More on why usage-billing makes more sense than freemium.

A fairly undeniable fact regarding Kamikaze pilots is that they didn’t only remove themselves from the gene pool, they also took out a lot of people who were simply playing by the ‘normal’ rules of war. A similar fate awaits not only those companies hell-bent on distributing their wares for free, but also for many of their more conventional competitors who will get caught by the shrapnel.

Like all good sects, the free web has its own illustrious leader, Chris Anderson, author of the The Long Tail and Free. Anderson has refined his notion of free over time, so that it now no longer focuses primarily on the ad-funded model, preferring instead the trendier freemium concept. His philosophy can be summed up by his phrase “give away the head and charge for the tail”.

The core of the free argument is based on the notion of the distribution cost of digital content being close enough to zero to round down. Hence it doesn’t cost much to give away your product to anyone who wants it, and in doing so you create an opportunity to offer a premium version to your grateful users.

In a review of Anderson’s new book, New Yorker writer Malcom Gladwell points out that even if the distribution cost is very, very small, when you multiply it by a number that is very, very large, for example the number of videos streamed by YouTube, you still get a figure with lots of digits before the decimal point. In other words, if your free offering is really popular, the zero-distribution-cost part of the freemium equation stops being true.

Taking this idea one step further, let’s look at an industry that really does have zero (rather than negligible) distribution costs, and was already doing free when the internet was just a twinkle in TBL’s eye.

Nothing New Under the Sun
In eighties England there were only four TV channels. It is an uncomfortable fact that just about any Englishman in his forties can tell you what triple salco or puissance mean. There were two ad-funded networks, ITV and Channel 4 – the other two channels were provided by the ad-free BBC.

Things must have been pretty sweet for the two ad-funded networks, and even the arrival of two fledgling satellite broadcasters (BSB and Sky) towards the end of the decade did little to upset the status-quo. Most commentators argued that people were very unlikely to pay for a service that could be had for free – and they were almost right. The cost of buying rights to premium content combined with the expenditure required to operate their respective services pushed the two struggling entrants into a shotgun marriage, and British Sky Broadcasting (now known simply as ‘Sky’) was formed.

But then a strange and important thing happened. It appeared the Great British Public was not universally happy with its free programming. It transpired that many people were actually willing to pay for content. Some wanted to see live sporting events, others wanted to see films before they were available to rent, and others just liked the idea of surfing dozens and dozens of channels. Whatever the underlying reasons, the UK rapidly transformed itself into a nation of satellite viewers – as evinced by receiving dishes now nestling like mushrooms on every other house in the UK.

In a move that must surely give the freesters food for thought, an entire nation went from free to paid – and moreover in a industry with a distribution model that represents the best case scenario for free.

Nothing is Identical Under the Sun, Either
Okay, TV in eighties England is not the same as the internet now. Viewers had practically no choice, and the arrival of the satellite operators actually reduced the quality of the free offering, because the satellite operators bought-up the rights to the content people most wanted to watch. Barriers to entry are also very different: it costs a lot less to create a website than a TV network. This last point is important, because when there are only a handful of national operators, there is no real need to do any marketing – everybody knows who you are. One of the key aims of the freemium model is to get your service noticed by letting your free offering do the marketing for you.

Nevertheless the principle still holds: we have seen that under certain circumstances people are willing to pay for content - even in the face of a free alternative. And more to the point: people are willing to pay for the head (the most popular content), even if the (admittedly short) tail can be had for free. This is the exact antithesis of Anderson’s argument.

Any physicist or mathematician can tell you that while you can’t prove a theory with a single example, you can certainly disprove one. So does this imply the free web, and freemium in particular, is based on the incorrect premise that the easier content can be distributed, the less likely people will pay for it? Unfortunately not. Business is not science, and what holds in one market doesn’t necessary hold in another. But we have shown that the pro-free arguments, which at first glance can seem so alluring and self-validating, are in fact nothing of the sort.

That said, the disturbing reality is that, based on current alternatives, the free web – at least in the guise of the freemium model – makes a lot of sense. Or, more succinctly, in the land of blind web revenue models, one-eyed freemium is king.

The real problem, therefore, is not so much with freemium (chronically flawed as it is) - but with the lack of any viable alternatives.

The Paid Web
There are two basic ways of billing for stuff on the web: subscriptions and micro-payments. We have already argued strongly against subscriptions on this blog, and will therefore proffer only a brief summary here: subscriptions, whether on the web or anywhere else, are a form of lock-in. Users know this and so view them with suspicion. People don’t take out subscriptions to restaurants or shops, for example, because they don’t want to limit their future choices. Most importantly, they resent paying for something they either no longer use, or use less than expected.

