Vonage: It's All About Customer NPV

VOIP pioneer Vonage filed for its IPO last week and the financials contained in its S1 have set off a lot of discussion.  Om Malik, who has followed Vonage very closely over the years, thinks that the S1 contains more red flags than Boston's Big Dig, while Michael Parekh is having a serious case of late 1990's deja vu thanks to Vonage's enormous losses.

As for me, I think the S1 makes it plain as day that Vonage is simply following the same game plan that the major online trading firms followed in the late 1990s.  As I outlined in a mid-2004 post, the key metric for Vonage is not current period Net Income, but Net Present Value (NPV) per customer.  As long as the incremental customer NPV is sufficiently high, Vonage should not only continue running losses, but it may actually make sense to dramatically increase near term losses in order to create more long term value.  Indeed, one look at the financials tells you that this is exactly what Vonage did in 2005.

While this is great in theory, up until now there has been no hard data publicly available to see if Vonage was in fact succeeding with this strategy.  While the recently filed S1 doesn't have all of the information needed to make a 100% accurate assessment of Vonage's incremental customer NPV, it has more than enough data to make a decent estimate and so with that in mind I dusted off some of my old online trading customer NPV models (I covered the online trading industry when I was a Wall Street analyst) and took a crack at it.

The 5 Keys To Estimating NPV
Customer NPV models generally have 5 key pieces of data.  Below I have identified each piece of data and described what, if any data is available from Vonage.

1. Net acquisition costs per customer.   Acquisition costs are probably the most important part of an NPV model.  In the S1, Vonage claims that its marketing cost per customer in Q3 05 was $209. This is a bit understated though because this figure does not include the costs associated with deploying customer premise equipment (ATAs) which appear as "Direct Cost of Goods Sold" in the financials.  These costs in turn are somewhat offset by "Customer Equipment and Shipping" revenue.  Including these additional figures results in a line acquisition cost of $233.24, which is further reduced to $203.25 by the activation fee of $29.99.  The actual cost is probably a bit higher because Vonage accounts for cash incentives and rebates as a contra against revenue, but these costs don't appear to be highly significant right now so I am going to go with a $203.25 cost in my model.
2. Average customer life.  The key determinant of average customer life is what's commonly known as "churn".  The higher the churn, the lower the average customer life, the less aggregate cash flow a given customer will generate.  In the S1, Vonage inexplicably decided to calculate churn based on customers as opposed to lines, even though it doesn't provide any statistics on how many customers it has each quarter.  That said, it's possible to calculate the line churn easily and it has been 2.1%/month for the past two quarters.  At this churn rate, the average customer life is roughly 4.2 years.  According to the S1, Vonage is assuming a 5 year average customer life as that is how long they are choosing to amortize their activation fees.  I am going to be generous and give them their 5 year customer life under the assumption their accountants signed off on it so it and it’s not that far from the range they have been running lately.
3. Average yearly customer revenues.  With fixed rate monthly subscriptions, Vonage's revenue/line tends to be relatively stable although it has been declining over time thanks to price cuts.  In Q3 2005, the average monthly telephony revenue per line was $24.84 or about $298/year.  This includes a small portion of the activation fees, so I am going to use a $292/year figure to account for the activation fees after doing some rough estimates.
4. Operating margins before customer acquisition costs.  In an NPV analysis you have to separate out customer acquisition costs from on-going operating costs in order to determine what the "out year" margins are on customer cash flows.  In Vonage's case, when you strip out all of the marketing costs as well as all the equipment costs and revenues, you get an operating margin of 5.5% in Q3 05.  This is down sharply from Q1 of 05 when operating margin before marketing costs peaked at 22.2%.  It's not completely clear why margins have declined so much (they should be increasing) but it's possible that one-time costs associated with the IPO and some e911 requirements led to a spike in expenses.  In light of this, I am going to give them the benefit of the doubt and use a 20% operating margin in Year 1 although it will be very interesting to see what direction these margins go in Q4.
5. Discount Rates.  Discount rates are a fairly subjective matter when it comes to NPV analyses.  At a minimum, discount rates should at least be set to the expected rate of inflation so that you have a "real dollar" cash flow.  However, most organizations use a discount rate in excess of inflation, usually one that is either their Weighted Average Cost of Capital (WACC) or some targeted premium to WACC (to create economic value and/or sustain their premium market valuations).   For high risk private companies a 10%+ discount rate is typical.  In my online trading models I used to set the discount rate equal to the market consensus long term growth rate estimate (which was usually in the 20-25% range) under the conservative (and somewhat convoluted) theory that this meant people were expecting the company to grow earnings 25% a year and thus any NPV analysis should reflect those expectations.  I don’t know what Vonage’s market value will be, nor do I know it’s WACC, but I do know that they have used a lot of venture capital to fund their build out and the long term IRR of early stage venture capital is around 18%.  Given this I am going to use a discount rate of 18% even though this is admittedly mixing apples and oranges when it comes to finance theory.

