Telecosm - Ike Elliott

"Toes" Starts a Blog!

I have just discovered that Rob Powell started a blog last week, and for those that track Level 3 Communications, this is big news.  That's because Rob Powell is one of the best of the "anonymous analysts" that haunt the Level 3 Investor Village message board, where he goes by the alias "toes_that_twinkle." 

In my occasional trips to the Level 3 Investor Village message board, I have learned to scan for posts from "toes," because they are almost always thoughtful, never ranting, and most importantly, on the mark.  He really digs through the SEC reports, and as a result has very deep and informed perspective on Level 3.  If you track Level 3, you are going to want to add Rob's new blog to your RSS reader. 

To start, I highly recommend Rob's post on the "Virtuous Circle" that awaits Level 3 if the stock rises enough to start calling their convertible debt. 

Rob, welcome to my blogroll!  I will be an avid reader of your posts.

Churn Week Continues: Sprint's Struggles

Sprint Nextel announced their first quarter earnings yesterday, and it looks like their struggle to keep subscribers in their Nextel unit are getting worse.  Yesterday's post focused on Vonage's atrocious 3.3% monthly churn, and today, we'll look at Sprint's 2.45% churn in post-paid wireless subsribers.

Here is Sprint's pattern of post-paid subscriber losses since the beginning of 2007:

	1Q07	2Q07	3Q07	4Q07	1Q08
Net change in Sprint post-paid subscribers	(220,000)	16,000	(337,000)	(683,000)	(1,070,000)

Sprint did say that they expect an improvement in churn in the 2nd quarter, but til now, churn has actually worsened.  Last quarter, and the quarter before, post-paid churn was 2.3%.

It is enormously damaging to a recurring revenue service business to experience churn above 2% per month.

For example, did you know that if you assume a constant rate of gross subscriber additions, a service provider's churn rate can predict the maximum size of the service provider's subscriber base?  The higher the churn rate, the lower the maximum size of your customer base.  Tomorrow, we'll look at how this works.

Vonage: Mixed 1st Quarter Results

Vonage announced their first quarter earnings last Thursday, and while the company is still in a precarious position, it may be able to survive in the short term.  However, investors have to be wondering if "survival" is anything to be excited about, since the cost of adding net subscribers is skyrocketing.

The good:

• Revenue per month per line is up to $28.85, up from $28.18 last quarter.  This is quite a bit higher than I had forecast.
• Quarterly revenue was over $225M, about $4M higher than I had expected, mostly due to the larger increase in revenue per month per line.
• SG&A was lower, down to about 35% of revenue, down from about 37% of revenue last quarter.

The bad:

• Monthly customer churn is getting worse, now at 3.3%, up from 3.0% last quarter and up from 2.4% a year ago.
• Gross subscriber adds continue to decline.  The company added 281,329 subscriber lines, down from 283,907 last quarter and down from  332,493 a year ago. 
• Costs per subscriber were up, with direct cost of goods per sub per month jumping to $2.83 (up from $2.29) , and cost of telephony service per sub per month jumping to $7.26 (up from $7.11)
• As a result, gross margin is down to about 65%, down from 66.7% last quarter.

The ugly:

• Vonage still has not finalized their debt refinancing.  While it seems they are making progress in negotiations, until the refinancing is completed it represents a significant risk to the company.
• While Vonage added 281,329 subscribers, they lost 251,196 subscribers due to churn, netting only 30,133 subscriber additions.  The company's marketing cost per gross subscriber line addition was $216.47, for a total marketing cost of about $61M.
• Vonage prefers to report their marketing cost per gross subscriber addition.  It looks a lot worse, though, if you view it as marketing cost per net subscriber addition:
  

  Vonage	1Q07	4Q07	1Q08
  Marketing Cost Per Net Subscriber Addition	$     549	$   1,130	$    2,021

Bottom line, at $2,021 in marketing cost per net subscriber add, you need to keep a $28/month customer for six years, just to pay back the marketing cost of adding the subscriber.  Then you can start to repay the cost of goods involved in serving that customer for the past six years.  That's not a sustainable business.

This is why I have focused on Vonage's churn stat as the number to watch in past posts on Vonage.  Unfortunately, the company has yet to demonstrate its ability to reverse the negative trend in churn.  Until Vonage can do that, the future of their business is at risk.