Subscriptions can work when people are willing to exchange a degree of product choice for some sort of benefit (most often a discount). This can occur with magazines or newspapers, when the customer realizes they would have probably bought the publication anyway. But in the fast-moving world of the web, where people don’t know what’s coming next and so don’t want to limit their choices, and where there’s so much free stuff available anyway, subscriptions (even in the context of freemium conversion) don’t work.

As to micro-payments - the practice of charging a tiny sum for each incremental bit of service or content - Clay Shirky killed that idea long ago. He quite rightly pointed out that regardless of the size for the requested payment, the ‘mental transaction costs’ associated with an endless stream of tiny decisions rapidly add up to one big and annoying cost in terms of time (our most valuable resource) and inconvenience. Micro-payments are even worse than subscriptions.

A Thought Experiment
Einstein loved thought experiments (and look where they got him) – so let’s try one here. In a way it’s ironic that we’ve just held up newspapers and magazines as an example of how subscriptions can work in the real world, when these very same two categories are getting absolutely hammered on the web. So let’s balance the books and use newspapers as the basis for our though experiment.

Let us imagine a magical world in which a copy of a newspaper can be printed and distributed for free to every household in the world. Let us also imagine that each issue has a special microchip embedded inside. Many people will pick up their free paper every morning and throw it straight in the trash. Others may skim the front and back pages over breakfast and then throw it in the trash. But a certain percentage (dependent mainly on how good the newspaper is) will actually open it up and read inside. This is were the microchip comes in – it’s clever enough to tell when the newspaper is being read properly.

After an acceptable period of product engagement, it beeps and says, “Good morning, I can see you’re really getting into the paper today. How about we charge you ¢20 every time you read more than half a dozen articles?”. The reader has a little think and asks, “What if one day I don’t read the paper, or only read four articles?”. “Then we won’t charge you”, chirps the microchip. Our reader likes what he’s hearing, but he thinks a bit longer and then adds “Clay Shirky says I’ll rapidly get fed up deciding every morning whether I want to be charged for reading the paper”. The microchip was expecting this one: “Clay’s right. We only ask you this once whether you are willing to pay for reading the paper thoroughly. If you agree, you’ll just get a monthly bill, based on how much you’ve read”. “What if I find a better newspaper and stop reading this one?” asks the reader. “Then you’ll never hear from us again” says the chip.

Our reader puts the paper down and begins to reason: what’s my downside? I can’t buy a good newspaper for ¢20, and if I don’t read it, I don’t pay. No annoying interruptions. No sections I don’t have access to. I only pay for what I consume, and if I stop reading it altogether, it just disappears. Finally he exclaims, “Microchip dude, you gotcha self a deal!”.

Fantasy, eh? Not entirely. What a lot of people don’t seem to grasp, or at least leverage, is the fact that the web is bi-directional. This model can’t work for TV or satellite because they are broadcast mediums - the information only flows one way. But it can and will work on the net.

Not Quite Cloud Billing
We call this model micro-billing. One decision, one bill, very small payments.

In many ways it’s similar to the way billing works in the cloud, in that you only pay for what you use. But there are key differences. In the cloud, billing generally starts the moment you begin using the service, but micro billing uses a usage threshold, below which you don’t pay. It’s very important that people are able to try products without paying – this is the best part of the freemium model. It makes sense for content owners to leverage the web’s negligible distribution costs to get their products in front of consumers. The product functions as its own marketing device.

Micro-billing also has a maximum usage fee. In our thought experiment the threshold payment and the maximum payment were the same (¢20), but in some situations it makes sense for them to be different. The key point is that a potential customer should always be able to work out the most the service will cost. People hate getting into open-ended commitments.

Finally, a word on price. How can a quality newspaper only cost ¢20? Well, for the same reason you can find electronic gadgets in cereal packets: volume. With over one billion people connected to the web, companies with content that can be digitized have to learn to offset the loss of their semi-monopolistic status in the physical world, with access to this much larger, albeit more competitive, online market. In a key difference with the freemium model, all active users pay. Some may pay more than others – but they all pay. When you have a very large customer base and such low variable costs, you can charge much less for your product and still maintain a decent return on capital.

