A brief postmortem on the wireless Internet utility

Nearly all of the city-scale, mainly WiFi-based wireless ISPs of the past three years are dead. Some, like Philadelphia, lumber on as zombie ventures. A few small town systems will continue to operate as long as the social and political consensus supports the subsidy required. And there are a couple of big city projects that haven’t burned through their initial operating capital yet.

But the rest are dead. The disease that killed them was cash flow hemorrhage, brought on by virulent churn.

Churn measures the percentage of total subscribers who cancel service and have to be replaced, in a given period of time. It also lets you calculate subscriber lifetime. In the mobile phone industry, a typical 2.5% monthly churn rate results in an average subscriber lifetime of about 40 months. Take an ARPU (monthly revenue per sub) of, say, $53, subtract $30 a month to provide service to a subscriber, and there’s sufficient cash flow over that period to pay off a subscriber acquisition cost (SAC) of perhaps $400 and still have something left over to improve the balance sheet, or grow the business, or even pay dividends. The business model works, although different companies implement it in a variety of ways with a wide range of results.

Let’s run the numbers for a municipal (or municipally bounded) wireless ISP (WISP). First of all, ARPU is limited by competition from DSL. At a typical monthly cost of $20 (forgetting for the moment promotional rates that are a few dollars less initially), and with performance metrics far superior to WiFi, DSL keeps WISP rates in the $15 range. (DSL’s superior performance is constantly debated, mostly by wireless equipment manufacturers and other vendors, but the hard fact is that it’s almost always faster, usually has significantly greater effective market coverage and, most importantly, always delivers a more consistent subscriber experience. Some will argue this assertion, but real world results trump arguments).

From that $15, subtract $12 for a small system (5,000 subs, say) and $8 for a large system (more like 50,000 subs) to pay for the cost of providing service. These cost figures are highly optimistic, it’s very possible to see a monthly operating cost of twice that range. But for the sake of discussion, let’s start with a rosy operating cost scenario.

In the best case, then, you have $7 per month left over to pay down SAC, put towards growing your subscriber base and improving your network. Let’s start with SAC. Take just the cost of providing and supporting the installation of CPE (consumer premise equipment, in other words the wireless bridge every subscriber needs to access the service reliably from homes and businesses), the direct cost of selling and activating a new customer, and a little indirect marketing cost, and you’re over $200. But let’s say $200 for SAC.

With $7 of operating cash flow, you’ll need 26 months just to break even on the average subscriber. If you can hang on to that subscriber for as long as mobile phone company does, you have a makeable business case. Unfortunately, WISPs don’t, and can’t, manage that essential trick. WISP churn rates are 2, 3, 4 times and more that of mobile phone companies. At a 7.5% monthly churn rate, which is not particularly high for a WISP, your subscriber lifetime is only 13 months, half what you need to pay the cost of getting and serving a sub. Even a 5% churn rate won’t get you there, and that’s wildly optimistic.

It’s easy to build a spreadsheet and plug in numbers that make it work – lower churn, lower operating cost, lower SAC, higher ARPU. And dozens of would-be municipal wireless ISP operators did just that. But the real world doesn’t pay attention to spreadsheet models and powerpoint presentations.

An $8 operating cost, $15 monthly rate and a $200 SAC are difficult to achieve, but possible. What’s not possible is a monthly churn rate much under 7% or so. And that’s what kills the model. The annual loss (or subsidy) is in the hundreds of thousands of dollars for a small system and in the millions for a large one.

The high churn rate is a direct and inevitable consequence of the competitive position of a WISP versus DSL and other wired technologies. If significantly superior service is available for $20 a month – and it is – people who rely on Internet access (which these days is nearly everyone who has it – it’s long past being a luxury) will pay the extra $5 if they can afford it. Unless they don’t want to sign up for a minimum term of a year or can’t pass the phone company’s credit check standards.

So as a competitive tactic (and often as a matter of public policy), WISPs either adopt easier credit standards and shorter terms or, more usually, all but eliminate those requirements. As a result, the core subscriber profile leans heavily towards households with lower disposable income and credit scores, and people who don’t plan to be in town very long. With this subscriber profile, even a well designed and operated WISP is going to have a high churn rate, well out of reach of a sustainable enterprise.

Mobile workers are touted as the sweet spot for municipal broadband, but they tend to give up on service and churn out too, for a couple of reasons. First, WiFi-based WISPs are not optimized for truly mobile service. When you’re driving around in a car, for example, handoffs from one access point to another are problematic. Second, service ends at the city limits, and it’s a rare private sector or government worker whose job is limited to a single city, except for city employees themselves. Mobile data service from cellular providers is a far better solution to both problems, and people with job-related needs quickly migrate to those platforms.

The grand municipal WISP ventures of the past three years died when the cash transfusions stopped. In some cases, they simply ran out of capital. In others, they had unworkable business models, often resulting from unrealistic demands by policy makers for free service and various other perks. When the subsidies, explicit or otherwise, stopped, the systems went dark. RIP.