Tag: Box Office

Why I Think Theaters Will Return in May: Forecasting When Society Can Reopen

A character in The Sun Also Rises described going into bankruptcy as happening “gradually and then all at once.” This apt description has been applied to everything from debt insolvency to failed democracies. 

The US recovery from Covid-19 will likely feel the same way.

Right now, most of the coverage I read is fairly pessimistic and cynical. On one hand, I get it. The US just had an awful winter. Covid-19 will claim 500,000 deaths in the US alone, and that number is growing. Even positive coverage is framed with caution:

IMAGE 1 - Calm Variant Storm

IMAGE 2 - Experts Still Worried

IMAGE 3 - Reopening v03

On the other hand, doesn’t the US have a lot of good news on the Covid front? Is this pessimism still warranted? Forecasting doom and gloom could be as inaccurate as forecasting sunshine and roses. Where are we really headed as a society? Given the huge implications for the economy, this is a tremendously important question that very, very few people are answering:

When will society reopen after the vaccine rollout?

For my purposes, a lack of forecasting about the future is particularly painful. While lots of entertainment is consumed at home (TV, streaming, video games) lots more is experience live and outdoors (theaters, concerts, sporting events, theme parks). As I usually do when I can’t find an answer to a question I need to know, I tried to answer it myself. 

So here’s a report on what I found. The outline of this (long) article is roughly:

– Bottom Line, Up Front
– The Problem: the legal/regulatory thresholds for society to reopen
– The Assumptions/Inputs: the key assumptions/inputs to build a reopening model
– My Hypothesis: Effective vaccines widely distributed will crush the death toll
– The Model: My results!
– The Key Metrics To Track Going Forward

Let’s get to the results.

Bottom Line Up Front

I’m optimistic that theaters will be open by May 7th across the US, including New York and Los Angeles, the day Black Widow is set to premiere in the US. Black Widow itself may not premiere on that date, though whether or not it moves dates in the next few weeks will be driven by uncertainty, since Disney will have to make the call in about four weeks.

After building a vaccine distribution model (with three scenarios), I’m confident that by early May, between 33%-44% of the population will be fully vaccinated, and up to 61% will have had a first dose. Given the initial data that even one dose provides nearly full efficacy, this level of vaccination will likely decrease deaths and hospitalizations by 76% or more.

It is much more uncertain what the volume of cases will look like by May 7th. The key metrics to focus on, for studios and other entertainment companies, will be the pace of vaccinations, the current case loads, and the rate of death.

(Initially, I had a series of caveats to this controversial article. I can hear the skeptics, “You’re not an epidemiologist so why should we listen to you?” To save space, I’m moving those to their own future article. To address the key caveat, yes, I’m not an epidemiologist. 

Frankly, most companies cannot afford to hire epidemiologists to forecast disease outbreaks. And even if they could, as we’ll see, their models have their own uncertainty. Instead, many companies will likely have their strategy teams doing this analysis themselves and/or rely on models from companies like McKinsey/Goldman Sachs, which are, you guessed it, put together by people like myself. The difference is I make my models/methodology public.)

The Problem: When Will Theaters Open Specifically?

Today, I’m focused on the movie theater problem. (Hopefully, I’ll get to sports and concerts in future updates.) The problem for big studios is distribution: when they release a big tentpole film, they want it widely available. Specifically, this means Los Angeles and New York, and other big cities. They’d ideally want theaters at 100% capacity, but would likely accept 50% capacity constraints.

Thus, the question is whether those specific cities (and America broadly) will be mostly reopen by May 7th, which is when Black Widow premieres. (Or Memorial Day weekend, 28-May-2021 when Fast 9 is set to premiere.) The studios will need to make this call about eight weeks prior, so that they can schedule promotional advertising campaigns. Disney has said they’ll likely make the decision on Black Widow in the middle of March.

To reopen theaters, local counties will need to have Covid outbreaks under control, using whatever guidelines local states have established. In California, for example, this is a series of “tiers”, of which Los Angeles is firmly in the first, most restrictive tier. From the Los Angeles Times:

IMAGE 4 - Reopening v01

And our status under those definitions:

IMAGE 5 - CA Status LA Times

(Though, like all things Covid-19, there is actually a more restrictive tier of “stay-at-home orders”. California’s stay-at-home order was lifted in January.) 

Tying this to theaters, in California, “red/tier 2” counties can have theaters at 25% capacity or under 100 people, whichever is lower. “Orange/tier 3” counties can have 50% capacity or 200 people, whichever is lower. “Yellow/tier 4″ presumably would be fully open.

Covid Act Now, uses a different set of guidelines, and you can see how various metro areas stack up:

IMAGE 6 - Covid Act Now Metro

No matter which numbers you use, the same basic correlation is the same: The lower the number of cases, the more indoor businesses which can be open, which includes theaters.

