Amazon Prime's Sports Rights Strategy

In 2019, Amazon experimented with a purchase of 20 Boxing Day Premier League football (not soccer!) games in the UK. This was the company's first acquisition of sports rights. Over 4 years later, we estimate AMZN will spend up to $5bn per year on sports rights globally by 2025.

This interview with a Former leader of AMZN Prime Video's Sport business is an exploration of how AMZN acquires, prices, and monetises sports rights. We explore how Amazon chooses certain rights packages, calculating LTV, and how it competes with ESPN, NFLX, and now Apple.

Live sports is unique; fans both buy subscriptions and watch ads. AMZN saw record new Prime subs after acquiring the NFL and Premier League rights.

Not only does acquiring live sports rights drive new Prime subs, but AMZN also calculates the gross profit from net parcels shipped and net ad revenue shown on Twitch or Prime based on customer segment and demographic.

if I'm buying Champions League rights in Italy, I might acquire three million subs on Prime; I'm just making numbers up. Two million are male, and one million are female. Females normally buy 50 parcels a year, and males buy 10, and then you work out the lifetime value estimates of each…And it's a hugely collaborative effort because each country has its own P&L, so Prime Video says, we want to acquire Champions League football in Italy, for example. With Prime Video, we can drive this much acquisition engagement and advertising revenue, and then Prime Italy, in retail, will say, we think there’s this benefit to it, which equals X. That totals how we evaluate revenue and what we're willing to bid for. - Former AMZN Prime Video Executive

AMZN isn't the only new entrant acquiring sports rights; Apple have recently dipped their toe in with the MLB. Netflix has also been inquiring.

Apple acquired MLS, globally, and MLB, so they've pulled the trigger on US sport. The MLS Global sport was interesting because that was the first time a global deal had been done. They're making moves into the space. Netflix, I know they inquire, but I don't know that they've actually bid. - Former AMZN Prime Video Executive

We believe it's only a matter of time before NFLX buys live sports rights. Assets like NFL or NBA are so expensive that the economics would break if NFLX only had SVOD business model. The combination of AVOD and dynamic ad insertion changes the economics for NFLX.

The challenge is that dynamic ad insertion is not available across every country, so at the moment, it's very much the one ad to everyone, and then you can prove on the measurement side what return it's driving. The real win is to have dynamic ad insertion, so I know you're watching, or if you like this content, you've bought this previously, then I'll serve it to you. - Former AMZN Prime Video Executive

However, doesn't this just increase the stakes for AMZN to play? It now spends as much as NFLX per year on video and music. With near 100% household penetration, spending $1bn per year to acquire NFL rights must lead to a lot of new parcels shipped or ads served to make the economics work. And the CPM's need to be closer to $100 rather than $20. This is only possible with dynamic ad insertion on CTV. We will explore this next month when we publish our research on AMZN ads and TTD.

Future of Google Search?

Fascinating interview on the future of Google Search with a Former GOOG executive who spent over 12 years at the company working across all products.

There is an interesting example of how Google’s experience and tech for GOOG Assistant provides an insight into how new conversational-like Search UI’s could replace the current search user flow:

The fundamental development model was still much more heuristics based, rather than learned based. What I mean by that is, all of the use cases that worked really well and reliably, were hard coded in. Somebody had gone in and said, "Here are 43 different ways you can ask for what the weather is". Sometimes you would say, "Palo Alto weather", sometimes you would say "Weather in Palo Alto", sometimes you would say, "What's the weather tomorrow or day after?" We have language scientists writing all this down, and then the computer would basically go through a bunch of if/then statements. That just becomes really cumbersome once you get to two or three levels of back and forth, because there's just so many different permutations that each area is creating. I think that we hit that ceiling of, okay, we can't do complicated things this way. You can't code in 'how to order a pizza', because there's literally 10,000 different possibilities of how you would order a pizza. Then Domino's will introduce a 11th one, and now suddenly there's 10,000 more in line, and now there's 30,000 more and on and on. I think that what's really interesting is some of these large language models, which are based on different technology that are not so heuristic driven, but rather learned models. I think they can facilitate a richer conversation.
Now, the downside is, they're not super reliable, because you don't know what this model has learned. Nobody has programmed it in, and so you may end up with Coke on your pizza or something because somebody wrote some blog and the thing crawled it or whatever. I think that's the cutting edge of where this technology is, which is, how do you get the best of both worlds, which is the breadth and capability of the large-scale models, but the reliability and safety that people would expect from a service from Google around accuracy

It’s hard to believe any company other than GOOG with its Search data can fall behind in training such models. We’d be surprised not to see a material update of GOOG’s capabilities by year end.

Is Constellation Software's Employee Churn a Problem?

This interview is with a Former CSU Portfolio Head who spent over 12 years at the company and now works for a 'competitor'. The discussion focuses on the potential increased competition to CSU.

One driver of competition that continually comes up in our research is that of employee churn. More specifically, those senior employees that leave to create similar vehicles to acquire similar software businesses. We have a list of over 25 small private CSU replicas and many of the leaders previously had somewhat senior roles at CSU.