New Music: Phil Stacey

My colleague Greg Menard (CTO of Unity Business Networks) turned me on to a new CD by Phil Stacey that is excellent.  Phil Stacey was one of the finalists on American Idol last year, so you may know about him already.  His self-titled release is in the country genre, yes, but not over-the-top country like the kind that keeps me from listening to country stations.  He's got a great voice that could sing other styles, and the arrangements are the kind of "barely country" arrangements that without the slide guitar you might not know were country. 

Anyway, believe it or not, I have never watched an episode of American Idol, and I usually avoid country music like I avoid house cleaning, but there are some really good songs on this CD.  I especially liked "If You Didn't Love Me", "You Are Mine", and "Identity".   It's hard for me to say it, but I guess I do like some country music.

A Day In The Studio

I took the day off from blogging yesterday, choosing instead to spend the entire day at Derryberry Audio, a recording studio in Westminster, Colorado, and it was a blast.  It was a day of great fun, and sometimes exhaustion, as a fantastic group of volunteer musicians came together to lay the foundation tracks for seven songs. 

It was my first time in the studio since my college band recorded a song for a James Madison University compilation album back in 1985.  Yesterday's recording session was much more intense.  Back in 1985 it only one song and it was lead guitar player John Romeo's music.  This time, it was seven of my own compositions.

A couple of years ago I started to write music, which has been a life-long dream but a dream that I had never really pursued...so I decided to change that, and got some rudimentary recording gear and started to put together some ideas.  Over the past two years I have written about 15 songs, and I worked with a bunch of local musicians from my church and the surrounding area to put together a low-budget plan to record seven of the songs, market them on iTunes, and donate all of the proceeds to charity.  We hope to have the songs finished and ready for sale by the end of the summer.

I am thrilled with the musicians who have donated their time and talents to the project so far...they are incredible.  I want to thank them personally here:

Electric Guitar: Mitchell Ashley
Acoustic and Electric Guitar: Gordon Clouse
Piano: Jeremiah Horner
Bass: Skye Perry
Drums: Rich Guess
Production and Engineering: Jeremiah Horner

Jeremiah did a fantastic job in the production role, it was a pleasure working with him.  Also, I want to especially thank our sponsor, Tim Draughon, for making this project possible, and also thank Mark Derryberry, owner of the studio where we recorded and enjoyed fantastic hospitality. 

I think everybody is having a great time with the project and we are looking forward to our next studio session.  I'll keep you posted on our progress!

Inflationary Trends in Retail Internet Access

Isn't competition supposed keep prices in check?  Apparently not in retail Internet access these days. All three of the big phone companies are making noises about price increases for their retail Internet access services:

• In February, the Chicago Tribune reported AT&T is raising prices $5 per month on their DSL-based Internet access services.
• Verizon announced on their late-April quarterly earnings call that they plan to raise Internet access prices on their FiOS based services in the 2nd quarter by similar amounts.  Verizon also plans to raise prices on FiOS-based video services.
• Qwest made a similar announcement on their quarterly earnings call last week, saying they will raise prices on DSL-based Internet access by as much as 11 percent.

Will the inflationary environment in retail Internet access creep into the wholesale Internet access market as well?  If so, that would be a huge turnaround in the wholesale Internet access market (where large companies, content providers, and carriers are charged by the megabit for access to an Internet backbone), which has experienced price declines in the 25% to 35% range for several years. 

Sprint and Nextel's Excellent Adventure

It's been a big week for Sprint Nextel, who for the last several years has been telecom's most notable slacker.  Now, Sprint has leveraged its assets and its position to cobble together an alliance of convenience with cable companies, Google, Intel, and yes, Clearwire, in the form of a joint venture that helps Sprint cover the cost of its big WiMax bet.

The week started with rumors that Deutsche Telekom was thinking about buying Sprint.  The next day, we heard that Sprint may sell its Nextel unit.  Today, the long-rumored cable-Google-Clearwire-Sprint-Intel alliance has come together.  It seems that Dan Hesse, Sprint Nextel's CEO, is making progress in Kansas...at least it's the kind of progress his shareholders might want to see.  Om Malik believes this is the start of a hack job in which WiMax is the first of three separate parcels for sale, with the others being Nextel, and then the remaining Sprint properties. 