As cases decline, so do deaths, after about a three week lag. Crucially, as vaccines are widely distributed, deaths may decline even further and faster than cases. If deaths drop and stay low, even if cases come back, some folks could argue that the values of reopening society outweigh the risks of disease spread. (In short, if a pandemic is raging, but no one is dying, is it a pandemic?)

So really there are two scenarios for reopening. One is the “public health” requirements, which means eradicating cases. The second is the “demand side” requirement, which is that deaths and hospitalizations are low enough that society wants to reopen.

The Key Assumptions/Inputs

The reason I couldn’t build a model before 2021 on Covid-19 was that forecasting the pace of the disease was nearly impossible. With so much uncertainty and unknown variables, most models have failed to effectively forecast what will happen in the next two to three months. 

Vaccines, on the other hand, are an easier modeling challenge. We have fairly reliable data sets on distribution and fairly robust (and growing) knowledge about how effective they are. Let’s explain those inputs/assumptions.

Input 1: The pace of vaccination is increasing in the US, and could accelerate further.

No matter how you slice the data, the rate of vaccinations is growing in the US, and it’s a story that most outlets mention, but with the pessimistic caveat that everyone wishes they were even higher. (Yes, we do.) But they’re getting higher every week. In a few weeks, they will be shockingly high. Here’s Our World in Data’s look:

IMAGE 7 - OWiD Vaccine Rate

Or take Kevin Drum’s take, looking at peaks:

IMAGE 8 - K Drum Update

The US was able to double vaccination capacity in ten days starting in January 4th (350K per day, seven day average) to January 14th (700K). I start with January 4th, since that’s when most healthcare workers returned from Christmas break, and all state and local governments focused full-time on the vaccination program. Then, it took 26 days to double again (from January 14th to February 9th). Depending on when we get to 2 million doses, it may have taken only 30 days to have doubled from 1 million doses to 2 million. (We’re currently at about 1.7 million per day.)

We will likely be delivering two million doses per day, as this week the Biden administration committed to delivering 13.5 million doses to states at a minimum. On top of this, the Biden administration is delivering two million doses directly to pharmacies and one million doses to community health centers. If all those doses are used next week, we’ll exceed 2 million doses administered per day. Further, the CDC expects 200 million total doses distributed to states by the end of March, which implies a daily rate of 3.5 million doses per day in March.

Input 2: One Vaccine Dose Begins Providing Protection

The initial data from the UK and Israel shows that even one vaccine dose provides a high level of protection. This does not mean individuals shouldn’t get a second booster shot. They clearly should. Very likely, though, the first dose provides immediate impacts on infections, hospitalizations, deaths and even transmission. More and more data is coming out which supports this conclusion.

Input 3: The Covid-19 vaccines work. Extremely well.

It seems crazy to have to repeat this, but it should be noted that the vaccines work really, really well. You probably saw this tweet from Brown epidemiologist Dr. Ashish Jha, but it’s worth repeating:

Moreover, studies conducted in the United Kingdom and Israel on vaccine roll outs confirm the efficacy of these vaccines. In addition to preventing symptomatic illness, the vaccines also drastically reduce hospitalization and death. Finally, initial data also suggests that in addition to preventing death, it looks likely that the vaccines prevent transmission to other individuals.

Input 4: Covid-19 is most severe for older individuals, who the US is prioritizing in the vaccine rollout.

While the coronavirus can and does kill all ages, one of the more clear trends is that the virus disproportionately kills older individuals. Here are a few looks at this:

IMAGE 9 - Deaths by Age

IMAGE 10 - Deaths by Age v02

Fortunately, in January, after vaccinating healthcare workers, the CDC changed guidance to focus on high-risk (meaning older) populations. This means the impact of effective vaccines will be even greater than general distribution because it will decrease hospitalizations and deaths of those most likely to be hospitalized and die. 

If we can achieve high levels of vaccinations in those groups (say 80% of a given age population vaccinated) then the results will be dramatic. Vaccinating 80% of individuals over 75 will lower total deaths by 48%, individuals over 65 will lower deaths by 65% and vaccinating individuals over the age of 50 will lower total deaths by 76%. (And if we can achieve 100% vaccination in the most at risk groups? We’d lower total potential deaths 95%!)

Input 5: The current lockdowns are having a dramatic effect.

Meanwhile, as the Covid Tracking Project’s data show, the number of cases are plummeting in America. The declines are dramatic:

IMAGE 11 - Covid Tracking

Whatever the cause—lock downs, seasonality, growing natural immunity, some vaccine prevention—this is very good news for the United States. It means cases are trending down, right when vaccine distribution is ramping up.

