There’s a fun tidbit in this massive must-read article on Kotaku about eSports. Here’s the quote I love:
A 2017 Morgan Stanley report leaked to Kotaku claimed that, in its first year, the Overwatch League could conceivably generate $720 million in revenue, about the same as World Wrestling Entertainment. By 2022, says Goldman Sachs, viewership of pros playing competitive games…may be on par with the National Football League’s viewership today.
Those estimates are just…nuts. Here’s the thing. I believe those reports made them. What I am more skeptical about is if those reports gave their confidence intervals. Basically, was Goldman Sachs estimating their prediction, or their best case scenario? The best case isn’t a prediction, but your hopes and dreams.
My goal is to never do that. Here’s the thing about predicting the future: it’s hard. A lot of the techno-futurists and streaming-vangelists of the world are better than me in that regard. They know how the streaming wars will end and can predict that with unerring accuracy. I’m not that talented.
Instead, I try to tell you how this business works and try to do so with some modesty. Even with my backwards looking estimates, I want to give you my 90% confidence intervals. That’s why for my big, big analysis articles (explainer here) I build multiple scenarios. (I also try to show you my math. And explain what and how I did it.)
Which brings us to Game of Thrones. In my Decider article, I had to leave out some explanations for space. I’m taking those and giving them to you now in a Q&A with myself I’m calling my “director’s commentary”. (See Part I here.)
Let’s start with uncertainty. What was the biggest variables in the model?
I have two areas that could really swing the model. First, in other revenue, home entertainment and merchandise could have been even bigger than I thought. Again, in winner-take-all, as box office regularly and reliably shows us, the winners are multiples larger than everyone else.
Then, in subscription revenue, the numbers again could be huge. And since I don’t have the actual viewership behavior, it’s really a question of attribution. As is, I had it at 57% of total revenue but there’s a real possibility that this show is even more important for HBO, especially on the international and digital sides.
What about costs? Are they certain?
The costs are pretty relatively certain. The production costs get leaked every season it seems and even marketing costs were leaked this year.
So everything else you’re very, very confident in?
Well, no I’m always wracked by uncertainty. Which in a columnist, I understand is a weakness. (If you fail to project anything less than strength, they’ll get you, like Theon getting got by the Iron Born in Game of Thrones.) At a high level, here’s my confidence in my inputs:
As you can see, there are always lots of estimates in a model like this. And estimates are better than “guesses” but a far cry from leaks or facts. And even estimates can range from pretty high confidence like the merchandise take or studio distribution fees, to pretty all over the place like tax credits. (I’ve seen the rates for Ireland, but it’s unclear how much GoT films specifically in Ireland, so hard to know. There have been some leaks, but only selectively for certain years too.)
So the revenue estimates will provide the biggest range in the final estimate?
Yep. Oh and one more. Profit participants.
(I had called this “talent participations”, but agents aren’t talent. Shade thrown.)
Specifically, the actors and GRRM. I’m fairly confident the showrunners are getting 10% of the MAGR profits. My gut is GRRM is between 5-10% of that (either matching the showrunners or equal share to them), and the actors could be in that range. And for the initial agreements, I’m pretty confident in those guesses. But…
…what about the renegotiations?
That’s the big uncertainty. After season 4, this show is a monster hit, and HBO needs everyone on board to see it to its conclusion. That’s where the power starts to shift to the talent. And even if this show loves to kill off major characters, killing off Jon Snow or Daenarys Targaryen just seems unlikely. So how much higher did all the talent negotiate in the backend? I don’t know.
Okay, let’s get to the high and low cases now that we know the uncertainty. Can you walk us through the changes for the high case model.
From the top.
My working theory is that Game of Thrones is the biggest TV series in terms of units sold this decade, and maybe this century. But is that over a billion dollars in retail sales, as I’ve heard? Again, I heard a rumor around 2016 that Game of Thrones had already earned a billion dollars for HBO. That’s through just six seasons. Of course, I didn’t ask the source if that was gross retail sales or net receipts. But that’s my starting point for a high estimate.
For the high case, I used that estimate I had that GoT did about 1 billion with a B in home entertainment. I assumed that wasn’t including digital, and that got me an estimate of about 25,000,000 units sold. So I used a multiplier of 4 on my initial home entertainment estimates. This seems really high, but this fall, collecting the entire series on DVD will be a big seller.
As for the low case, I didn’t want to drop it too low as we had a real starting point in the sales for the start of the series. So I used a .8 drop. For digital, I used the same multipliers as physical.
First, is the home entertainment bucket. Or even more? I spent a lot of time looking for this, and couldn’t find any estimates, so had to use my own estimates.
I tried to come up with a higher fee for the international sales, but I couldn’t do it. I think my initial estimate is probably the top range. HBO sells their series to a lot of cable/satellite providers as a library deal, which limits the amount they can renegotiate. This limits HBO’s downside but also their upside. Further, I’m pretty confident in the starting point since that number was leaked to the press. However, I did make a low case by assuming that HBO couldn’t get the 50% of production costs to go up after season 4, instead having them increase at a fixed rate.
Library value is one of the very real but very future estimates in this whole process. I assumed that each episode would bring in 125% of its production costs going into the future. To make this number, I’d been playing with this table that projected how much money HBO could make of Game of Thrones for the next decade and beyond.