The issue is preserving the people. They have a strong culture with strong intrinsic motivation, but the extrinsic is missing, which is why I left. They couldn't keep up with compensation in the market. Decentralized decision making means it's up to managers of the operating companies or portfolios to figure it out, and there wasn't a lot of proactivity to do that. That is problematic for M&A which is the bread and butter of Constellation and you've seen several departures there. Times are no longer so good, so Constellation could suddenly be a great place to be. The trickier piece is how to incent CEOs and management teams to focus on organic growth, because the compensation scheme is not set up to motivate that type of behavior. - Former Constellation Software Portfolio Head

Given CSU is in the capital allocation business, is the company losing important capital allocators?

CSU's proxy suggests 'executives' are paid an annual cash bonus based on ROIC and net revenue growth. 75% of this bonus is reinvested in CSU common equity locked-up via escrow for a minimum of 4 years. However, it's not clear exactly how many individuals participate in the scheme. Who really has the chance to get into this bonus scheme? How much does a Volaris M&A Head earn relative to competition, for example?

But how many employees really should be included in the bonus scheme? Maybe employee churn isn't such a material risk to CSU because the model is such an ultra-pure play on niche, VMS companies. It acquires <$6m, legacy market leaders in small markets. It's a crystal clear playbook; as close to programmatic acquisition as possible. M&A employees are scouting deals and following a clear playbook. How much shareholder value are they really adding if there are such strict guidelines?

Although we believe only those with responsibilities of the whole organisation should be compensated with equity, the cash compensation for those below the top level of seniority should be highly competitive. As CSU delegates M&A responsibility down the organisation, incentivising and retaining junior M&A employees could become a serious limitation to growth. Or worse, lead to increasingly deteriorating ROIIC due to poor capital allocation.

We see organic growth as more of a risk. Mainly because CSU acquires 100% of legacy VMS companies and the founder leaves / checks out. With CSU's comp plan, how does it really incentivise those closest to the VMS opco to drive organic growth? Questions for the AGM possibly.

Lousiana-Pacific & US Building Products Competition

Louisiana-Pacific has refocused its business on higher-margin building products. The company is currently strong in the Midwest and is focusing its efforts on grabbing market share in the coastal markets.

This former LP sales executive describes the strengths and weaknesses of LP's products and the opportunities to expand into other markets.

On new construction, it is builder-driven. They’re going to offer a product and they’re typically going to get rebates from one of the large companies, whether it be vinyl etc. That’s where the battle is fought and that’s where we’ll win or lose new construction; it’s mainly with the builder. For repair/remodel, it’s with the installer or the one pitching it, which is usually the installing company. I was part of a session where we talked to individuals about their siding choice, and the customer doesn’t know LP outside of the Midwest region; even in Illinois, the customer doesn’t know what it is. They know Hardie because a Hardie board is kind of like Kleenex. They know what vinyl is, and if the homeowner wants to go bigger and better, they’ll choose Hardie. The installers love LP because it’s easy to install but they don’t do a lot of selling, they don’t want to do too much selling, they want the easier sale and vinyl has been the easier sale."

Epiroc vs Sandvik

Sandvik and Epiroc dominate the mining equipment market. Over two thirds of Epiroc's revenue is from the aftermarket. This Former Glencore Director highlights disintermediation risk that could affect the profitable aftermarket earnings stream. With >45% of Epiroc's EBIT from the aftermarket, a decline in spare parts volume or margins could be a significant risk.

"Not OEM, Epiroc do not make filters and buy them from Donaldson Filters, who will supply at 50% of the cost. Traders were also bringing the same Epiroc parts at much lower prices. We initially thought they were pirated spares, but later found out they were the original Epiroc or Sandvik spares. They opened up their businesses in America and offered country wise pricing because it has become a global market. Before people sign contracts, they benchmark spares prices to see if it makes sense. Mining is done for the world and everybody has hand to mouth cost controls. You need to strategize because sometimes CAT offers better notice so it depends on your mine. Most mines are trying to optimize the cost of spares so there will be more competition. - Former Glencore Director

Marlowe PLC

In 2020, UK-listed Marlowe acquired Ellis Whittam for £60m, its largest acquisition to date. This led to 11 further acquisitions and a rebranding to WorkNest which now creates 50% of the GRC segment’s profits.

This former WorkNest executive describes the business and how it was eventually acquired by Marlowe:

"They have started to do it with bolt-on services to the employment law, HR and health and safety products, which are becoming increasingly important. A good example is HR software, where you can input employee details so they can request holidays which their line managers can authorize through the software. It can also record sick absences, return to work interviews and run your annual appraisal processes. Ellis Whittam had a poor version of that. Just before I left, Marlowe bought Youmanage for just over £1 million. At that price, you either get something very early stages or not very good, and Youmanage is not good. Peninsula have their own in Bright HR, which is absolutely superb and good enough to be a lead product. WorkNest currently have a bolt-on product to their employment law and the HR service which is not good enough to be a lead product."

With the Optima Healthand VinciWorks acquisitions, it is clear that Marlowe is adding software services to its more established business service companies. However, there seems to be doubt as to the effectiveness of the integrations between the different companies and offerings.