Input 6: Case rates by population

To understand the impact on cases, beyond deaths, it is important to know how cases are distributed in the population. Unlike deaths, cases are fairly well aligned with the population. Meaning, folks will test positive for Covid-19 roughly correlated with their percent of the population. 

IMAGE 12 - Case and death and population

This is important because it could mean that as vaccines are rolling out to younger populations, the case rate could flatten or even increase, but deaths will not. There is already evidence that case loads are dropping in certain age groups in the UK, Israel and US, but as you can see that may not impact the overall case load significantly.

My Working Hypothesis

Add these inputs together, and this is my working hypothesis:

If current lockdowns can drive down cases…
Which will drive down hospitalizations/deaths…
And if the United States can vaccinate the highest-risk groups…
And if we can continue to increase the vaccines distributed per day/week…
Since vaccines prevent deaths and hospitalization…

Then once we get the current hospitalization rate down, it will stay down until the next flu season (next November or December). If hospitalizations are down, deaths will stay down as well.

This will allow states to reopen, including theaters.

The Model

So that’s the working theory. Let’s turn this into a model.

Step 1: Vaccine Distribution

To start, I sketched out some vaccine distribution scenarios. To start, I drew a linear model in vaccine distribution to see how it was growing. Then, I made a simpler second model based on potential vaccine supply. This weekly model uses big round numbers, but is more aggressive than the linear model and based on CDC guidance about distributing 200 million doses by the end of March. Lastly, I made a very conservative model based on plateauing vaccine supply at 2 million per week starting in February. 

First, here’s the linear model to show the logic:

IMAGE 13 - Scatter Plot

Here’s the table of doses by week:

IMAGE 14 - Table per Day

And a chart of that…

IMAGE 14 - Chart

That’s a lot of number, so here is a summary by month of the “reasonable” model.

IMAGE 15 - Chart

The monthly model confirms that vaccine makers could indeed hit the aggressive targets. I built the conservative model based on misreading the director of the CDC Rochelle Walensky estimating that we’ll have 200 million doses by the end of March. (I thought she had said 200 million by the end of April.) At a rate of 200 million doses by the end of March, the CDC is basically forecasting that we’ll be vaccinating 3.5 million people per day at some point in March. That’s my “reasonable” but aggressive model.

Step 2: The number of vaccinated individuals

So that’s the first step. We can reasonably forecast how many doses we’ll have distributed by future dates. Yet, some folks will insist on distinguishing between shots and those who’ve been fully-vaccinated. I went further and also calculated those who will be two weeks from their first dose, since data shows this too provides quite a bit of protection. Specifically, we need to know these dates by the start of May, our presumed launch of Black Widow, 

Here is my rough counts for those, and to put them into context, the percent of the population over the age of 18 who would be covered:

IMAGE 16 - Numebr Vaccinated

IMAGE 16 - Chart

(Why over 18? Because the vaccine is only approved for those 16 and over, and Covid-19 has very little impact on ages younger than that. Plus, frankly, it is about 250 million individuals which is a nice round number.)

In the worst case scenario, by the time Black Widow premieres, only 30% of the population will be fully-vaccinated. 

However, in the best case scenario, if you include natural immunity, one dose vaccines and count folks who are two weeks from their first shot, then up to 69% of the population will have immunity to Covid-19. So we’ll be on the verge of herd immunity by the end of April!

Let’s be clear on these assumptions:

– Some portion of Americans have already had and recovered from Covid-19 acquiring natural immunity to it. This floor is at least 27 million confirmed cases. (About 10% of the population.) The high end is unknown, but Harvard epidemiologist Michael Mina’s forecast is 40%. Let’s split the difference and call it 20%.

– I used the date five weeks prior since Moderna is three weeks, then two weeks to be fully effective. Yes, Pfizer is four weeks, but again this difference is minimal overall.

Step 3: The number of deaths prevented. 

Again, the impact of Covid-19 is not equal across the population. It is very much tilted towards older individuals. Meaning, when they are vaccinated, the odds of dying from Covid-19 decease by 92% or more.

To figure out how to vaccinate the right folks, I simply took the CDC data and 2019 census information, and assumed an 80% vaccination rate:

IMAGE 17 - Deaths and Cumulative Distribution

In other words, to prevent 76% of deaths, we need to vaccinate 80% of everyone over the age of 50. The key assumption is that we can achieve 80% full vaccination by group. Most surveys put vaccine hesitancy at 70% of the population, but only about 10% is hard core obstinacy to any vaccine. So I took a number in the middle.

So how many people do we actually need to vaccinate to get to that 3/4th decrease in total deaths going forward? Well, here you go:

IMAGE 18 - % Needed to Vaccinate

Of course, not all doses will go to healthcare workers and individuals over the age of 65, especially as counties and states begin vaccinating more essential workers. (Like Los Angeles, who is moving onto food workers, teachers and remaining public safety officials after those older than 65.)

