Amazon Retail: The Rails of E-Commerce

Is e-commerce a good business? Can a company consistently ship ~$15 parcels with free delivery within 24 hours and make money? E-commerce profitability seems to be the center of the argument for AMZN sceptics we speak to. Investors question whether heavy investments in fulfilment centers will yield sufficient returns.

After interviewing an AMZN 3P seller, with 16 years experience scaling small brands and running AMZN 3P accounts for large Fortune 100 brands, we published our analysis exploring the history of AMZN’s Retail Marketplace Strategy, its underlying economics, and offer a new lens to view the business as Retail Infrastructure as a Service.

We share some snippets from our work below. The full interview and analysis is available to members.

Over the last year, AMZN hasn’t been the only one struggling to make e-commerce profitable. Its sellers are too:

With Amazon FBA, it’s 15% listing fee, 15% FBA fee, plus your storage fees, plus various return fees and things like that; hidden fees. They raised the fees this year, in January, then they implemented a fuel surcharge, and then they just implemented a surcharge for the fourth quarter for warehouse expansion…so all these fees build up to the point where we have one client who was on FBA for years and we pulled them off FBA. They can actually build their own warehouse, staff it, handle their own pick, pack and ship for less money than they can with FBA. - 3P FBA Seller

FBA sellers can pay up to 60% of sales to AMZN. This excludes the product cost. A sellers’ COGS has to be below 35-45% of sales to earn money selling via FBA.

Bears argue this becomes self-selecting; 3P sellers can only afford to supply the higher-margin products and therefore leave lower-margin items for AMZN to sell 1P.

If there is one chart that epitomises the bear case, it’s likely this:

Source: AMZN 10-K, IP
Source: AMZN 10-K, IP

From 2015, AMZN’s pure e-commerce profit per unit significantly declined and ads and prime subscription margins have plugged the gap. But is this really so different to Costco or Walmart? Excluding ads and Prime is akin to excluding Walmart’s slotting and placement fees or Costco membership revenue from its EBIT.

Both Costco and Amazon follow a similar strategy: offer unbeatable consumer surplus by selling high-margin memberships. Yet, they are currently yielding different results.

Source: IP Estimates, AMZN, COST
Source: IP Estimates, AMZN, COST

There are two opposing views on AMZN's poor profitability. On one hand, there is the argument that AMZN has overbuilt its FC network in a mature e-commerce market and is ultimately in a bad business of shipping cheap widgets at very low margins. Given the fixed asset growth, AMZN will struggle to earn an adequate ROIC.

The other side, and more through the infrastructure lens, suggests online penetration is early and AMZN has built the foundations to run the rails of e-commerce. As e-commerce penetration grows, AMZN will continue to offer superior consumer surplus but yield even better financial results than brick and mortar competitors.

How will it earn superior EBIT / GMV vs brick and mortar competitors?

By growing 3P sales through AMZN’s rails and, more importantly, leveraging its 1P data through advertising.

In our full analysis, we compare Costco to AMZN, and explore the history of Amazon’s marketplace strategy and the potential underlying unit economics.

In the coming weeks, we will continue to pick apart Amazon focusing on its fulfilment process advantages and the potential third pillar on top of AWS and Retail: advertising.


This was an interesting exchange with an experienced FBA seller:

Q: What about Shopify’s new logistics service they’re trying to compete with FBA?

Where is it? It’s like Walmart coming out and saying we’re going to be better than Amazon; they’re been saying that for 20 years. They were going to show Amazon how to sell. I don’t know if you’ve ever sold on, but it’s no Amazon. The customers are on, that’s where you can sell your product. The question is, how do you use Amazon to make money. - 3P FBA Seller

Topicus & Constellation Software

Ever wondered what it must be like to work with Mark Leonard? A seller of Dutch VMS business to Topicus provides a tiny glimpse into the level of detail at CSU:

Businesses have autonomy and MDs make their own decisions, but they have to report everything. New initiatives are entered into a template, and it needs permission from the top to run it. You can raise prices but if it affects revenues, you need permission, whereas you can change the building you rent or hire new personnel without permission. They employ an 18 month rolling forecast. We bought a competitor of my old company and I thought I could integrate Van Brug modules, but I failed with that. Revenues were sometimes low and each month for three years I was asked why I didn't follow up on my promise. It is all about reporting figures and you are constantly checked. The templates you complete are combined with bigger templates which go to CSI. Mark Leonard knows my KPIs better than I do.

Sunbelt, H&E, URI & US Equipment Rental Demand and Rates

We’re curious about the equipment rental business: it enjoys local economics of scale, flexibility to manage FCF in a downturn, and, weirdly enough, all current management teams are confident top line will grow double-digit even in a potential recession.

We interviewed a Former H&E and Ashtead executive with 25 years experience in equipment rental to discuss major risks through this cycle.

One interesting observation was the potential change in rates as inflation nomralises:

Even though there is positive cash flow in a downturn, an extended downturn will definitely hurt in the long term. That combined with inflation, buying equipment is getting more expensive. It was getting more expensive before the overall inflation that we're experiencing in our economy. The labor costs and additional safety features I've never seen before, and I've had these discussions with people in the industry. The acceleration of the prices of equipment over the past five or six years has been unlike any other period.

Rental companies have been passing on the price increases to customers. This is great, until it’s not.

Once the supply demand equalizes more than it is today, as people fleet up it will equalize. Then you will continue to buy more expensive fleet and you will not have the ability to raise rates. I'm only saying that based on past experience.

But maybe this isn’t too much of a risk? Sunbelt is spending ~$1bn replenishing and growing fleet on $11bn+ total equipment value. As long as there is demand to utilise the fleet, a decline in rates shouldn’t impact margins too much. Especially when specialty is growing and its at 70%+ EBITDA margins!

Fedex Freight vs Old Dominion

This interview with a Former Fedex SVP for 35 years, one of the most senior executives we’ve ever spoken to in the freight industry, shares insights into LTL trucking. This comment on the pros and cons of decentralisation vs centralisation at the terminal level stood out:

because of the centralization of FedEx, they had done away with P&L at site-level. Each of ODFL’s terminals has a P&L that the center or terminal manager is responsible for and the sales team reports out of that organization, so there is a lot more control and focus. In FedEx, the sales team is literally part of another unit – FedEx Services – which is IT, sales and marketing. It’s a double-edged sword; on the one hand, if not managed well, the sites will optimize for themselves and not for the overall business and that becomes a problem, but it also puts a high degree of focus at each terminal on hitting their profit targets. There are several flaws with site-level P&Ls because you have to allocate certain costs and there is always debate about how much SG&A to allocate, but at the end of the day, it does give the terminal higher motivation and incentive. - Former Fedex SVP

Azenta Life Sciences, Biostorage, & Medical Lab Equipment

Through acquisition, Azenta Life Sciences has quickly grown into a leading biotechnology services company. This interview with a Former Brooks Automation executive and current customer of the business, discusses the strengths and areas of improvement for Azenta:

"That is the biggest problem, hands down, that they have not done a great job bringing in all of these disparate companies and shaping a culture that's conducive between the different lines of P&L."

Azenta acquired GENEWIZ in 2018 to offer complementary services. According to the executive, the sales force has not been well integrated despite a large overlap in customers. If Azenta can address this, it could provide a strong foundation to capture a secularly growing market.