The new joint venture helps fill a need for each of the participants:

• Sprint gets $3.2B in funding for its $5B WiMax rollout, and sheds much of the operational responsibility for the rollout, allowing it to focus on the atrocious customer churn in their wireless business.
• Comcast and Time Warner Cable get to partially fill a gaping hole in their product bundle: wireless phone and Internet access.
• Google gets a captive platform for its Android intitiative, which it needs in order to help close the gap with Microsoft and others in the crowded mobile operating system field.
• Intel gets a boost for the WiMax technology in which it has invested so heavily.
• Clearwire renews its WiMax relationship with Sprint, helping ensure the viability of both Clearwire and the WiMax technology on which it has placed its bets.

It's a deal that had to happen.  However, the history of joint ventures with widely-varied participants with competing interests is spotty, so success is far from certain.

Meanwhile, this development is just one of two big developments for mobile Internet access this week.  The other is from T-Mobile, as reported by Brough Turner on his Communications blog:

T-Mobile USA has launched their first 3G service using the spectrum they won in the 2006 AWS auctions.  For now, it's only New York City, but Reuters reports that T-Mobile plans to launch in 20 to 25 new markets by the end of the year and T-Mobile's stated intention is a full national HSPA network. In 2009, this will be our fourth national 3G network fully capable of multi-Mbps down and multi-hundreds Kbps up.

So, between AT&T, Verizon Wireless, Sprint Nextel, T-Mobile, and the new Clearwire WiMax joint venture, Brough predicts that we will have five competing wireless broadband Internet access networks by 2010.  I wouldn't be surprised to see this consolidate back to four players before 2010, with T-Mobile and Sprint joining forces.

Anyway, this looks like good news for consumers, and good news for companies selling bandwidth, like Level 3 Communications. Fatter wireless links, like those promised by WiMax, will need fatter fiber-optic networks to carry all of that traffic.

Interview: Envysion's Matt Steinfort and Rob Hagens

I recently had the pleasure of interviewing two leaders from one of the up-and-coming Software-as-a-Service (SaaS) companies here in the Boulder, Colorado area: Envysion.  I was lucky to get to work with both Matt Steinfort (Envysion's CEO) and Rob Hagens (CTO) while we worked at Level 3 Communications.  Since leaving Level 3, Matt and Rob and the Envysion team (including Darren Loher, another outstanding engineer from Level 3) have been doing some pretty innovative work around video services on the Internet.  But I've said enough...let's let Matt and Rob tell the story:

1. What does Envysion do?

Steinfort: Envysion provides Managed Video as a Service (MVaaS) to customers to enable them to better understand and manage their numerous remote locations.  Said in plainer English, we provide a video service that gives people access to live and recorded video in a very simple and scalable way so that they can understand what is going on in their locations, learn from it, and improve the way that they operate.

2. What is Envysion’s unique competitive advantage?

Steinfort: We’ve got a number of competitive advantages.  Among the most compelling of these are the ease of use and scalability of our service.  There are a lot of companies out there today that tout video as a great operational improvement tool, but their solutions were designed to work in a single site at a time and to be used by someone in IT or security.  Our solution is the only enterprise grade video service that works easily within the constraints of large customers’ network and IT capabilities and is easy enough to use that everyone from the CEO to a brand new marketing intern could use it.

3. How much bandwidth does a typical Envysion video stream use? 

Hagens: There is not a black and white answer to this because you have to factor both quality and resolution into the equation. Quality is a measure of how lossy the compression algorithm is run at. You could run 10 fps at low quality over 50Kbps, but the picture would look "pixilated" or "blocky".

Resolution is a measure of the total number of pixels that are used to render the image. Resolution in the video world is typically described as the number of pixels used horizontally and vertically. Today we support 320 by 240 resolution. DVD quality is typically 720 by 480. That is 4 times as many pixels as 320 by 240 which would drive much higher bandwidth.

A typical camera at 10 frames per second, 320 by 240 resolution would consume about 150Kbps with good quality. Envysion video is always recorded at a specific resolution, frame rate and quality. However, because of all the complexity described above, Envysion has built a dynamic bandwidth adjustment into its streaming technology. When viewing the video remotely, the system will dynamically adjust the bandwidth consumed in order to "fit" within the Internet access connection being used by the customer. This happens automatically without any customer intervention. This is a good example of how Envysion Video "just works better" across the Internet.

Steinfort: That's one of our key differentiators, our ability to dynamically throttle the amount of bandwidth that we use to stream video based on what is available.  It enables us to work on low speed connections such as ISDN and satellite, and ensures we don’t ever clog the networks that are also being used for mission critical applications (like credit cards processing).