Given the vaccination rates above, we can see that it is very, very likely we’ll have fully vaccinated 40% of the population with either one or two doses, including most healthcare workers and folks over 65. A big portion will also likely be those 50 to 65. If we include people getting only 1 dose of Johnson & Johnson, then we’ll almost certainly have vaccinated all high risk groups.

My model forecasts that by May 7th, we’ll have lowered the ceiling of potential deaths by 76%. If this widespread vaccination results in decreased case loads and transmissions, the actual death rate could be much, much lower. This is essentially the “ceiling” of deaths.

Step 4: The Los Angeles Specific Model

This model, so far, has only addressed vaccinations and deaths. What about cases?

As I set up in the problem, the primary criteria to release Black Widow (and other big studio films) in theaters is whether or not the coastal cities are reopened. To answer this, ideally, I’d build a model forecasting cases in both of those cities. Given that I live in Los Angeles, I pulled the numbers there to see how far LA is from reopening. I’m assuming that Los Angeles and New York are roughly correlated with each other, and their outbreaks are also roughly correlated with national outbreaks.

(This assumption is both fine and could be horribly wrong. The “summer surge” mainly took place in “Sun belt” states, whereas the first surge took place in north Eastern states. However the last surge took place in every state simultaneously.)

I showed Los Angeles’ current performance on state re-openings, but it’s worth noting that most metrics are tied to case loads. If it goes down, ICU capacity, case positivity rate and the equity will trend downward as well. Here’s the current case trend line:

IMAGE 19 - Case Trend Line

Now the question is can we model how vaccine distribution could impact case levels going forward?

And no, I don’t think we can. 

I’ve done a lot of forecasting so far, but every number is from a fairly reliable source. The vaccine distribution plan is fairly well reported, and its growth is easy to forecast. (Again, look at that straight line!) The impact of the vaccinations is also fairly well known. Thus, we can confidently predict a coming drop in deaths that will stay low, if we vaccinate the most at risk groups.

Cases, though, are a different ball game. 

Just look at US trendiness in the past. If you started in the middle of October, and just extrapolated forward, you’d have missed the December spike. Or if you started in the middle of January, you’d have forecast cases to stay high. This actually happened with the CDC forecast. In the middle of January, their model of models forecast 1.5 million cases by the middle of February, with a floor over a million.

IMAGE 20 - Forecast

Instead? Cases are currently at 600K and falling.

IMAGE 21 - Cases Feb

I don’t blame the CDC. Modeling seasonality, societal behavior and mostly a brand new virus is incredibly tricky, and these epidemiological models have trouble with it. Again, no blame here from a fellow modeler. I’m just acknowledging the limitations of modeling.

Add it up, and I won’t forecast the case rate/total cases in Los Angeles by the beginning of May. There are too many unknown variables. Indeed, I think both the best case (cases stay very low, due to natural immunity and expanding vaccinations) and the worst case (case rates rebound after widespread reopening, potentially driven by more transmissible variants) are both possible by May. The only thing I am fairly confident (as steps one though three show) is that deaths will stay down. 

At best, what we can say is that as vaccinated rates go up, the peaks of the worst metrics will be limited. Think of it like this: the infection rate is the number of people an individual with symptomatic Covid transmits it too. If it is 2.5, that means one individual gives it to 2.5 individuals. If half of all the people someone meets are protected from Covid-19, then the max number would now be half of that, or 1.25. 

The Key Metrics Going Forward

Notably, I haven’t provided any probabilities thus far. If I apply probabilities to events two months out, my error bars would need to be very, very wide. Those probabilities wouldn’t be worth much more than guesses. 

Instead, I’m going to provide a scorecard of key metrics. The higher the scorecard, the more confident we can be that society will returning to normal. You can apply your own probabilities based on the numbers. These metrics will have three parts: vaccinations, deaths and case loads. 

Screen Shot 2021-02-18 at 5.02.44 PM

I plan to think on these metrics a bit before I do an update, hopefully next week. As I said, I won’t provide predictions, except for the vaccine roll out, but will color code which metrics are moving in the right directions.

As I said at the start, if you read all the way down here, as vaccine distribution picks up, the Covid-19 pandemic will end gradually, then all of a sudden.

In the meantime…

Please do whatever you can to prevent the spread of Covid-19, focusing on the best practices at preventing spread. 

1. Wear a mask. In fact two, or the highest quality you can afford.
2. Avoiding indoor gatherings until you are fully-vaccinated.
3. By all means, get vaccinated as soon as you are able to.
4. Help older relatives get their vaccinations by whatever means necessary.
5. Spread positive news about the efficacy and safety of vaccines and that they will enable our society to reopen. Vaccine hesitancy is driven by vaccine skepticism, a large amount of which is coming from the media. This includes skepticism about how vaccines prevent hospitalizations and death. Do your part to spread the good news and not Covid-19 hysteria.

