastructure of a business like Walmart or other large brick-and-mortar retailers, you'll see a difference. These traditional retailers receive products from manufacturers or brands at their distribution centers. These products remain at the distribution center until they're shipped out to expensive retail stores located on high streets.

There's a lot of complexity involved in determining which products go to which stores, managing the replenishment cycles of these products, and ensuring the shelves are always stocked. Additionally, these high street stores need to be staffed with pleasant employees to enhance the shopping experience. This can be a complex business, with a wide range of experiences possible at the same retailer because every store is unique. Each store carries different products and employs different staff, which can lead to varying customer experiences.

On the other hand, if you compare this to Amazon or the foundational structure of ecommerce, the original equipment manufacturers (OEMs) or third-party merchants ship their products to fulfillment centers, which are typically larger than those of legacy players. This allows for more stock. Instead of shipping to expensive fixed assets, leased or owned retail stores, they ship directly to the end customer.

They fundamentally overlook a significant and expensive part of the whole business, which is arguably more complex to manage in terms of the human aspect and the experience of visiting a store versus buying online. For instance, Amazon only has to manage one website in every country. It's one user interface, a fixed expense. They can stock many more products than Walmart has to manage across the US.

In my view, it's arguably a more robust structure that lends itself to more scale benefits, right, in terms of the foundations of the business. Then, of course, there are the benefits of offering an endless shelf, and the auction dynamics that can be added, which Amazon has done. One fixed expense in a website, and then you can, I guess, make more use of your assets. You have lower fixed assets compared to a brick- and-mortar retailer, especially the incremental fixed assets.

You can sell more products through those assets, leading to higher returns on investment. When you add the third-party marketplace business, it becomes even more attractive. There's no doubt that it has generated high returns, historically. The original Amazon, you can go back and see the cash flow it generates, the free cash flow that financed the whole business, the underlying returns. Even five years ago, you could see the returns in North America on EBIT at 4%.

There's no doubt it makes money. What Amazon has done historically, going back to the last 20 years, is that they were able to offer an endless aisle, make more use of their assets, sell more products through the same dollar of assets, and drive a higher return.

Clearly, Bezos learned from businesses like Costco. They don't rest on their laurels and use that return to retire. Instead, they give it back to customers. They've continuously passed that back on in the form of quicker shipping, cheaper shipping, better service, and better prices for customers. This has essentially solidified their position in ecommerce in the core countries they operate in.

Fundamentally, historically, in my mind, this is a better business because it lends itself more to scale. It can grow not only SKUs, but also as ecommerce penetration grows. So you have this very scalable asset with an entrenched position, which can pretty much grow its product line and also grow its market size with the growing trend of the internet.

The story of the last 20 years has become questionable today because it's got much more complicated. It's about how much you want to share with the customer. They started off with free shipping, which was great. But now they have same-day delivery, they own aircraft for shipping, and they own their own trucks.

It's gone from being like eBay, plus owning a bit of delivery, to now running the whole end-to-end process and also building out too much stuff. This has muddied the waters for investors, who are really questioning how profitable this really is. If you are going to push same-day delivery for all these parcels when the economics don't work, because you want to maximize that lifetime value in the long run. And that's why the stock trades where it does.

In your attempt to construct the framework of ecommerce, I'd like to discuss the perspective you've been writing about. Specifically, how should we understand what this business is striving for? Given your viewpoint, one might question the ultimate profitability of the business. Is it as lucrative as people initially thought? This largely depends on how you frame the business's operations and Jassy's current priorities.

Jassy has experience in building marketplaces. Although he hasn't built a retail marketplace, which is a different ballgame, he seems to be on the right track with initiatives like Buy with Prime. Whether these initiatives will succeed is another question.

In Jassy's mind, he aims to transform this into a third-party marketplace and provide the infrastructure for others to sell their products, similar to AWS's role in cloud computing. It's uncertain whether this will work, but I believe it has a good chance of becoming the infrastructure for a significant portion of ecommerce.

Currently, Ecommerce accounts for about 20% of the addressable retail market in the US. I have no doubt that this percentage will increase over the next 10 to 20 years. They may have too much capacity now, but in the long run, they will fill it up quickly.

The question is, how will merchants and customers want to transact online? Amazon can provide a platform for these transactions. I believe this is Jassy's plan. Traditionally, Amazon has functioned as either a first-party or third-party merchant. They purchase products wholesale and sell them, or they serve as a third-party marketplace. They also handle logistics if required.

Jassy's strategy with Buy with Prime, for example, is to offer to handle logistics even for those who sell on their own websites, similar to what UPS or FedEx does, but with an attached warehouse. The key questions are, how good are Amazon and what is the cost, what is the unit cost of shipping a parcel? And is it attractive for third party merchants to either go via FBA or even when they're shipping themselves, do they want to get into the complications of managing a 3PL relationship and shipping product? That's what I think Jassy is doing.

Whether third-party merchants and, more importantly, larger retailers want to deal with Amazon is another question. Would these large retailers ever use Amazon for backend shipping? My base case is probably not, and I'm quite skeptical.

