After 24 hours with DirecTV now...

[SpaethCo]

I don't know much of anything about the pricing for millions of CDN endpoints like that, so if you could share your math like I did it would help the rest of us understand the numbers. Obviously stuff like that is subject to change since the technology marches on, but even as a snapshot of "the situation in early 2019" it would be valuable.

I'm also interested to see ncted's updated numbers on this.

I took a stab at the numbers a little over 2 years ago, I found financial costs from the last DirecTV financial statement that line up pretty well with the satellite distribution costs you outline above: DIRECTV NOW LAUNCH

 

[ncted]

19,000,000 subscribers x 2.5 people per household x 5 hours of viewing per day x 30 days = 7,125,000,000 hours of viewing per month

4 Mbps (bitrate DNow uses for HD streams) x 60 seconds x 60 minutes = 14,400 Mbph (Megabits per hour)

14,400 Mbph / 8 bits per Byte = 1800 MBph = 1.8 GBph (GigaBytes per hour)

1.8 GBph x 7,125,000,000 hours = 12,825,000,000 GB

12,825,000,000 GB x $0.0015/GB = $19,237,500/month

$19,237,500/month x 12 months = $230,850,000/yr

That cost per gig is less than half of the biggest contract I've ever seen as a large volume Akamai customer. Now clearly AT&T is not paying that much now, but that would be a very big bill if they ever did get a customer base that large. If you see any errors, let me know. Like, I said, it was a quick and dirty calculation.

 

[slice1900]

I'm also interested to see ncted's updated numbers on this.

I took a stab at the numbers a little over 2 years ago, I found financial costs from the last DirecTV financial statement that line up pretty well with the satellite distribution costs you outline above: DIRECTV NOW LAUNCH

Hey nice find, those broadcast operations expenses can be added to what I already calculated for as the build/launch costs for the satellite to get the all-in satellite delivery cost. Those operation expenses come out to $2 per customer per month, which includes MANY expenses that would also be incurred for streaming - because that mention "including costs of collecting signals for our local channel offerings" may account for half the cost by itself! Add it all up and you get $2.50 per customer per month. Now we have some exact figures to compare with.

One quibble with your streaming calculations. You assumed each customer watches the equivalent of one stream five hours a day. That's probably fine if someone lives alone, but a family probably watches more than that since they won't all watch the same thing. But the big one is this - how do they avoid streaming when people aren't watching???

When I watch my Tivo, I turn on my TV. My Tivo is already turned on and doing its thing. If I watched streaming content the same way, I'd be watching 168 hours a week, not 35 hours a week! Now sure, the streaming box can be smart about it and disable the streaming if the HDMI sink it is connected to (the TV) is powered off. What if you have your streaming box connected to an AVR, or HDMI switch? Those are ALWAYS ON, so the streaming box will not have any way to tell if you aren't watching. I guess if you haven't hit the remote in a few hours it can ask "are you still there?" and if you don't respond it can turn off, but that's still going to lead to hours of unnecessary streaming on EVERY streaming box every time it is used.

There are also other costs to account for, like the cost of storing all those cloud DVR recordings. If you know how many hours each customer gets, assume they keep it 75% full (which is probably right for limited storage) you could easily figure out how much that would cost on say Amazon in bulk. We can assume all video will be encoded in HEVC, since Directv's new IP offering will be using HEVC capable (C71KW) clients.

 

[SpaethCo]

One quibble with your streaming calculations. You assumed each customer watches the equivalent of one stream five hours a day. That's probably fine if someone lives alone, but a family probably watches more than that since they won't all watch the same thing. But the big one is this - how do they avoid streaming when people aren't watching???

For every calculation, I tried to error on the side that would be favorable to streaming costs. Viewership numbers, in particular, are hard -- which is why I stuck with the average. My gut feeling is that commercial accounts would skew the numbers way up, from small doctors offices that have just 1 feed playing in the lobby all day long, to sports bars that have 5+ unique feeds every game day. I stuck with the home viewer average and assumed 1 screen per subscriber to try to intentionally skew the costs to work out on the side of streaming. The 4mbps streaming rate is also lower than what we still see today: DTVN's top streaming rate is still around 8mpbs, Hulu & Vue are pushing 6.5mbps, Sling is coming in around 5.5mbps.

