Top 10 Most Popular Articles in 2022

1. Amazon: Deconstructing the Retail P&L

2. IP ANALYSIS: Halma, Danaher, CSU & Serial Acquirer Org Structures

3. IP ANALYSIS: AWS: Long-Run and Normalised FCF Margin Estimates

4. IP ANALYSIS: Amazon Retail Unit Economics

5. Constellation Software: Culture & M&A

6. Carvana Finance GPU Analysis

7. IP ANALYSIS: Costco's Changing Philosophy

8. Ferrari: Values and Culture of a Prestigious Automaker

9. AWS, GCP, Azure, Oracle & Cloud Hyperscaler Competition

10. IP Investor Dialogue: Investor Dialogue: Carvana vs CarMax

5 Primary Research Insights from 2022

Amazon Retail P&L Deconstruction

In 2022, three out of the top 5 most popular pieces were on AMZN. Somewhat counterintuitively, Amazon, a service and brand nearly every US household knows, may be the most misunderstood large cap stock globally. We believe this is partly due to their poor disclosure, but also because retail and cloud computing are very different businesses.

In our research, we often came across investors who understand one segment far more than the other. In fact, the thesis of most shareholders was solely based on AWS and not Retail. This was interesting to us and suggested AMZN was more complex or less well understood by investors than we originally thought.

We spent Q2/3 splitting the P&L into Retail and AWS in an effort to estimate the current and long-run EBIT / GMV for the core e-commerce business. We conducted a few interviews and wrote our learnings in various pieces of analysis.

One insight from the work was that the e-commerce gross margin has been stable at ~23-25% over the last 6 years but the contribution margin has declined significantly due to huge fulfilment expenses. Same and next-day delivery is very expensive.

Source: AMZN, IP
Source: AMZN, IP

By splitting out the retail P&L, it's clear advertising revenue has been plugging a huge gap in the e-commerce unit economics.

Source: IP, AMZN
Source: IP, AMZN

Another insight from our primary research was that ~15-20% of operating expenses in the 'Technology & Content' Retail P&L could be for ‘science projects’ unrelated to e-commerce. This distorts the underlying online retail economics.

As FC utilisation and e-commerce penetration increases, we believe a fully-loaded EBIT margin for AMZN's standalone e-commerce is closer to 10% than 5%.

But this doesn't mean shareholders will see all this income. This would mean it's Day 2. And this is half the challenge of owning AMZN's equity today: you have to trust management to allocate huge $ of FCF into finding the new AWS. And it's possible that Jeff's ego, the current retail economics, and the performance of some of the current bets make this too much of a hurdle for some investors. Hence why the stock is flattish since 2018.

In 2023, we're watching Jassy's actions carefully to understand how serious Amazon is about proving its unit economics. If he significantly cuts the retail opex lines in 23, the narrative may change pretty quickly.

Our work on AMZN advertising was also a huge insight into the non-endemic ad opportunity that we will be discussing next week.

CSU & Serial Acquirer Org Structures

The popularity of our analysis on Halma, CSU, & Serial Acquirer Org Structures is interesting. We spent Q1 trying to understand how CSU is the only company in history to consistently buy over 100 small companies each year. CSU completed over 134 transactions in 2022, one every other working day. The scale of this M&A is unparalleled. We are curious as to why other serial acquirers like Halma or Indutrade struggle beyond 20 transactions per year.

In short, Leonard’s genius is that he has built a system to delegate M&A the furthest down any organisation.

If Danaher has the DBS, Leonard has built an adaptive system that has institutionalized M&A principles all levels of the org. These M&A principles filter down from Leonard to Operating Group’s who train Portfolio Managers. The idea is that PMs can scale their BU’s into an Operating Group at CSU. Like Topicus, at a given scale, Operating Group's can then be spun into separate listed entities and the process repeats. - In Practise Analysis

We also share other insights to why VMS is more scalable and easier to programmatically acquire than the niche engineering businesses that Halma, Judges, Indutrade or Danaher focus on.

The key takeaway from our work was that CSU could in fact be at an inflection point. If Leonard can combine larger, cheaper acquisitions like Allscripts with 100+ <$5m deals, CSU could grow for longer than we originally expected at similar ROIC's.

Costco's Changing Philosophy

The insight from our interview with a 30+ year veteran at Costco is that the company culture has been potentially changing since Sinegal retired.

For instance, the Kirkland Signature water, the 16.9oz bottle. I saw buyers working tirelessly to get to $2.99. This became the Holy Grail because $2.99 is unheard of. But the vendor made a case to make more margin so Craig [and the leadership] let them take a 20-cent price increase. Jim would never have allowed that and would have challenged everyone to find efficiencies and cost savings. - Former Executive at COST

COST is renowned for its strict gross margin discipline and deep collaboration with suppliers to cut costs to continuously add value back to customers. Under Craig Jelinek's leadership, there seem to be slight changes to the gross margin policy.