(Please click the continuation link below for the rest of the interview)

Thomas Howe on the Future of Voice

Thomas Howe has written a very interesting post on "Keys to Understanding the Future of Voice."  Mr. Howe's company specializes in integrating voice applications into business processes, sometimes called "voice mashups."  He often has a forward-looking perspective on where VoIP and voice applications are heading.  Here are a few excerpts of his post:

"Voice applications have not significantly changed in the past few decades. Our attempts at creating successful and plentiful voice applications have repeatedly failed. We failed to create significant new applications with CTI; we have failed to create them with VoIP. We are poised for failure with IMS."

"If it ever was a serious debate, it is one no longer. In the merger of the computer network and the telephone network, the computer network is dominant and pervasive. In other words, the Internet is the dog and the phone is the tail."

Good stuff.  Click here to get the rest of the post!

LEC Line Loss Summary, 2007

For the past week, I've been digging into the massive losses in Local Exchange Carrier Switched Access lines, trying to quantify where all of these lines are going.  I looked at how cable voice services, fixed-mobile substitution, and VoIP are all contributing to LEC line losses.  Now it's time to summarize all of the data in a single table:

LEC Access Line Migration, 2007	EOY 2006	EOY 2007	Change in "Lines"	Estimated Contribution to LEC Line Losses
AT&T (landlines)	66,469,000	61,582,000	(4,887,000)	(4,887,000)
Verizon (landlines)	45,100,000	41,441,000	(3,659,000)	(3,659,000)
Qwest	12,299,465	11,500,000	(799,465)	(799,465)
Comcast Digital Voice	1,900,000	4,377,000	2,477,000	2,477,000
Time Warner Cable Voice	1,860,000	2,900,000	1,040,000	1,040,000
CableVision Voice	1,209,000	1,592,000	383,000	383,000
Charter Voice	445,800	959,300	513,500	513,500
Vonage	2,224,111	2,580,227	356,116	213,670
Skype Real Users in USA*	6,230,000	8,000,000	1,770,000	0
AT&T (wireless)	60,962,000	70,067,000	9,105,000	1,379,853
Verizon Wireless	59,029,650	65,700,000	6,670,350	1,010,884
Sprint Nextel	53,100,000	53,800,000	700,000	106,084
T-Mobile USA	25,041,000	28,685,000	3,644,000	552,244
MetroPCS Communications	2,940,986	3,962,786	1,021,800	154,853
Leap Wireless	2,226,000	2,860,000	634,000	96,082
Estimated Other Cable Line Migration				367,792
Estimated Other Wireless Line Migration				93,455
Estimated Other VoIP Line Migration				28,036
Estimated Business Line Migration				467,273
Estimated Fax and Modem Line Disconnection				461,740
			Total	-

*Note: Skype "Real Users" figures are from Hudson Barton's Borderless Communicator blog.  He has an interesting related post on the LEC line loss theme today.

To create this table I relied on data from annual reports of public companies, data from the FCC Annual Report and Analysis of Competitive Market Conditions With Respect to Commercial Mobile Services, comments from Telecosm readers, and a little guesswork on the margins. 

Estimating fixed-mobile substitution is the hard part of this table.  The FCC report estimates that between the 2nd half of 2005 and the 2nd half of 2006, the percentage of U.S. households that only had wireless phone service grew about 4% to something between 11% and 12% of all 110M U.S. households.  I'm guessing that by the end of 2007 the percentage of U.S. households that used wireless as their primary phone line had grown to about 15.3%.  However, some of the incremental growth in wireless-only households did not switch from LEC services, instead switching from cable and other VoIP services, so I tried to account for that in the table.

The "Estimated Other Cable Line Migration" is estimated at 4% of the total line losses at the big three LECs, to account for Cox Communications, who is privately held, and to account for the host of smaller cable companies.

I'm guessing that VoIP accounted for only about 2.6% of the overall LEC line losses in 2007, due to the fact that some of Vonage's customers are overseas, and to account for the fact that VoIP providers struggled or went out of business in 2007 (SunRocket).