Sources

I linked to most data I used to put this together. However, a few websites provide regular updates. I recommend…

Our Wold in Data
Covid Act Now
The Covid Tracking Project
The Los Angeles Time
Bloomberg Vaccine Tracker
Nate Silver’s Twitter feed, who close followers will recognize a few tweets from.

Most Important Story of the Week – 25 Sep 20: We’re Heading for the (Almost) Worst Case Scenario For Theaters

Last week was a big one for me as I tore through a lot of Mulan data to produce my soon-to-be-biggest article of all time, “1.2 million Folks Bought Mulan on Disney+”. (It looks like it will dethrone the previous champion, “Netflix is a Broadcast Channel”.)

It’s been four weeks since I checked in on the health of theaters, let’s make that the most important story of the week.

Most Important Story of the Week – We’re Heading for the (Almost) Worst Case Scenario For Theaters

I try to think about things probabilistically. As Nate Silver would recommend. The world has lots of randomness, so events and different outcomes have different probabilities.

When I made my forecast of Coronavirus’s impact on theaters for a consulting client, I had a median case of theaters reopening in August. And it almost happened, but for a summer surge in cases. The worst case was that theaters would stay closed through 2020. We’re not quite to that worst case, but we’re close.

We’re partially opened in America as 70% of theaters are allowed to be open, but the studios are pulling their tent poles until the biggest markets reopen. Given that the US still accounts for 30-50% of a film’s total box office, America’s uncertain situation is scaring off all the big studio releases.

Which is a shame, because the rest of the world is doing much better. They’ve opened and after a few weeks most customers returned. Yet the US uncertainty (combined with global piracy, which is another shame) has held all the big studios from releasing their true tentpoles. The news of the last few weeks is that studios waited to see what Tenet would do, and found it wanting. 

Thus, Wonder Woman: 1984 moved to the end of the year (Christmas Day) and Black Widow moved to 2021. Though not all of Disney’s slate, as Soul is still holding onto Thanksgiving. And Universal moved up a few kids films to try their new PVOD strategy.

So I wouldn’t say we’re in the darkest timeline for theaters, but we’re closing in on it. November and December will have a lot of weight to pull to bring studios and theaters through.

Other Contender for Story of the Week – The Tik Tok Deal and Global Entertainment

Every newsletter I follow has been tracking the ins-and-outs of this story. But I waited. Would it be Microsoft? Or Oracle? Or Walmart? Or none of the above?Twists, turns and…we’ve ended up in almost the exact same place?

It’s like that quote from the Red Queen: you can run all day and end up in the exact same place. (Hat tip to the The Lost World novel for writing about that and logging it in my head from (is this right?) 25 years ago.)

All that has really changed is that Byte Dance has a new 20% owner of Tik Tok (Oracle) and it gets to keep operating in the United States. But it keeps its algorithm and presumably spy software in China.

Does this have implications from global entertainment? Assuredly, though let’s not go too far.

Clearly, China and America are headed for a new “Cold War” or “Bipolar” economic landscape. I’m not breaking news telling you that. President Trump has also escalated the situation with his proposed bans on TikTok and WeChat.

Not that this economic nationalism is unprecedented. China has banned US apps and companies for years. The biggest challenge for both EU and US companies and their nation-state champions is that there really is an unfairness in the global business situation. Netflix, Amazon, Google and others can’t operate in China due to protection laws. Yet, the EU and USA (and most OECD nations) pride themselves on allowing free and open markets. Which lets in Chinese champions.

This makes a seemingly unfair balance of power. (Though I could defend why China does it, and that reason is because US firms have definitely exploited smaller economies over the years. China has now largely avoided that fate. But this isn’t a politics website, I’m merely trying to explain why China is doing what they do.)

Where do we go from here? It’s unclear. Both presidential candidates seem concerned about China, so presumably restrictive measures could remain in place, with a Biden administration administering them a little more fairly/objectively. Long term, this could really hurt global business strategies with prominent Chinese ties.

That’s Disney, primarily, but really all the studios. One of the changes to my film model I’ve been thinking of making is to update the box office to: US, China and Rest of World. Since China is so protective, it keeps an outsized amount of profits in that country. (Only 25 cents of every box office dollar goes back to the studios. And even those can be hard to pull out.) If companies need to increasingly make “non-China included” strategic plans, that has lower global upside everywhere.

Entertainment Strategy Guy Update – The MLB-Turner Extend Their Deal with a 7% Year-Over-Year Increase

What? 7%? You saw the 65% jump in value reported in the press, didn’t you? 