In the freemium model, on average 97% of people use a service without paying for it. This means the price for the paying users is much higher than it could/should be because 3% of the user base is effectively subsidizing the remainder. Micro-billing significantly lowers the cost of a service, increasing the likelihood people will regard it as good value, and hence sign-up for being billed

Sayonara Baby
Free to use (as opposed to try) is a dangerous trend, not least because it also punishes those current and potential competitors who can see it doesn’t make sense. Tim Cohn put it best when he asked “is free the new barrier to entry”.

Fortunately, the example of the TV market in England shows that all is not lost – it is possible to introduce a paid model into a market which was previously all free. But the model has to be right. Currently, there is no compelling paid model for the web, and so freemium wins by default.

We believe usage-based micro-billing is the paid model the web has been waiting for.

Photo Credit

Or why usage billing makes more sense

In the freemium model, a free version of a product is used as a marketing tool to entice people to sign-up for an extended paid version. The term was first coined on Fred Wilson’s blog back in 2006.

Freemium has a great deal of support on the net, especially after advertising rates plummeted, and the practice of buying-out companies based purely on their reach has gone the way of the codpiece. These days investors and potential company buyers want to hear the sweet jangle of coins dropping into the coin box. Well-known services like Flickr and Skype, as well as many others, use the freemium model.

Close, but no cigar
In an excellent article, Free! Why $0.00 Is the Future of Business, Chris Anderson, author of The Long Tail and soon-to-be-released Free, explains why this “freaky land of the free” makes perfect sense – at least to him. He begins by pointing out that the concept of giving away something in the hope of charging for something else is anything but new, describing Gillette’s groundbreaking idea of giving away razors and then charging for blades. He also cites the example of the Brazilian Banda Calypso, who intentionally make their music and sleeve art available online, so that street vendors can create and sell cheap CDs, legally keeping all the money. The buzz created around the CDs permits Banda Calypso to sell-out its live concerts.

Banda Calypso’s business model is great. It leverages the negligible distribution cost of digital content to get its promotional material (the music) in the hands of an army of marketeers, who are happy take on the production costs, and plaster the band’s image all over the place in order to promote both their own and the band’s interests. It’s beautiful - but it’s not freemium.

As Anderson points out, both are examples of a general form of cross-subsidy, were one type of product is given away so as to charge for another. Neither Gillette’s original strategy, nor Banda Calypso’s, meet the common definition of freemium as it exists on the web, which dictates that both the free and premium versions are variations of the same product. To be genuinely freemium, Gillette would have to offer a two blade head for free, and charge for a four-bladed one; Calypso would have to give away most of their concert seats, and only charge for the best ones.

The pitch
The logic behind freemium is that for many websites the marginal cost of providing the service is very low. And as hardware and cloud services continue to get cheaper, it gets lower by the day. Take advantage of this to create a large base of non-paying users, then put your faith in the law of averages and watch some of them convert to the paid service.

Even if the number of paying users is only a tiny fraction of the freeloaders, that’s still more users than you would have got by just offering a straight subscription service, and their acquisition cost (ie the cost of running the free service) is significantly lower than a large-scale advertising and referral program.

Very nice. So much so that Sean O’Malley was driven to ask whether freemium can save web 2.0? - without, it’s worth adding, suggesting that it could.

Freemium is not a revenue model - it’s a marketing technique. The revenue model is still subscription based. The core idea is that giving away a product version is a more cost-effective way of driving subscriptions than traditional methods. And it probably is. But let’s take a closer look at the opportunity cost of letting anybody and everybody use your service for free.

Under the hood
There are hundreds of web pages discussing freemium. Based on nothing more sophisticated than a simple frequency count within many of these pages, an estimate of the average conversion rate for freemium comes in at around 3%. It’s true that some verticals have higher rates (5-10%) but it’s equally true that others have much lower ones (0.5 – 0.05%).

So let’s assume that a hypothetical and profitable website is using freemium, and charges $10/month for its paid service. This means the company is earning an average 30¢ from all active users. Let’s also assume the core product is a spreadsheet application, since it’s a common metaphor.

Okay. So you’re a happy and contented user of the free version. You think it’s really cool that you can use such a fantastic program at no cost. Then one day you need to save your spreadsheet in CSV format. No problem, just click on the ’save as CSV’ button. You guessed it: “The save as CSV option is only available to premium users, click here to upgrade”. This kind of message is an extraordinarily effective way of making even the most mild-mannered person shout obscenities. So what do you do? Pay $10 to save your file as a CSV? Perhaps you do. Perhaps…

The website owner knows these messages are annoying and can frequently be counterproductive. There are number of ways of minimizing them. The first is to make the CSV option available in the free version too. But you don’t have to be Warren Buffet to realize this approach doesn’t scale. If all the useful functionality is free, why would anyone upgrade to the premium version? (A whole micro industry has cropped-up offering advice on how to optimize the free/paid balance). A more cunning technique is to not show the CSV button in the free version. If you can’t click on it, you can’t get pissed off when it doesn’t work. But now there’s a new problem: if you can’t see what your missing, why would you ever upgrade?

