After Sirius XM Holdings (NASDAQ:SIRI) posted a self-pay monthly churn rate of just 1.6%, a number of those commenting on Seeking Alpha wanted to know what I would have to say about the number. Well, first of all, as CEO Jim Meyer said on the conference call [my emphasis added]:
Our churn rate and new car and used car conversion were very strong. Self-pay churn was just 1.6%, one of the best results in the company's history. Vehicle related and non-pay deactivations were up slightly year-over-year, but this was offset by a huge decline in voluntary churn. Put quite simply, fewer people called us to cancel. When you have a subscriber base as big as ours, every tenth of a point change in monthly self-pay churn equals the difference of about 70,000 subscribers in a quarter or 280,000 net subscribers in a year.
So this improvement in churn of nearly 20 basis points drove a lot of upside in the quarter. We believe our retention performance will continue to be extremely solid in the back half of 2015 although we don't see 1.6% churn as the new normal.
In response to the question, "...could you maybe just walk us through historically what have been the moving parts in terms of what drives churn and maybe just give us some sense as to where you think you can come out for the rest of the year?" CFO David Frear later added some detail:
...it's three primary components: you have non-pay, you have vehicle turnover and you have other voluntary churn. And where we've seen a real departure from expectations is in the other voluntary, and there're just fewer people that are going out and effectively choosing that free option that Jim talked about than we had expected. It's a good beat, we're glad to have it, but very difficult to predict how people are going to act going forward. So we've had a long-term expectation of churn in sort of the 1.8% to 2% range. We are beating that this year. We hope to continue to beat it, but we're cautious on the outlook.
My answer to the question about what I would have to say is, "It was a great number. It's hard to argue when the company states 'it was one of the best results in the company's history.'" However - and you all knew that there would be a however - the question for investors is whether this was a one quarter fluke or whether the drivers that worked favorably for Sirius in Q2 can be maintained? And more importantly, what else do we know about the churn statistic and why it has been improving.
Before going further it may be useful to discuss how Sirius defines churn. In one sense, the churn statistic is a measure of subscriber loyalty. It tells us how many subscribers that had decided to pay for the service (what Sirius terms "self-pay subscribers") cease paying for that service. It is a fairly simple mathematical formula that takes the number of these self-pay cancellations that occurred during the quarter (or the year for the annual result) and divides it by the average daily number of self-pay subscribers during that period, and divides that result by the number of months in the period.
While the percentage - whether 1.6% or the historical norm of 1.8% to 2% that Frear mentioned - may seem small, it is a very large number. And, as Meyer noted, "every tenth of a point change in monthly self-pay churn equals the difference of about 70,000 subscribers in a quarter or 280,000" for the year. That also means that even a low 1.6% translates to 1,120,000 for the quarter and a rate of 4.5 million for the year. Even if churn is only 1.8%, the low end of the historical range, the number of self-pay deactivations for the year would exceed 5 million.
Investors should note that Frear has also said about churn: "The biggest increase in churn once again came from vehicle-related turnover." In other words, one of the biggest drivers of churn has been caused by self-pay subscribers who bought a new car and cancelled the subscription on their former car. These subscribers had no need to maintain their subscription since the new vehicle being purchased usually came with a new free trial that typically lasted from three to twelve months. This free trial is essentially a joint offering from Sirius and the OEM.
The OEM receives a subsidy from Sirius to install the satellite radio in certain vehicles. Then, the OEM would also be eligible to receive a share of the future revenues generated from that radio if the owner becomes a self-pay subscriber. In some cases, in addition to the subsidy, certain OEMs pay a fee to Sirius in exchange for a longer trial and/or a larger share of the potential future revenue if the purchaser becomes a self-pay subscriber. These latter trials are called paid promotional subscribers. The paid promotional trials are regularly reported by Sirius while the unpaid trials are never reported in the 10K or 10Q. (The unpaid trials can be roughly derived from the various documents, certain reported metrics, new car sales and the conference calls.)