Well, the key is context and the Entertainment Strategy Guy is nothing but context. When I see big splashy deals, my first question is the time period. In this case, a seven year extension. Then I take the two numbers and plot the CAGR. I put the average deal value in the middle of the deal (since leagues like to have revenue increase on a flat rate). Then I make my chart:

Screen Shot 2020-09-25 at 9.30.17 AM

As for the strategy, the next deal that shows a decrease in prices will be the first deal to show a decrease. Sports continue to be the source of programming keeping the linear channels alive, and the remaining linear players are paying a lot for them. And the bubble with 5-10% average increases in price each year has stayed on track.

Data of the Week – A Few Data Points on Subscribers (Peacock, NY Times The Daily and Shudder)

If “apples-to-apples” is the theme of the week, then I need to put the context right up front for these numbers. One of the numbers is “US only”. One is “US plus”. And two are global. Do not confuse them, since it really does change the denominator. (330 million versus 7 billion!)

First, Peacock, while explaining the increasing centralization of all NBC-Universal decisions under Peacock, Comcast let slip to the Wall Street Journal that they have gotten up to “15 million sign-ups” from the 10 million they announced in their July earnings report.

Next, Shudder, which is available in the United States, UK and some other territories, has reached the 1 million subscriber milestone.

Third, the New York Times “The Daily” podcast now reaches 4 million folks. Which is a huge number, but again don’t assume they’re all Americans.

The Athletic has also purportedly reached 1 million subscribers. While this is technically a global number, odds are it is driven much more by US customers. The caveat is that The Athletic has so aggressively discounted its business model that we don’t know what a subscriber’s actual ARPU is.

Other Contenders for Most Important Story

Disneyland (and Friends in California) Wants to Reopen

If you’ve been reading the EntStrategyGuy for any length of time, you’ll know that theme parks are a big part of Disney’s revenue stream. (Even more so than toys, which often get the credit.)

Hence, each week and month that Covid-19 keeps theme parks shuttered in California is a significant hit do Disney’s top and bottom lines. This week Disneyland, Knott’s Berry Farm and others publicly called on Governor Gavin Newsom to allow them to reopen. They noted that the reopenings in Florida and Europe haven’t seen accompanying surges in transmission, which surprised me. (Disneyland Hong Kong, however, was shut after reopening for having an outbreak.) 

Notably, some theme park-adjacent businesses are opening, like the Los Angeles Zoo. So curious to see when Newsom changes on this. 

DC Comics/DC Universe Staff Sees Layoffs

This is a few weeks old, but it is important enough news that I didn’t want to skip it. Warner Media is cutting staff at DC. If comic books can be the “R&D” department of a movie studio–and look at Disney, they are–then why would you cut the staff?

Of course, layoffs are complicated. Sometimes organizations really do have bloat. Sometimes they really do have redundant capabilities. But this seems like some creative executives were swept up in this part of the Warner Media reorganization. Meaning long term the cost cutting now could hurt the creative output of the future. Comic books will never be the cash cow that turns around AT&T’s fortunes, but having a strong DC could help grow HBO Max.

M&A Updates – Ion Networks is Acquired by EW Scripps

Some more merger action! This time Ion Networks is getting acquired by EW Scripps. I’ve long appreciated Ion Network’s business model. Ion Networks realized that if they owned a broadcast channel, cable and satellite providers must carry their programming. They bought up broadcast stations, and then ran cheap reruns. It’s been surprisingly successful for them:

image-1-estimates

Lots of News with No News – The Emmys!!!

I put less emphasis on The Emmy’s than anyone else. From a business perspective, I just don’t think they tell us much about what customers want or how businesses are doing. (They mostly tell you who spends the most on Emmy campaigning, as brutal as that sounds.)

The story was Schitt’s Creek, which went from nothing to something with a run on Netflix. Using the “Netflix is a Broadcast Channel” thinking, though, this makes sense. It’s like a show went from a small cable channel to running on NBC. Since it was good, naturally it had a boom in viewers.

1.2 Million Folks Bought Mulan in the US During It’s Opening Weeknd: The (Not) Definitive Analysis of Disney’s Mulan Experiment

How many folks bought Mulan?

That’s the buzziest question in the streaming wars right now.

Since we don’t know, we’re left to pick at the analytics tea leaves. Fortunately, as each day passes, we’ve got more tea leaves to pick through.

(Partly, the question is relevant because it gets to the buzziest question, “Who’s winning, Tenet or Mulan?”. I’ll answer that on Wednesday.)

Far from throwing my hands up, I’ve started to realize these tea leaves are signal not noise. So if/until Disney tells us otherwise, I’ve done my best to compile all the Mulan on Disney+ data we have. Consider this a “meta-analysis” on Mulan. First, I’ll summarize each data source and what it tells us, next I’ll try to compare this to Trolls: World Tour, then I’ll compare all the data sources, and finally I’ll make my estimates for Mulan’s performance.