Enter the sneak preview. At regular intervals, free users are given access to some premium features. They can now see what they’re missing, and might even be persuaded into hitting the mythical upgrade button when the functionality disappears at the end of the promotion. Very clever. But let’s dig a little deeper…

The free version of any freemium service, must, by definition, be useful – otherwise even the freeloaders wouldn’t use it. You can’t give them ‘copy’ and hide ‘paste’ behind the pay wall. So what goes into the premium version? Functionality which is occasionally very handy, but not fundamental to core usage. Like a save as CSV option. And what are free users expected to think when they are given a time-limited opportunity to use these nice-but-non-core-features? Yay! For a limited time I can save all my spreadsheets as CSVs! Game changer! Not. Like your nail clippers, you only miss this kind of functionality when you need but can’t find it.

All this assumes the website owner is in control of what’s in the different versions of the product. But if your spreadsheet competitor offers the CSV option in their free version then you probably will too. If you don’t and it doesn’t hurt you, then perhaps people aren’t saving that many spreadsheets in CSV format, and so why would they pay for that functionality anyway?

What we have here is the risk of a glorious race to the bottom. Your competitors will offer more and more free functionality so as to take market share. All this is great for your users, but not so good for your shareholders.

Setting aside the fact that all the time dedicated to deciding what should be free and what shouldn’t - often backed-up by complex A-B testing – could be better employed by improving the core product. Or that your competitors are ruining the market by dumping evermore free stuff on grateful users, there is one more reason why freemium is not the savior it’s frequently held up to be - and it’s a humdinger.

The Deal-Breaker
Defenders of the freemium model could point out the spreadsheet example is somewhat contrived, or that it focuses solely on functionality. A common freemium approach is to offer full functionality in both versions, and to simply offer more of something in the paid version. More storage, more uploads, more downloads.

Fair enough. But here’s one thing freemium fans can’t deny: in their model, a tiny minority of paid users subsidizes the service for everybody. It is this simple fact that makes the freemium model self-defeating, because, for the numbers to work, the price of the paid service must be set artificially high.

In other words, for the marketing side of the freemium model to work, the free offering must be functional and useful, otherwise no-one would use it. The premium version, therefore, can only contain an incremental benefit. But the difference between free and $10, or $5, or even $1, is anything but incremental.

The key flaw in the premium model is that it requires people to overpay for the inherently incremental advantage of the premium service. It has to do this, because the premium users must support the whole user base.

As we saw above, the spreadsheet service could, in theory, be profitably supplied to the entire user base at 30¢ per user per month. But there is a problem. Even if the conversion rate would be expected to rise steeply, it would not reach 100% since some people just refuse to pay, at any price. Josh Kopelman calls this resistance the Penny Gap.

The Penny Gap and the Dollar Trap
In his famous blog post, Kopelman explains why the normal price/demand relationship breaks down on the net when the price hits zero. He points out that it’s significantly harder to get someone to go from $0 to $1 than it is to get them to go from $1 to $2 - even if the gap is the same. The penny gap offers indirect support to the freemium model because it says that no matter how low the price is, some users just won’t cough up. On that basis, it’s better to get a tiny minority of users to effectively pay for everyone else.

There’s no doubt the penny gap exists. So the obvious question is: why?

Does Amazon suffer from the penny gap? It doesn’t look like it. People seem more than happy to pay for the convenience of having their stuff delivered to their door. So what is it about certain e-commerce services that allow them to convince people to open their wallets?

One argument is that people are more willing to pay for tangible goods. Those same goods they can see for sale in the shops, which come with a price tag attached, and are rarely given away for free. The implication is that the free web has made a rod for its own back: by giving away so much stuff, the perceived value of that stuff has fallen to zero, thus making it very hard to get large numbers of people to buy any type of web service at any price. This is an intriguing idea, and worthy of a blog post in its own right. But for today at least, save for pointing out that the Apple App Store doesn’t seem to have this problem, we’re not going to go there.

Another possible explanation (also supported by the App Store) is that there is a fundamental difference between a one-off purchase and a subscription. The recurring nature of subscriptions are exactly what make them so attractive to companies – but users aren’t stupid. They can add up. A monthly subscription measured in pennies will add up to a cost measured in dollars over time, regardless of how often the service is actually used.

