Beyond the Deadweight Loss of “All You Can Eat” Subscriptions

In the continuing conundrum of pricing digital content, “all you can eat” unlimited subscriptions have become the latest thing to try/hate. We see it in music, with the Taylor Swift withdrawal from Spotify, and in e-books, with angst over Amazon Unlimited and other subscription services.

At a most basic level, isn’t it clearly perverse that:

• I pay the same for Spotify whether I listen for 2 hrs/mo or 12 hr/day?
• I pay the same for Amazon Unlimited whether I read 1 book/mo or 1/day?
• I pay the same for Netflix whether I watch 2 hrs/mo or 12 hrs/day?
• I pay the same for NYTimes.com whether I read 20 stories/mo or 50/day?

The problem is obvious on the surface, but the fix has not been at all clear.

• All you can eat (AYCE) encourages wasted resources ( a “deadweight loss” as economists say*), and this is especially painful to the creators of content (musicians and authors) who find themselves getting pennies for their hard-earned work.
• But all the conventional set-price alternatives are also very problematic — especially for digital content, which can be replicated at near-zero cost, to economically serve almost anyone who sees value in it.

The problem is that we have had no efficient way to price for a value that varies from person to person and from time to time. For example:

• AYCE under-prices to heavy users, and overprices to light users.
• Set-prices of any kind underprice to those who most value an item or are most able to afford it, and overprice to those who find limited value or have limited means.

This results in a deadweight loss to our economy — a loss to society — because the set price excludes the large numbers of potential consumers who would gladly pay less, and thus would (1) obtain value and (2) generate a profit to the seller.

This is partly a matter of usage volume. Pricing for usage has always been a problem (think of long distance phone bills, and now cellular data), but now it is far more widespread and problematic for products and services that involve significant costs of human creation but low cost of replication. Many have identified the need** to find new business models, but with little clear direction as to how.

I suggest that this derives from an underlying blind-spot — we are stuck in our habit of thinking that prices should be set by sellers, rather than embracing the new dynamics of networked e-commerce that make it practical to dynamically set individualized prices (see my earlier post, “So Last Century…”). The new FairPay pricing architecture promises to move us to a new economics, by setting prices that correspond to a particular individual’s usage context and value received. This has long been an economic ideal (perfect price discrimination), but FairPay shows how the digital world offers a new way to approach this ideal.

Pricing for value

When you think about what is economically efficient, and what is fair to both parties, it is clear that prices should correspond the the value of the experience. Value is not just how much you eat, of what and when, but how good is it — a particular and ever-changing mix of how tasty, exciting, nutritious, timely, sustaining, …and how costly to provide.

Businesses can exploit other models, but that is often unfair to consumers, and wasteful to society as a whole. But pricing for value is challenging, and so is rarely done for consumer services. Let’s look at some of the commonly used options, all of which are good for some of the people, some of the time — but not for others.