(I covered some of these data points in a column and Tweet thread two weeks ago. Today, I’m updating all that data and tossing in my estimates at the end. Also, if you’re new to the EntStrategyGuy, my newsletter goes out every two weeks with links to my writings and the favorite things I read over the last two weeks.)

To start, though…

Bottom Line, Up Front

Don’t want to read the entire thing? Fine, here are the talking points you can deliver confidently without reading the whole article.

— The story about Mulan’s performance is remarkably consistent, if you ensure you are comparing “Apples to Apples”.
— Right now, I’m fairly confident at estimating that its opening weekend Mulan was purchased about 1.2 million times. (Other estimates range between 1 to 1.5 million, giving us a fairly tight range.)
— That implies that it made about $36 million on its opening in total revenue.
— Based on its rapid decay, the Trolls: World Tour comp and the fact that it will only be in PVOD for 8 weeks, I estimate Mulan will generate about $90 million in US sales over its lifetime. (Based on the estiamtes, this could be as smalls $75 million and as high as $135 million.)

What We Know: 6 Different Sources Tell a Remarkably Similar Story about Mulan

Disney took a big swing by releasing Mulan straight to Disney+ (and only Disney+) for $30 a pop. That left multiple analytics firms—each vying to get new customers to buy its data, a important point about self-interest to note—to fill in the gap. Reelgood said one thing about the popularity; Samba TV said something else; Antenna said something else and then Yahoo took 7 Park’s data in a completely different direction.

The better analogy than tea leaves is actually the old parable about the elephant and the five blind men. Each grabs a different part of the elephant, so feels something different. That applies to our measurement firms. One is measuring viewership; another purchases; another app downloads. Toss in different time periods and sources, and it seems bewildering.

But if you put the whole picture together, it’s not that confusing. After 7 Park put out a great thread clarifying their data this weekend, I’m fairly convinced each source is telling the roughly same picture.

Source 1: Google Trends

This source is so easy anyone can use it. So be careful. Google tracks search traffic data which has been shown to be a very good proxy for interest. Here’s the time period going back to when Covid-19 started featuring top streaming films:

G Trends - PVOD Comparison

What’s the simple takeaway? Interest in Hamilton far outpaced anything else in the straight-to-streaming space. (See my article in Decider for details.) This, for me, is the context of Mulan.

However, since we’re triangulating on Disney+, it’s also worth looking at a “Disney+ only” look:

G Trends - PVOD Disney Only v02

Mulan was big, but paled in comparison to Hamilton.

Source 2: Antenna

Antenna tracks subscription behavior across a range of services such as iTunes, Amazon Fire TV, Roku, Google Play and others. Last week, they released their analysis of Mulan’s opening weekend in this great chart:

Antenna Longer Time Period

This is the most skeptical look I have of Mulan’s huge driver in interest from Disney. Yes, it helped boost sign-ups for Disney+, but less than any other major theatrical driver of the last few months. Also note how this aligns/correlates with Google Trend data, but not perfectly. Black is King did better than Mulan, according to Antenna, but Google Trends has lower interest. (Google Trends has more interest in Artemis Fowl than Black is King.) 

There is a similar story with Frozen 2 driving more sign-ups than Onward according to Antenna, and Google Trends telling an opposite story. (This explanation is fairly simple: Frozen 2 launched right as lockdowns started, so that’s more the story of lockdowns driving parents to subscriber, not interest in Frozen 2.)

Antenna’s data goes further on Mulan. They also used their data to breakdown Mulan purchases by sign-up time period. 

Antenna Subscriber Percentages

Antenna also  released purchases by sign-up time period. So I took those numbers, and combined them with the above chart to give us this estimate of the average % of subscribers who dropped $30 on Mulan:

Antenna Subscriber Purchas Rate

Save that number, we’ll get back to it. But it’s not the only look Antenna provided. They gave some data to LightShed Partners (and then tweeted it), which compares daily sales of various PVOD releases with “purchases by day”:

Antenna Daily Purchases

This is great because we can use a few numbers to compare Trolls: World Tour sales to Mulan. Hang on to this number too. And pay attention to those steep decay curves.

Source 3: 7 Park

7Park is another data analytics firm, though they don’t clarify where and how their data is collected. However, they have been releasing streaming data for a while now.

7 Park entered the data fray this week with a buzzy article on Yahoo, that slightly oversold the analysis. 7 Park measured, through the first 12 days of September (which covers through Saturday of Mulan’s second weekend), the percentage of users who watched Mulan among all Disney+ users during the time period measured. That italicized portion is key. Which is why Mulan could get 29% of streams during its opening weekend, but then a much smaller number when you look at Q3 to date:

7 Park Long Time Period

How does that 10.3% compare to Antenna and Google Trends? Favorably. As 7 Park pointed out in their thread, the demand ratio from Hamilton to Mulan matches Google Trend very well. As for their data versus Antenna, they measure different things. One compares to subscriber base while the other compares to active users. Assuming active users are between 50-75% of the total subscriber based, then the numbers tell a similar story.