One move by Sirius that has lowered the reported churn in the past couple of years was a change in the calculation of churn. In the 2013 filings, the following sentence was added to the description of changes in churn [my emphasis added]: "The decrease was due to a higher mix of existing subscribers migrating to paid trials in new vehicles which are not included in average self-pay churn."
The key words are "migrating to paid trials," and I found this change particularly strange. It means that if a self-pay subscriber has a GMC pickup truck, and trades it in a for a new GMC pickup truck that comes with an unpaid trial, that subscriber becomes part of the churn statistic. If that trial eventually converts to a self-pay subscriber, it makes no difference, and that trade-in remains as a cancellation for the calculation of churn.
On the other hand, if that same self-pay subscriber trades that same GMC truck in for a new Ford or Dodge truck that comes with a paid promotional trial, the self-pay subscriber is not counted as a cancellation for the calculation of the churn statistic for the duration of the trial. If, at the end of the trial, the purchaser chooses not to convert that Ford or Dodge truck to a self-pay subscription, at that point the subscriber counts as a cancellation. And, if that purchaser converts to self-pay, then, according to Sirius, no cancellation has taken place.
Some might argue that any self-pay subscriber that trades in a vehicle should not be part of churn since the odds are that they are very likely to again convert to a self-pay subscriber at the end of the trial. Others might argue that since there are no guarantees that the self-pay subscriber would convert at the end of a trial, they should be counted as churn. At one time I was firmly in the latter camp for a number of reasons.
The most obvious point in favor of delaying the recognition of the cancellation is that if someone had previously found value in the programming and decided to pay for the service, they should be more likely than not to become a self-pay subscriber at the conclusion of the trial. That said, one of the metrics regularly reported by Sirius - new car conversion percentage - would seem to suggest otherwise. The likelihood of any new car trial converting from trial to self-pay subscriber hasn't been above 50% since before the merger. It was 47.5% in 2008, the year of the merger, and it has trended lower to the current level of 41% as the percentage of new cars equipped with satellite radios has increased.
Persuasive arguments can likely be made for both positions, but the bottom line is that any time a self-pay subscriber trades in their vehicle for a new one with a free trial, the revenue that Sirius receives from that self-pay subscriber will drop. Clearly, if the buyer purchases a vehicle from an OEM that has opted to go with unpaid trials, Sirius has lost three months of revenue. If the buyer goes with an OEM that chose the paid promotional trial option, the monthly revenue will also decline. We know this because of the following statement in the most recent 10Q [emphasis added]:
For the three months ended June 30, 2015 and 2014, ARPU was $12.42 and $12.36, respectively. For the six months ended June 30, 2015 and 2014, the ARPU was $12.34 and $12.27, respectively. The increase was driven primarily by the contribution of the U.S. Music Royalty Fee, partially offset by growth in subscription discounts offered through customer acquisition and retention programs, and a shift to longer-term promotional data service plans with lower rates.
Regardless of how one views a modified description of churn, a case can certainly be made that it is one reason the churn figure has declined. Is there another reasons that churn has declined? It would seem that there have been new customer service initiatives that have also helped reduce churn. Here are a few comments by Meyer from the Q3 2014 conference call [emphasis added]:
...we are utilizing advanced tools to better segment our marketing and customer service efforts. ...We've launched a new marketing database which, along with specially-priced programming packages, such as the one for Joel Osteen Radio, should help us attract and retain more customers.
We've identified our most tenured and loyal customers to ensure we give them a great experience when they call into our call centers for help. We are using our website to more prominently promote the varied pricing we offer, and we will continue to experiment and innovate when it comes to packages and pricing.
Our service continuity effort, which ensures SiriusXM customers continue to subscribe when they transition from one car to the next, has produced good early results to help ensure our customers never go without SiriusXM.