More recently, we also explored Costco’s international opportunity which shares insights into the potential store count and why Canada has a structurally higher margin than the US.

AWS Long-Run Margins

Following on from our AMZN retail work, we spent Q3 deconstructing AWS' normalized FCF margins. We interviewed various executives on running datacenters, Graviton, Kubernetes, and AWS' risk to long-run margin pressure.

Given our lack of technical experience, we've historically felt more comfortable underwriting the moat of AMZN retail vs AWS. Our big concern was AWS' long run margins.

This question is difficult to answer given AMZN's opaque disclosure masks how EC2, S3, and other PaaS services like Redshift drive margins. And it's very difficult to even split these business lines up given they are so intertwined. But the potential insight is that maybe AMZN's moat is simply in its physical assets; it's all about driving more compute through its data centers. Maybe AMZN doesn't mind losing too much revenue to SNOW, as long as SNOW customers run compute through EC2?

If we assume Redshift has SaaS-like 75% gross margins, AWS will be losing high-margin revenue to Snowflake, but gaining more compute and storage revenue. Greater EC2 and S3 revenue fills data center capacity and will drive higher gross margins. The net margin impact is unclear. - IP Analysis

In various pieces of analysis, we share insights on the three major risks we see for AWS: losing high-margin software revenue, heavy price competition with Azure / GCP, and EC2/S3 being displaced as a cloud computing platform. We also share why the risk of Kubernetes and open-source software commoditising compute providers is overblown.

Overall, AWS is an infrastructure business, just as retail is. Cloud computing is infrastructure that frees up dev time to focus on more productive code similar to how FBA allows merchants to focus on producing great products. However, unlike retail, cloud computing is an oligopoly at worst and duopoly at best.

We explore why we think AWS' margins are sustainable through 2025 and shared our analysis and an estimate of normalised FCF margins for AWS.

Source: IP, AMZN
Source: IP, AMZN

If we assume AWS grows revenue at 10% for the next 2 years and earns a normalised 30% FCF margin, AMZN is trading at 29x AWS FCF only.

Ashtead’s Increased Resiliency

The key insight from our Q4 work on AHT.LN is that the growth in specialty revenue and more rational rental pricing architecture has made Ashtead and United Rentals far more durable businesses today relative to 2008/9.

In the GFC, dollar utilisation dropped 35% as rental companies flogged equipment at any price to increase utilisation. Today, services like Rouse provide real-time pricing to help rental companies manage ROI's across SKU's. This pricing discipline was put to test during COVID:

Pricing optimization and the ability to price to a mobile app wasn't around; we all priced from a piece of paper with stated rates and applied an available discount. From a discipline standpoint, you saw it shining in 2020 during Covid. The fleet dropped 5% and utilization went down to 58%, and I thought rates would drop 8% to 10%, but they only dropped 4% to 5% because they knew what historical prices were. Rouse has over 250 US customer's data which show the rates on a forklift 90 days ago, so we need to price at that rate or above. The industry did a good job controlling the fleet level and not forcing themselves to over discount. The utilization went backwards in 2008 and 2009, and everybody rented equipment to cover interest and depreciation. - Former VP at URI

Combine pricing rationality with 25%+ diversified, non-construction specialty mix with 100+ ROI’s and companies like Sunbelt and United look much more durable than 10 years ago.

Thank You

Thank you for all your support last year. We couldn’t be more excited about the future of In Practise.

We continue to be highly selective in partnering with funds that have a quality-tilt and low portfolio turnover. On average, we believe these funds produce the best primary research to evaluate the long-term prospects of companies. The consistent feedback we receive suggests this layer of curation produces higher-signal executive interviews and is complimentary to more scaled libraries in the market.

It’s also interesting to note that 3 of the Top 5 most-viewed articles in 2022 were not interviews but our own research analysis. This reinforces our belief that investors don’t necessarily want interviews, but wish to understand, build or test an investment thesis. Although we plan to double the interviews published in the next 12 months, we are increasingly focused on differentiated, non-interview content. This is where we can excel. Expect more analysis and new, differentiated content formats.

We’ve spent the last 2 years building the team, brand, and technological infrastructure of a best-in-class research platform for a very specific user: a long-term, quality-focused public equity investor. On the foundations of a quality, curated primary research service, we plan to combine software and differentiated content to build an enterprise investment research platform solely for quality public equity investors. This excites us - roll on 2023!