One last assumption I have to make is what will happen to annual revenues and margins in the “out years”.  Vonage has already cut prices several times and it stands to reason that they may do so again in the future.  In addition, pre-marketing operating margins are likely to change in the future, although margin pressure from price cuts may be more than offset by the benefits of increased operating scale.  For example, in the late 1990’s many of the online trading firms ran pre-marketing operating margins in the 20-30% range, but now those margins are now in the 40-50% range thanks to scale effects.  To account for these dynamics, I am going to assume a 5% year decline in revenue and a 5% year increase in operating margins (which leads to a 24.3% pre-marketing operating margin 5 years out).  I have no idea what these will actually be but I think that these are reasonable guesses.

And The Answer Is…
So where do all these assumptions get us?  As the table below illustrates, these assumptions, plus the hard data from the S1, generate an incremental customer NPV of $(21.39) for Vonage in Q3 of 2005.

Obviously, a negative NPV, even a small one like $21, is not a good thing and may indicate that Vonage is actually destroying value (at least relative to return expectations) despite its rapid growth.

This conclusion is tempered somewhat by the following facts though:

1. This is admittedly a very simplistic customer NPV model based on incomplete data and tenuous assumptions.  However, anyone thinking about buying Vonage’s stock will basically have to go through the same exercise with the same data, so they will face the same challenges and may come to similar conclusions.
2. Vonage’s customer acquisition costs may be inflated due to their very heavy ad spending in 2005.  In the online trading industry, many companies had similar spikes in spending and saw greatly reduced customer acquisition costs after they ramped down that spending thanks to the residual brand-value created by the advertising spikes.   Given this, it is possible that Vonage could lower customer acquisition costs in the near future.
3. Vonage’s pre-marketing operating margin may increase much faster than modeled.  There’s a strong argument that Vonage will be able increase its operating margins significantly, not only due to normal operating scale, but due to lower bandwidth costs and lower termination costs as Vonage processes more “on us” calls that stay within it’s own network.

All that said though, Vonage is clearly cutting things pretty close.  For comparisons sake, the last model I did for E*Trade in June 1999 showed that they had a $299 acquisition cost but a NPV of +$140 at a 26% discount rate which clearly provides a lot more room for error than Vonage currently has.

A Simple Matter of Price
What this all boils down to is really more of a valuation question than anything else.  At a 5% discount rate (which amazingly is higher than the yield on 30yr treasuries right now), Vonage is actually generating +$48 in value for each customer it added in Q3 2005, but at an 18% discount rate they are losing $21.  This means that investors should theoretically be willing to pay a price that equates to a discount rate of somewhere between 5% and 18%.   Just what price they settle on will probably depend on how persuasive Vonage’s management team is on their road show.

Admittedly, this post fails to address some of the larger (and arguably much more important) strategic issues facing Vonage such as competition from the MSOs, regulatory risk, and “free” VOIP from the Skype’s of the world, but from a pure historical #s perspective you can’t really say that Vonage isn’t creating value, you can just argue that they aren’t currently creating enough value to justify the return expectations embedded in the capital they are using.  That is potentially bad news for Vonage’s VCs but not necessarily bad news for public investors if they price the shares correctly.  It will be interesting to see just what that price is.