In the role of replacing existing broadcast TV: if you take the reasonable range of all variables, skew them all toward favoring a streaming solution, in the end it still ends up costing more money.

There are also other costs to account for, like the cost of storing all those cloud DVR recordings. If you know how many hours each customer gets, assume they keep it 75% full (which is probably right for limited storage) you could easily figure out how much that would cost on say Amazon in bulk. We can assume all video will be encoded in HEVC, since Directv's new IP offering will be using HEVC capable (C71KW) clients.

Interestingly enough, cloud DVR is possibly the best case of where streaming efficiency can shine. The key here is that for any given time block on a channel, only a single recording is made. If 10 people all record the same programming block, their "DVR" is just a logical pointer that authorizes them to play from that one stored copy. Once the last person "deletes" their recording, the last pointer is removed, the space is reclaimed.

 

[ncted]

One quibble with your streaming calculations. You assumed each customer watches the equivalent of one stream five hours a day. That's probably fine if someone lives alone, but a family probably watches more than that since they won't all watch the same thing. But the big one is this - how do they avoid streaming when people aren't watching???

When I watch my Tivo, I turn on my TV. My Tivo is already turned on and doing its thing. If I watched streaming content the same way, I'd be watching 168 hours a week, not 35 hours a week! Now sure, the streaming box can be smart about it and disable the streaming if the HDMI sink it is connected to (the TV) is powered off. What if you have your streaming box connected to an AVR, or HDMI switch? Those are ALWAYS ON, so the streaming box will not have any way to tell if you aren't watching. I guess if you haven't hit the remote in a few hours it can ask "are you still there?" and if you don't respond it can turn off, but that's still going to lead to hours of unnecessary streaming on EVERY streaming box every time it is used.

Not sure if you were commenting on my calculations or SpaethCo. I multiplied the 5 hours per day times the avg. household size in the US, 2.5 people. But, you do make a good point. The amount streamed and not watched might make it much higher.

 

[slice1900]

Interestingly enough, cloud DVR is possibly the best case of where streaming efficiency can shine. The key here is that for any given time block on a channel, only a single recording is made. If 10 people all record the same programming block, their "DVR" is just a logical pointer that authorizes them to play from that one stored copy. Once the last person "deletes" their recording, the last pointer is removed, the space is reclaimed.

It should be that way, but court cases regarding the legality of cloud DVR (which was challenged by content owners) require that each recording be unique - they can't use deduplication-like methods to save storage space. If a million people record Big Bang Theory, they need 1000000 * however many GB that recording requires. No one has since challenged it with "well what if my storage system does it for me automatically, do I have to turn off that functionality?" but realistically we already have cloud providers offering multiple GB of storage to people for "free" so it probably is cheaper to just do the way the court requires than to pay lawyers to try and change that.

 

[slice1900]

Not sure if you were commenting on my calculations or SpaethCo. I multiplied the 5 hours per day times the avg. household size in the US, 2.5 people. But, you do make a good point. The amount streamed and not watched might make it much higher.

Yes, his was probably the amount actually watched in a single person household, yours would be for a family. They'd have to find some way to avoid the unwatched streaming, that's a lot of wasted money they'd have to get a handle on (without making the experience too annoying, i.e. asking "are you still there?" every time you haven't touched the remote for an hour)

There's also a lot of people who leave the TV on all day as background noise. My mom does this all the time, and has a TV in her bedroom and one in the kitchen going most of the time she's both home and awake. She'd be really annoyed if it would cut out after a while just because they aren't in front of it to hit the remote, and wouldn't consider a TV product that did that a viable alternative to her cable at any price.