The pie chart below summarizes the data for the major causes of LEC Line Losses:

Pandora: A Power Tool For New Music Discovery

This blog is primarily a technology blog, and I like to focus on IP multimedia applications, but now and then, mostly on weekends, I get the urge to blog about other stuff that is important to me, like MUSIC.  I am a fool for good music, and I'm really not constrained by genre.  Mostly, I just like discovering new artists and get a thrill from enjoying their music for a few years before anybody else has discovered them.  That's why when my friend and colleague Bob Paulsen, President of Unity Business Networks, showed me Pandora, I was all over it.  And the cool thing is that it is an IP multimedia application, so I don't have to feel like I am going off-topic when I blog about it.

Put simply, Pandora is a tool for customizing your own Internet radio stations.  You can create a new radio station just by entering the name of an artist you like, or the name of a song you like, and Pandora plays song after song after song that are similar to the song or artist that you used to seed your radio station.  For example, I created a Dave Matthews Band station the other day and listened to it all day long, and Pandora played a lot of live versions of DMB songs that I knew, and also played a lot of U2, Pink Floyd, Jack Johnson, O.A.R., Pearl Jam, Ben Harper, even some Stealers Wheel.  As I write this post I am listening to my DMB radio station, and a song from their Live at Radio City Music Hall album, Eh Hee, came on...its one of the few DMB albums I don't own and I had never heard the song before.  Pretty awesome song.

Being smart about selecting which songs to play is part of the secret that makes Pandora special.  The algorithm relies on data from the Music Genome Project, which has captured hundreds of attributes about just about every song out there.  The algorithm starts out selecting songs based on the seed song or artist name, and then you "tune" the algorithm by telling it which songs you like and which ones you didn't.  You can also add new seed songs or artists to existing stations. 

Other things that make Pandora special: 

• It is really easy to use
• It is free! (there are options for paid subscriptions that remove the advertising).

Pandora does have some other limitations.  There is a limit to the number of different radio stations you can create, but it is a really high limit, so I haven't found it to be a problem.  Also, you can't force Pandora to play only the artist that you selected...you have to let it play what it wants to play, just like a radio station.  However, you can skip songs, which is better than a radio station...there is just a limit of 6 skips per station per hour. 

I am finding Pandora to be a fabulous tool for discovering new music.  I now have a Pinback station and a Le Loup station, for two of the bands I have recently discovered.

I encourage you to try it out.

How Much Is Pure-Play VoIP Contributing to LEC Line Losses?

This week, I've been investigating the phone companies' loss of over nine million traditional phone lines in 2007, trying to figure out where all of those lines went.  So far, we have seen that about half of them went to cable voice services, and a big chunk of the rest went to people replacing their home landline phone service with wireless phone service.  But there are a few other categories that accounted for at least some of the line losses.

One of these categories is pure-play VoIP services, such as Vonage, Skype, and all of their much smaller brethren.  Over the course of 2007, here is how Vonage and Skype grew their subscriber bases:

Consumer VoIP	EOY 1006	EOY 2007	"Lines" Gained
Vonage	2,224,111	2,580,227	356,116
Skype "Real Users"*	21,677,314	27,518,590	5,841,276
Total	23,901,425	30,098,817	6,197,392

*NOTE: the Skype "Real User" statistics is from Hudson Barton's Borderless Communicator blog, and is the best measure I have seen of the portion of Skype subscribers that really use the Skype service.

Of course, many of the users of these "over-the-top" VoIP services use the service to augment a wireline or wireless phone service, rather than using the VoIP service as their primary home phone line.   I believe that is especially the case with most Skype users.  Vonage users, on the other hand, often use their service to replace their primary home phone line. 

With that in mind, I think that pure play consumer VoIP companies caused only a small number of the 9 million line losses experienced by incumbents.  I'd assign maybe 6% of the total line thefts to the consumer VoIP bandits. 

There is one more potential cause for LEC line losses: business line losses to facilities-based competitive local exchange carriers.  Most CLECs have traditionally re-sold the incumbent's wireline plant in order to reach customers, so those lines aren't really "lost" since the incumbent LEC is still selling the line on the wholesale side.  However, some CLECs have connected their own fiber optic plants to larger office buildings in major metropolitan areas, and can sell services to the businesses that operate out of those buildings.  Examples of these kinds of CLECs include Level 3 Communications and Zayo Bandwidth.  I don't have good data on how much of the LEC line losses could be attributed to competitive bandwidth providers, but I suspect that this percentage is also low, less than 10%. 

Monday, I'll look at all of the suspected causes of line losses in one summary post, as was suggested by one Telecosm reader.