There may always be a trade-off between lower prices/larger discounts and losing customers in order to maximize revenue. And although we haven't heard much during the past few conference calls about whether the low cost package featuring Osteen has had the desired results, it is the idea of using "advanced tools" to increase the acquisition and retention of self-pay subscribers that may be another reason for the improved churn statistic. Better records and instant availability of data could allow the customer service personnel to more accurately tailor the minimal level of discounting required to retain customers, or whether a less expensive package will convince a potential self-pay customer to subscribe.
While these actions were taking place, the company also announced that a couple of significant OEMs - Ford and Chrysler - were offering longer trials. These longer trials should have several positive effects on both the total Sirius subscriber population growth and self-pay subscriber population growth. First, with longer paid promotional trials, anyone that had no intention of becoming a self-pay subscriber remains in the total subscriber population longer than a typical three paid promotional month trial. While this specific change doesn't directly affect churn, it does have the benefit of showing an increase in total subscriber population. Second, the longer trials give potential subscribers more time to sample the various programming and discover programming that could convince them to become a self-pay subscriber.
Third, if a self-pay subscriber that may have been ready to cancel trades in their vehicle for one with a 6-12 month paid promotional trial, that cancellation has been pushed out by an additional 3-9 months vs. a three month paid promotional trial. While that only defers the cancellation, the near term effect would be a reduction in the churn statistic.
It is difficult to ascertain how much of an effect the longer trials have on population growth. We can see that the paid promotional trials previously had peaked at 4,911,733 at the end of Q3 2013. This was immediately prior to one large OEM (widely believed to be GM, although never confirmed by either Sirius or GM) moving from paid promotional trials to unpaid promotional trials. By the end of the following quarter, the number of paid promotional trials had dropped to 4,477,493. Not only was that decline the first quarterly decline since Q3 of 2011, but it was also the largest drop since the merger of Sirius and XM in 2008.
The longer trials, increased vehicle penetration rates and rising new car sales since the end of Q3 2013 have all contributed to the rising paid promotional trial population. That number reached a new high of 4,999,000 at the end of Q2. On the recent conference call, Meyer said about penetration:
Our new vehicle penetration rate in the second quarter was 72%, up from 69% a year ago. With growing confidence, we now see long-term penetration rates settling near 75% of total production, up from our previous expectation of around 70%.
This suggests that the paid promotional trials are also likely to continue to increase, even after new vehicle sales begin to flatten.
Another factor to consider when trying to understand whether the low churn rate can be maintained is the length of vehicle ownership. New car buyers tend to hold their cars for six years while used car buyers tend to hold their cars for four years. As Sirius increases its marketing push towards used car buyers, the more rapid rate of vehicle turnover should put additional pressure on the "vehicle related turnover" portion of the churn statistic.
Summary
The Q2 churn statistic was excellent. It is also somewhat misleading when compared to historical levels due to the change in treatment of vehicle turnover for those self-pay subscribers moving to paid promotional trials.
Was it a one quarter fluke? Not necessarily. The company can adjust the discounts and promotions it offers to customers calling in to cancel, and utilization of "advanced tools" indicates the company is doing exactly that. We are clearly seeing a move towards average subscriptions carrying a lower net cost to the customer, and we see this in the ARPU figure.
Would it surprise you to find out that despite growth in advertising revenue, price increases, the elimination of multi-year discounts and an increase in the Music Royalty Fee, the ARPU has declined since the end of 2014? In Q4 ARPU was $12.49. In the most recently concluded quarter it was $12.42.
As Meyer stated, "...we don't see 1.6% churn as the new normal." I don't either.
Disclosure: I am/we are long SIRI.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.
Additional disclosure: In addition to my long positions, I have January 2016 $4 covered calls written against a portion of my long positions and January 2016 $4.50 covered calls written against a majority of my long positions in Sirius XM. I also continue to make frequent short term trades on large blocks of Sirius XM on a regular basis. I may close the current call positions, open new call positions or buy or sell large blocks of Sirius at any time. I hold no positions, nor do I have any plans open any positions, in any other company mentioned in this article.