Source 4: Samba TV

Samba TV measures viewership on connected TVs specifically. Samba TV also ran an analysis on Mulan viewership, from the opening weekend, coming up with the number that 1.12 million folks purchased Mulan during the opening weekend. It’s unclear if this is connected TV’s only or if they extrapolated out to all customers. Does this match the other numbers? Yes, as we’ll see.

Source 5: Sensor Tower

Sensor Tower measures application downloads. For the streaming wars, they track how often folks are installing streaming application. (Hedgeye analyst extraordinaire Andrew Freedman uses their data to forecast Netflix and Disney+ subscribers fairly well.) According to Sensor Tower, Mulan drove a week-over-week increase in downloads of 68%, which compares to 79% for Hamilton during its opening weekend. This is a bit lower than the Antenna, 7Park or Google Trends data. Sensor Tower only tracks mobile viewing, which may explain the difference.

Source 5: Reelgood

The biggest outlier is Reelgood’s data. Reelgood is an application that helps folks find and curate their streaming offerings. Reelgood uses their data (they claim 2 million users) to then estimate demand for various titles. Here’s their chart with notably the streams as a percentage of top 20 streams.

Reelgood Top 20 copy

This genuinely surprised me since customers had to purchase Mulan, which should have decreased its viewership. Instead, in a follow up, Reelgood said that Mulan actually surpassed Hamilton, which only had 9.68% of streams. This is the only source that implies that demand for Mulan was higher than Hamilton. So it’s our biggest outlier.

Missing Sources

Just to note, of the major sources I track, Nielsen and Parrot Analytics both haven’t entered the Mulan fray. The reason is that both focus on TV series with their publicly available data. (Though Nielsen does have feature film viewership data.)

Trolls Would Tour Comparison

That’s the data, let’s make the comparisons. First, here is the leaked details or estimates of Trolls: World Tour’s performance.

Screen Shot 2020-09-21 at 12.59.08 PM

Unlike Disney (so far), Comcast was much more willing to leak positive data about their Trolls: World Tour experiment. A few things to note, these estimates aren’t quite as steep as Antenna’s data, but match real world churn/decay better. We’ve seen this with other streaming titles where the opening weekend is about half the viewership of the first month or so of a title. And then with trolls the opening month is about half the viewership of the title lifetime to date.

This point may be interesting, but its definitely possible that about as many folks watched Trolls: World Tour after it dropped to $6 to rent then watched at $20. This chart from The-Numbers shows how popular Trolls: World Tour was even 3 months after PVOD:

DEG At home

WIth these numbers, we can compare purchases between Trolls: World Tour and Mulan using Antenna’s data. I did this by measuring the various peaks in the above Antenna chart with purchases by day.  Which made this chart:

Antenna Demand as Trolls

Since they’re decaying at roughly the same rate, we can use this to estimate Mulan sales. In other words, I estimate that Mulan had about 61% of the sales of Trolls: World Tour on PVOD. The caveat is that Mulan is available in less places than Trolls or Scoob, meaning demand could have been as high, but without additional TVOD channels it reached less customers. But that still results in lower sales/demand.

Comparing all the Sources

Wow. So if you’re still with me, here’s my summary of everything we know. Here are the estimates I derived for purchases for the first weekend, where the data allowed me to make that estimate:

Summary Comparison v01

Let me explain this. Given that Antenna and 7 Park are percentages of subscribers or active users, the 15-35 million are potential ranges of Disney subscribers/users. Then I picked the number that is my current “best guess” for each. In other words, I think Disney+ has about 30 million US subscribers, and about 20 million active users in a given quarter. If you disagree, pick another input. For Samba TV, I just used their estimates. For Trolls: World Tour I multiplied the estimated 2.25 million Trolls opening weekend customers (40 million divided by $20) by 61%, the rough proportion from the chart above.

All these sources say about 1.1-1.4 million folks watched on the opening weekend. Splitting the difference, and picking the number I like best, gives me an estimate of 1.2 million.

From there, we can estimate lifetime sales. I’m using my estimate that opening weekend will generate 50% of the first month’s sales. Both Antenna and Google Trends back this up. For example, it has already seen a second weekend drop in demand of about 75% in Google Trends. Also, given this decay, I think its second month will only see about 20% more sales:

Summary Estimate Lifetime

Using best case scenarios (33% viewing in the second month, 1.5 million opening weekend), I get to $135 million lifetime PVOD. Using worst case, I get to $75 million.

Phew. I’m wiped out. There are tons more issues to unpack, especially how this compares to Tenet. But I’ll do that next time.