Then there are the people who pay for Dog TV, so they have something to leave on all day for their dog when they aren't even home...

 

[nelson61]

It does not take a rocket scientist to see where Directv lies in the AT&T priority list.
The margins stink.

Wireless and content are where the money will be going.

Error in my table. Total is 37.6 billion for communications group

 

[slice1900]

Directv's operating margin was almost double that when they were a separate company. Directv Now, their legacy landline stuff and Uverse are obviously dragging that unit down.

 

[nelson61]

Directv's operating margin was almost double that when they were a separate company. Directv Now, their legacy landline stuff and Uverse are obviously dragging that unit down.

At December 31, 2018, our total switched access lines were 10.0 million compared with 11.8 million at December 31, 2017. The number of U-verse voice connections (which use VoIP technology and therefore are not included in the access line total) decreased by 557,000 in the quarter, totaling 5.1 million at December 31, 2018, compared to 5.7 million at December 31, 2017

Looks to be a similar decrease to what Directv is reporting.

I would say Mobility is where the money is - and maybe content

 

[slice1900]

At December 31, 2018, our total switched access lines were 10.0 million compared with 11.8 million at December 31, 2017. The number of U-verse voice connections (which use VoIP technology and therefore are not included in the access line total) decreased by 557,000 in the quarter, totaling 5.1 million at December 31, 2018, compared to 5.7 million at December 31, 2017

Looks to be a similar decrease to what Directv is reporting.

I would say Mobility is where the money is - and maybe content

Landlines are about a decade ahead of satellite measured as where the lifecycle of each is in terms of market share erosion. Both technologies have the same problem in that as the userbase shrinks, you have to spread a fixed cost amongst fewer customers. DSL helps pay for it, but Centurylink is the only baby bell that invested in DSL (you can get up to 100/50 service in much of their footprint, and if they do g.DMT upgrades that will go much higher)

I'd have to think AT&T loses money outright on their landline and DSL service, if they had made investments in it and kept it more up to date the DSL side might be viable but they decided to abandon their copper and cherry pick affulent markets to install fiber.

In a decade this is where satellite will be, but luckily for AT&T they won't have regulations forcing them to keep satellite available so they will simply shut it down once the satellites begin running out of fuel.

 

[TheRatPatrol]

One thing I still don’t understand is how the streaming side of things cost less per month for the subscriber? Why can’t satellite and cable get the prices down to match the streaming prices? Is it all the extra fees we pay? Is $25 for whole home really necessary?

 

[slice1900]

One thing I still don’t understand is how the streaming side of things cost less per month for the subscriber? Why can’t satellite and cable get the prices down to match the streaming prices? Is it all the extra fees we pay? Is $25 for whole home really necessary?

Because Directv Now loses money. AT&T's CEO recently said something about raising the price of Directv Now by up to $20 and "thinning out the content" (i.e. removing a few of the less popular channels) to make it profitable.

https://www.multichannel.com/news/att-looks-to-reposition-directv-now

 

[camo]

FYI Directv Now was by far the worst streaming experience I've had plus expensive and tiered just like the satellite service which requires near top package for the most wanted channels like sports. More channels isn't always better. UI, DVR, 1080p @ 60fps buffer-free streaming with channels I want mainly sports are mostly covered by YTTV for $40 a month. If you must have more add Philo covers most of the missing channels like history, DIY, AHC, and a few others I watch on occasion for $16 additional.

 

[ncted]

One thing I still don’t understand is how the streaming side of things cost less per month for the subscriber? Why can’t satellite and cable get the prices down to match the streaming prices? Is it all the extra fees we pay? Is $25 for whole home really necessary?

https://www.multichannel.com/blog/youtube-tv-losing-money-but-is-there-path-profit

They are almost all losing money at this point. They will raise prices later.

 

[slice1900]

I read elsewhere that as of last year Youtube had content costs of $49 for that $40 package. Since networks schedule regular price increases in their contracts, no doubt that is over $50 now, so including all their other costs they probably need to be at $60 just to break even.