This library value is a combination of things. First, how much value will GoT have in attracting and retaining subscribers in the future for HBO? That will be very real as people will keep binge watching this show in the future. Especially when the spinoffs appear. Also, there is always the chance that HBO will sell the rights to Syfy, Amazon or whoever in ten years for a huge, billion dollar sum.
For my high case, I assumed $2 billion over the next ten years. Ask yourself, would Amazon pay $20 million per episode per year for the next five years? I’d say so. So that estimates feels reasonable. I used $564 billion for the low case–125% of production costs, discounted–or $7.7 million. Of course, if you did that you should account for the time value of money. Which is why my numbers are lower for the library value than the straight production costs.
Wait, you mentioned the time value of money. You love that. Did you factor that in on your initial GoT estimate?
No I didn’t. I’ll explain why next installment. Let’s stay focused.
Okay, any more revenue changes to the high and low cases?
The tales of Game of Thrones merchandise are legendary. Beer. Wine. Scotch. Board games. Puzzles. T-Shirts. Jewelry. That’s a lot of expensive stuff.
Again, this series was almost undoubtedly the biggest seller of merchandise for all primetime dramas. Ever. I’m confident in saying that because before GoT, adult TV merchandise just wasn’t a thing. As a bonus, GoT tied in the fanboy element with prestige drama. (Most of the other big franchises appeal to kids—Marvel, DC, Star Wars—whereas if you watch GoT with your elementary school children you should report yourself to child protective services.)
But if you look at my model, I don’t really have GoT making a ton of money on merch. That’s because in most licensing deals, companies only take home a percentage of the wholesale price, and that percentage is about 10%. That means in reality you’re at something like 5% of retail sales. So here’s my initial estimates:
(The blue is because I have these numbers coded as “inputs” in my model.)
So trying to push that up as high as I can, let’s say they got to about $1 billion in total merchandise this year. That’s about a third of what Star Wars did at its peak, and kids properties have many more product categories from toys to bed sheets to books and what not. So I can’t push that much higher. On the low end, I cap it out at about $400 million per year. I mean, I bought a lot of GoT branded beer. That gives me a low range of $2 billion to a high of $4 billion in total retail merchandise sales. That feels right.
You left out subscription revenue.
Yeah, that was the hardest to recalculate, again with a lot of uncertainty.
But given that you have subscription revenue generating 57% of the revenue for HBO, you had to make some changes, right?
I did, and that’s where the value of running different scenarios or simulations can start to be seen. The subscription revenue has four magic numbers—which I described last time—and another good estimate, but one that can definitely change over time/be off. And small changes in retained subscribers can have huge downstream impacts because of the customer lifetime values. So I updated the tables to these:
I left the CLVs alone from my initial estimate because—besides months subscribed—most of the other numbers are fairly reliable. I felt limited on the upside case by HBO’s profit. Basically, I just couldn’t attribute all of HBO’s profit to GoT. That’s too high.
Any other specific changes to the high case?
Not really. I left all the costs alone. Again, the production costs are pretty reliably leaked, and residuals are a specific calculation. (The residuals will go up as home entertainment and library sales go up.)
As for fees, the imputed license fee is a good deal for HBO in my estimated model, so it didn’t change in my high case. However, I kicked up HBO’s distribution fees to 40%. I don’t have experience with HBO specifically, but studios can find lots of ways to charge fees for selling in additional windows, with marketing expenses for everything. Also, for talent, I kicked their split down another 7% points to 20%. That feels low to me, but in the upside that HBO negotiated fantastically well, I could see it.
Can we see that high case?
Sure, here’s the bottom line result. I kick up the profit to $4.2 billion. As a reminder, this is the upper limit of my 90% confidence interval in what GoT.
Does the high case seem too high?
It sounds big right now, but not when you think that the show started in 2011 and will be valuable for decades to come. (Like Friends or Seinfeld or the Simpsons or Breaking Bad.) This is four billion over two decades. Especially the $1.5 billion in library value. HBO doesn’t have that money yet. It will come in over the years. Even the $2.5 billion in subscription revenue is spread out over 2011 to 2019, meaning it’s several hundred million per year of HBO’s $4-6 billion in revenue every year.
Did you many any specific changes to the low-case scenario?
Yeah, I made everything work against HBO. So the imputed license fee jumps up to 100% after season 4, assuming the talent negotiated really well. I also kicked talent participation to 35% and HBO’s distribution fee back down to 20%. Finally, I assumed that the leak that HBO only spent $7 million on marketing was a lie, and HBO spent $20 million per year from season 1 to 7 marketing this series.
Can we see the low-case?
Here you go. The headline is $1.15 billion in profit.
Wow, so again a pretty wide range, but you don’t think they’re losing money.
Yeah, a $3 billion dollar range. And yeah, I just can’t manipulate the numbers to get to a loss. A lot of people claim online—don’t believe anything you read on the internet—that every movie or TV show loses money. Frankly, even the best accountants can only rig the game so much.
HBO could/will ignore all the subscriber value, but I think as soon as season 1 hit the market, the international buyers grabbed 50% of the production costs, and the imputed license fee covered the rest. That’s my line 30, and even without subscription revenue this thing made money. Being top of the mountain in both home entertainment and merchandise just pushed it over the top.
Any other thoughts on the range being wide?
Yeah, as I built these scenarios, the subscriber attribution does an absolute ton to boost revenue. That’s a $1.6 billion range with just tiny manipulations of the data. Further, the library value is totally speculative on the future. Again, predicting the future is tough. At least, if you’re trying to predict it accurately.