• All You Can Eat subscriptions (AYCE at a set price — all the items you want, time-limited to a meal, a month or whatever) — This is simple and easy, but has all of the unfairness and inefficiency noted above. (These and variants are often referred to as paywalls.)
• Usage-related subscriptions (set prices for units of usage, a form of a la carte) — This can correlate well to usage as one important aspect of value, and thus be much more efficient and fair, but it is very unforgiving, and so consumers rightly shy away from it, especially if they are price sensitive. We all know of the old horror stories of teenagers getting surprised by cell phone bills for thousands of dollars, and are always very conscious of “the ticking meter,” Many variations on this have been applied, with mixed results, One variation is set prices for different tiers of usage (common with cellular data and previously for voice, but little used for content). This avoids surprises within one tier of usage, but confronts the user with a cutoff, after which they suddenly must jump a high price hurdle to the next tier for one more unit, or stop until the next cycle. An improvement on that adds “rollover minutes” in which unused usage credits are carried over into later periods (a doggie bag). These are workable and moderately efficient with regard to usage, but get complex, and still present the consumer with set-prices (and potential surprises) that may or may not match to the value to that user. (FairPay draws on aspects of these models in a more forgiving form.)
• Unit purchases of items (essentially AYCE at a set price, of a single reusable item, forever) — This is also simple and easy, and good for those who want to make full use of a given item (song, book, video, app, etc.) but tends to make it prohibitive to use very many items.
• Freemium subscriptions (AYCE at a set price, split between a free tier of items/services, and a paid tier) — This is just a variant version of AYCE that serves two customer segments, but only obtains fees from one. This has become very popular as perhaps the best solution readily available for many digital services, such as for Spotify, newspapers (metered or “soft” paywalls), and many others. But freemium is still a form of AYCE, with all its inefficiency, and it still has a set price, so the challenge of what price, at what level, remains (see post “Beyond Freemium…”). Multi-tier freemium models also exist, and add the efficiency of usage tiers or feature tiers to the basic idea of free+paid options. But still the tiers and prices are pre-set, and still do not adapt to the fact that value varies in many dimensions other than just quantity of usage. As to deadweight loss, freemium reduces some of that, by providing some value to consumers of the free tier — but it fails to provide the incremental value of the full service, or the revenue that many of those free tier users would be willing to pay. (FairPay offers similar benefits to freemium, in a more flexible and potentially efficient form.)

There are also many variants and other pricing strategies that deserve consideration, but have yet to prove widely effective for consumer markets. These may apply to items or subscriptions, and some involve a shared platform supporting multiple vendors. (Of course many other models exist — including other combinations of these techniques.)

• Micropayments (small set prices for each small unit of content) — This variant of usage-related subscriptions has a long history of conceptual appeal, but has the same problems of being unforgiving, with the nagging fear of the “ticking meter.” (To ease the logistics of handling small payments to many vendors, this is often applied across multiple vendors with a common payment platform, such as in app stores and in multi-publication subscription services.)
• Pay What You Want (PWYW) (AYCE of a single item, forever, but the consumer sets the price) — This gained fame after Radiohead offered a PWYW album, and has proven successful in some markets such as indie music, games, and e-books (see post “…Still Crazy…”). While it has been well established that most people will pay, and many will pay fairly or even generously, it is still very iffy as a sustainable business model. Its success so far has been greatest in special promotions, such as Radiohead, or Humble Bundle. (A limiting factor has been that most PWYW offers are in a single transaction setting — FairPay shifts this the setting to that of an ongoing relationship.)
• Pay What You Want “post-pricing” (PWYW, where the consumer sets the price after experiencing the items) — This “post-pricing” has been little used so far, but has a big advantage in that it let’s the user set a more generous price because they have confidence that they will not be disappointed — and it signals greater trust and confidence from the seller, to further encourage generous pricing. (One established form of post-pricing is shareware, but with only the choice between a set price or zero. Again a limiting factor has been the single transaction setting, unlike the ongoing relationship setting that is central to FairPay.)
• Tip-jars or Microdonations (blending PWYW with micropayments) — These have gained limited traction as a way for small sellers to pool voluntary payments using a common payment platform, such as for blogs. (Again, while pooled, this is purely voluntary, and not wrapped into a relationship-building structure as FairPay is.)
• Value-based or Performance-based pricing (collaborative post-pricing based on actual results of use) — Near the ideal of pricing for value with perfect price discrimination. As currently done, this has been generally impractical in consumer markets, but it has proven very effective and efficient for big-ticket industrial items or services, where the parties can agree on how to measure value and share in the value surplus that the product/service creates, and can do the analysis that takes. The advent of Big Data is making this more feasible and effective in a wider variety of businesses. (As explained below, FairPay provides a way to extend this to mass-customized consumer markets.)

What is fair to me is not fair to the next guy

While the deadweight loss of All You Can Eat is a big problem, usage pricing can only solve part of the problem of pricing digital, even if it could be done perfectly. Even at the same usage level, a price that is fair to me may not be fair to the next guy. What if I get great value from x number of items because they contribute significantly to my business, investments, health, hobbies, social life, cultural interests or whatever, but someone else may find some value in them, but not as much? Alternatively, what if the perceived value is the same (such as for health, or to a sports fan), but I am affluent and can easily pay a high price for that value, but someone else is a student, retired, or disadvantaged?