However, I don't know if Jassy can increase the third-party mix. It doesn't necessarily have to be sold on It just needs to utilize the fixed assets they own, the so-called rails or infrastructure for running parcels across the US and other major countries they operate in.

Can they displace the existing links in the market, such as FedEx, UPS, and middle mile, and the actual retailer? I don't know. But there's a lot to maximize. There are other costs in their income statement that are not strictly retail, which we can discuss later. But I believe that's Jassy's strategy.

While we're on the topic, you've spent a considerable amount of time examining the operations of the retail business, particularly the logistics. What insights have you gained from veteran logistics operators about the nature of Amazon's network, how it competes with businesses like FedEx and UPS, and how its setup compares to these companies?

We've published detailed information that explains step by step how Amazon's supply chain or logistics work compared to others. It's quite complex and varies based on the region of the country or even specific areas within the US. It also depends on the type of parcel. There's no one-size-fits-all approach.

Let's focus on the North American business for now.

Even within North America, there's no single approach. But to simplify, merchants ship their products into the Fulfillment Center (FC). There's an additional stage where they have a sorting center back into the FC. Products then leave the FC and are typically transported by middle-mile players like Less Than Truckload (LTL) carriers to another sorting hub. From there, they either go by air or truck to a delivery station, which is the last node closest to the customer. The products are then shipped out by a last-mile provider, which could be Amazon's own network, contractors they hire to fill gaps, or they go via UPS or the United States Postal System (USPS), depending on the geography and other factors. We've published estimates on how much each segment ships, for example, how much of the last mile they cover. But overall, there's a significant difference in the unit cost of shipping last mile between Amazon and other players. This difference is due to the density and structure of Amazon's network, which is difficult to replicate quickly.

You're saying that there are structural advantages inherent in Amazon's network design?

It's not too complicated. If you're building something from scratch, like FedEx, which was established decades ago, you should be able to create something with an advantage. If you can't, then you either don't understand the business or there's a problem.

With Amazon's scale to start with, it's fundamentally possible to build something superior. It's very nuanced, and there's not just one thing that gives Amazon its competitive advantage. In my view, Amazon's strength lies in many small, nuanced points or differences that collectively form one competitive advantage. Personally, I prefer a company that has many subtle, incremental differences that contribute to one competitive margin, rather than one big trademark or product as an advantage. This seems to me potentially much more durable and harder to compete with.

Amazon's advantage comes from various aspects, from how their fulfillment centers are set up, the technology they use, their acquisitions like Kiva and Sparrow, to their cost-cutting measures in sorting, picking, and packing. Their shipping process, including the way they receive and sort products at delivery stations, is also a factor. We wrote about single dispatch, which might seem obvious in hindsight, but legacy players typically have one dispatch a day for their express business.

For FedEx, this means that for about an hour a day, usually in the morning, all the delivery drivers back up their vans to the station, pick up the parcels, and then go out on their route and deliver them. The station employees then receive trucks of parcels to prepare for the next day.

Let's say you can ship out 100 parcels at a time, 10 vans and 10 parcels, and you ship out those all in the morning. You can only bring in 100 parcels in a whole day because you have to wait for all the parcels to come to then sort them out for the next morning. What Amazon does is they designed it so you have multi dispatch. You can send 100 parcels out in the morning, 100 at midday, and 100 in the late afternoon. So you're getting 300 parcels through the same fixed asset.

Amazon already ships more parcels in the US than FedEx per day. But what is even more interesting is that they do this with a fraction of the number of delivery stations that FedEx has. By my calculations, Amazon has around 600 to 650 delivery stations in the US, both small and large. FedEx has over 2,000, so Amazon has less than a third. They're shipping more products through fewer stations and they're doing more volume.

It's just a better foundation to serve and do this business. Now, that doesn't mean they haven't got too much capacity. That doesn't mean they've overbuilt. It doesn't mean that they're offering same day delivery when they shouldn't. But fundamentally, from a first principle basis, I don't see how they haven't got a real structural advantage in serving ecommerce demand over the next 15 to 20 years. To me, that matters more than whether they've got a bit more capacity in the next couple of years.

Indeed. This seems to be a part of the view you outlined, particularly a question you asked in our published analysis on the economics of the retail business. The question being, has the earnings power of the retail business significantly improved over the last few years?

If you look at the data, which I'll be publishing in the next few weeks, you can see exactly that. There has been a significant year-on-year improvement, in the last quarter. If you exclude the impairments and one-offs, they're already making improvements. However, when you examine the structure of the P&L, it's not as straightforward as it seems. This isn't like eBay where you have a clean retail P&L. There's a lot of other factors at play here. When you look at the margins or the EBIT margins of North America, it includes a variety of other elements.

Even if you just look at the gross and contribution margin of retail, when you go down to the tech and content line and the G&A line, it's grouped together in the consolidated numbers, both AWS and non-AWS. Even if you make some estimations because you've got the EBIT margin of AWS, you can net out what the opex of retail is. But then within that, you've got a variety of expenditures. They're investing in a range of projects, including satellites, which aren't retail.