There is also the fact that frequently companies make it very difficult to get out of subscriptions, using minimum terms and sometimes invoking penalty clauses for cessation. Even if un-subscribing from your service is only a mouse-click away, the whole concept may have been so tarred in the minds of consumers they don’t even give you the chance to explain how easy it is to drop your service.

There’s also the question of convenience. People don’t like typing their credit card details into payment systems, even if they’re not worried about fraud. It’s a pain and the work involved is the same regardless of whether you’re about to pay $30 or 30¢ – in fact, it seems almost stupid to pay 30¢ with a credit card. Amazon cracked this one right away when they created a system to store your details online. After the initial purchase, the subsequent ones are practically painless.

So, while acknowledging the Penny Gap exists, the reason for it existing – at least for subscription services – could have less to do with the gap between $0 and $1 and more to do with the very nature of subscriptions: the risk of lock-in, the fact that you still pay even if you don’t use the service, the inconvenience (and potential risk) of having to initiate each subscription with a separate credit card transaction.

Assuming we’re on the right lines then, and the problem is not with paying for web services per se, rather with all the baggage associated with subscriptions – what’s to be done? What’s the solution?

MicroBilling
Wikipedia defines microbilling as a service for mobile phones. Here we’ll use a different definition: microbilling involves charging usage-based fees for the use of web services. These micro payments across multiple websites are consolidated into one single monthly credit card or bank debit transaction.

Usage-based billing has been tried before, in a different context to web services, without much success. Micro-payments, at least for content, are pretty much discredited. Why bother even suggesting microbilling for web services has got a chance?

Because it solves all the problems associated with subscriptions. Because it can be made to work. Because it is elegant.

Let’s assume one or more clearing houses exist for handling microbilling. Users create an account, enter their payment details once and then they’re all set up. They find out about participating websites in the same way as they find out about freemium websites: virality, referrals, the news.

They chance upon a new spreadsheet service – it looks interesting. They create an account as normal, perhaps using their credentials from the microbilling service. They can try out the service for free. There is none of this free/premium nonsense – they have access to full functionality from day one.

The microbilling system has minimum and maximum usage thresholds. If the user starts using the service on a regular basis, sooner or later they’ll hit the minimum threshold. A message pops up, explaining that the threshold has been reached, and that if the user wants to continue using the service the minimum threshold payment is 25¢ and the maximum is 50¢ per month. There is a one-click button to agree to any (potential) billing. The user doesn’t need to be Fields Medal winner to work out that the most this service can cost per year is $6. The user clicks yes and goes back to using the product. The next month he goes on holiday and doesn’t use the service much. How much is he billed? Zero. The month after that he has a lot to catch-up on and uses it every day. Monthly bill: 50¢.

Perhaps in a years time, he finds a better spreadsheet app, and stops using the service. How does he un-subscribe? He doesn’t have to. He never subscribed in the first place. By simply not using the system, he won’t ever get billed for it again. He likes microbilling so much he explains it to his mother. She gets it the first time round.

He can log in to the microbilling system at any time and see the status of his current usage for each website, and also how close it is to the minimum and maximum thresholds. Once a month he gets debited for his consolidated usage across all his microbilling websites. He can see a highly itemized bill online, which gives the breakdown of the single transaction he sees on his credit card or bank statement. He never has to swear at a ‘do you want to upgrade?’ message again.

In a nutshell, by making the payment process fairer, painless and transparent, the Penny Gap can be bridged. With a greatly reduced Penny Gap, the base of paying users will expand significantly. With lots of paying users, websites can begin offering their services at a much fairer price because nobody is subsidizing anybody else. With such low prices for services, more people will sign-up, further expanding the user base and further driving down the cost of the service. Rinse and repeat.

Application or service providers can focus their time on innovating their core products, instead of continually testing different variations of the free/premium mix, or deciding what functionality should be included in this week’s sneak preview. This extra time will come in very handy because their users are no longer locked-in with subscriptions. If the website doesn’t continually improve, the users will migrate to greener (though, at these prices, not necessarily cheaper) pastures.

Website owners will wake up to the fact that frequently you can make more by billing a lot a people for a small amount, rather than billing a small number a large amount.

If microbilling can reach critical mass, there will be no going back.

David Semeria
LM Framework
June 2009

Disclosure
LM Framework will use microbilling through the P-BOS licensing system. This post basically summarizes the process we went through when we evaluated the freemium model internally, and as a consequence decided to create our own usage-based revenue model.