Simply charging based on units of usage still has the problem of the Long Tail of Prices. A number of consumers (the short “head”) who get high value or have high ability to pay could potentially be enticed pay more, and should do so to help support the creation of this value (much as patrons of cultural creation do). The “long tail” of very many consumers who would get less value or have less ability to pay, and so do not buy, might happily pay a lower price, and thus contribute added profit to the creator of that value. It is well know in economics that it is most efficient to sell to everyone who will pay a price greater than the marginal cost of production (which for digital products is nearly zero) — if you can get those willing to pay more to do so.

Units of usage are just not an adequate measure of value. (Why is it that very expensive gourmet restaurants serve very small portions with few complaints, while mass-market restaurants feature large portions — and, of course, “all you can eat” buffets?)

The efficiency of FairPay

Essentially FairPay takes the principles of the value-based and performance-based pricing that work increasingly well in B2B markets, and applies them in a lightweight and intuitive form to consumer markets. In doing this, it can flexibly blend features of many of the models described above in a new paradigm. Post-pricing is used in combination with aspects of freemium and PWYW, along with post-pricing, in a new way that gives buyers and sellers evenly balanced power to collaborate over time to set individualized fair prices in “dialogs about value” that can consider all of the relevant dimensions of value and fairness.

For more specifics of how FairPay changes the game to enable personalized pricing based on individual value see the other posts on this blog, and the FairPay Web site.

• The previous post digs into the specific example of music, providing a good example of the subtleties that FairPay can address, but music is a hard business to change because of the role of entrenched labels and other intermediaries that make it difficult for those selling music to experiment with new pricing models (especially after Steve Jobs got them to buy into iTunes!).
• Very much the same analysis can be done for other industries that have fewer structural constraints. For example, journalism, such as newspapers and magazines, may be much more amenable to FairPay in that there are no powerful middlemen to stand in the way of pricing innovation. Several prior posts address that (such as those relating to NYTimes, NYTimes Premier, Bezos and the Post, and Omidyar, and I expect more to follow). Briefly, there are the same issues of widely varying value propositions, and the need to get readers to pay a fair price to support the expense of high-quality journalism.
• Much the same applies to many other businesses, including e-books, video, apps, games, and other digital services. One post explores the rich nuance that FairPay can offer in dealing with the widely varying value of Travel Guides.

Of course it does no good to set individualized prices if those asked to pay more than others will not agree to do so. That is why perfect price discrimination has been so elusive. What is different about FairPay is how it nudges those who can and should pay more to agree that it is fair that they do so. That is addressed in the examples just referenced, and more fully in posts on Seller Control and on Making People Want to Pay You.

Isn’t it time to move beyond our deadweight losses and try a pricing model that addresses the dynamics and context-dependency of digital so that as many people as possible pay a fair price? …A model that enables a product or service to by provided to everyone who is willing to pay more than the very little that it costs — and enables the provider to gain a fair profit from all of them?

— — — — — — — — — —

*Not being expert in economic theory, I am not sure if all of the economic inefficiencies described here are strictly considered deadweight losses (and wonder how well economic theory addresses the new varieties of exchange enabled by digital) — but it is such a great term! My point is they are inefficiencies (in theory and in practice) — and that many sales that would create an economic surplus for both buyers and sellers do not happen because of these kinds of pricing inefficiencies.

Two interesting items on this theme are:

• The classic amusing paper, The Deadweight Loss of Christmas
• An interesting post on AYCE sushi bars

…And don’t start me on why I have to pay $4 per month for ESPN even though I never watch it!

**The music business has been one of the hardest hit by this and most outspoken about the need for new business models (as addressed in the prior post). For example, see recent pleas for new business models for music in Harvard Business Review, from an ex-Rhapsody exec, and in Music Industry Blog posts here and just earlier.