There are questions over capacity and shipping costs. For example, they've got to consider how much they can grow into their fulfillment network to reduce the fulfillment costs per unit and how to improve the shipping cost per unit. This depends on how Jeff and Jassy view same day versus two-day shipping and how aggressive they want to be on pushing that.

Then you've got the whole opex line, which includes tech and content, G&A, and a bunch of other stuff, which who knows what's in there that is not retail. You can make some estimates, and we've spoken to some executives that give us some estimates, but all we know is that it's a significant portion.

So, if you assume ecommerce grows, if you assume Amazon has a structural advantage, and if you assume they grow into their network, so the fulfillment and the shipping costs start to decrease, the gross and contribution margins are going to increase. If you were to net out a portion of the opex lines, you've got a large and profitable business with a real advantage, at least in the US, UK, Germany, and Japan.

Then there are other questions around shipping specifically, which is a bit more complex. This goes back to the Costco and the economies of scale framework. But how much do you really need to share? Like, do you need everything being the same day? No, obviously you don't. And frankly, do you need everything to be two-day? No, but getting to two days across the US is probably a pretty interesting spot.

So, I think they've got a lot of room to tweak that shipping line. If you want to ship something same day and the product is in San Francisco to New York, then you have to pay a premium. If Amazon can get to that point, who knows how much that shipping line can improve. But right now it's out of control. That shipping line has gone from being less than 10% of GMV to being 13% of GMV last year. Or even if you go on sales, it's obviously even higher. But on GMV it's gone from 8% in 2016 to 2017 to be 14%. It's a lot. Fulfillment cost has also exploded from 10% to 12%.

How has your understanding of these questions evolved over the last couple of months through your interactions with operators?

I tend to focus more on the long-term aspects, such as the structure and foundations of the company. I'm not overly concerned about whether they've overbuilt. It's quite clear that they've built too much square footage, which is a mistake every ecommerce company seems to have made. I am curious about how aggressively Jassy will work towards normalizing or reducing that space, but I don't think that will significantly impact the earning power in 2030. Admittedly, I'm biased because I believe retail is a safer bet. I'm not as familiar with the AWS business, but I feel more confident about predicting the earning power of retail than AWS in 2030.

Could you share some information about what we plan to publish over the next few weeks and months as we try to holistically understand this business, which is quite complex given Amazon's diverse involvements?

Certainly. We plan to conduct some research on Amazon advertising, a topic we haven't discussed yet but can perhaps cover another time. We'll be publishing a comprehensive review or research on Amazon advertising, the retail business, and more details on the logistical aspect. We aim to understand why or what we think matters versus incumbents to identify the potential 10-year or 20-year advantage that FedEx can't replicate. Despite their efforts, they're struggling to transition to multi dispatch and express business. It's just too difficult. We'll be publishing that, along with a range of other topics. We'll share some of the numbers from our research and our estimates of what Amazon could potentially achieve. We'll also include cost estimates based on our discussions with former Amazon employees regarding what they believe is a normalized return in pure ecommerce. However, these are rough estimates because it's incredibly complex and almost impossible to accurately determine the real return for ecommerce.

You've spent almost six months intensively researching this business, conducting around 10 to 20 interviews with operators. What are your thoughts on the quality of this underlying business? What stands out to you? Has your perspective changed at all?

Amazon is one of the most unique businesses that has ever existed. Jeff Bezos could potentially be recognized as one of the greatest business minds. It's unusual to find a company that sells cloud computing and parcels, and even more so when all of this has been built internally without acquisitions.

This is very different from other companies. For instance, the core of Microsoft is still the operating system and Windows, the software they built. This has been the foundation of Microsoft and it has evolved from there. Amazon, on the other hand, has taken all the free cash from retail and built a cloud computing business. It's fascinating to think about how someone could take cash and build something so completely different. But perhaps it's not so different in the eyes of innovators like Amazon.

What sets Amazon apart is its unique culture, a combination of pure innovation and capital allocation. Even if retail generates 100 billion in free cash flow in 2026 or 2027, as a minority investor, you're not likely to see that in your pocket. Jeff Bezos and Andy Jassy might decide to invest that money in another venture or they might create another AWS. They might even have planted the seeds of a 100 billion dollar advertising business.

Despite some challenges in recent years, what they've built with the amount of capital employed is unparalleled. They have another line, another segment in their accounts that has been growing. They've bought MGM and who knows what they're going to do next? If Jassy and Bezos decide to distribute dividends or buy back stock, that might signal a change in their philosophy. They have always said they want this business and its culture to endure forever.

Just as Warren Buffett has built a diversified, robust collection of assets that reflect US GDP, Bezos has created a culture that is going to innovate its way through. These are two different cultures of how to approach building a durable asset. However, there is a caveat that a poor leader could ruin this. Just as we rely on Buffett to make good purchases, we rely even more on Jassy and Bezos to continue innovating. I may be biased, but I believe that Amazon has an interesting business. Especially considering its current valuation of 900 billion, which I believe is far too low for what this business can earn in 10 years' time.