Why the internet is perfect for price discrimination

By Felix Salmon
September 3, 2013

Price discrimination is one of those concepts that only an economist could love. But the theory is clear: the more that a vendor can discriminate according to willingness to pay, the more value that vendor can add. Rory Sutherland uses air travel as an example: having a mix of classes allows price-sensitive people to pay low fares, while the rich have a large number of flights to choose from. On top of that, he could have added, airlines are extremely good at exercising price discrimination within classes, so that two people receiving identical service might be thousands of dollars apart in the amount they paid for their tickets.

Airlines have always used a multitude of proxies to determine customers’ willingness to pay: how far in advance people are booking, whether they’re going one-way or return, whether they’re staying a Saturday night, and so on and so forth. But nowadays, online, the amount of information that companies have about their customers has never been higher. And the obvious way to monetize that information is through price discrimination: charge the people with high willingness to pay more money than those who will only buy if the price is low. Adam Ozimek explains:

The more information we have, the more profitable first degree price discrimination will be. As big data and online buying increases the information that business have on us, the ease and profitability of first degree price discrimination will become difficult to resist.

Ozimek says that this kind of price discrimination will be “creepy, invasive, and unfair” — but it will at the same time result in superior products. He doesn’t mention this, but the obvious place for this kind of price discrimination is newspaper paywalls. The FT already does it: there’s no real list price for an FT subscription, and the paper basically just charges whatever it thinks it can get away with, given what it knows about you.

For newspapers with more price-sensitive readers, smart price discrimination is even more important. Ideally, you’d charge every reader just a little bit less than they were willing to pay — and you’d give your content away to the people who were willing to pay nothing. And here’s the thing: newspapers know a lot about their readers — and especially the regular readers who come back often enough to hit paywalls. To take a simple example, they know if those readers are looking at local sports reports, or whether they’re looking at general entertainment news. The former group will be much more willing to pay than the latter. But they’re also quite likely to know a lot more about you than that, including — if you’re someone who’s ever had a print subscription — exactly where you live.

The NYT will shortly roll out a new, lower-priced product, giving some subset of its news to people who don’t want to pay for the whole thing. But that’s just going further in the wrong direction. Already the pricing is sending all manner of bad messages: access to nytimes.com plus a phone app is $15 every four weeks, access to the website plus the tablet app is $20, and access to the website plus both phone and tablet apps is $35. Which logically means that access to the website itself is worthless. (If A+B=15, and A+C=20, and A+B+C=35, then A=0.)

The NYT is in the process of building new products, for which it can then charge varying amounts of money. This is a lot of work, and has reportedly created tensions between the CEO and the editor when some of the new products report up to the former rather than the latter, even when they’re being built by journalists. Instead of thinking in terms of creating a wide range of different products, then, maybe the NYT should just try to do the best journalism it can, and then sell that journalism at a wide range of different prices.

That would be a very tough decision to make: consumers, as a rule, viscerally dislike price discrimination. If you know that your friends and neighbors are getting exactly the same product that you are, but are paying a different price for it, then someone is going to feel ripped off. Still, at the margin, the NYT can start to implement something like this without having to charge different prices. For instance, it can hold off on the paywall for certain readers — the ones with the lowest willingness to pay — while putting it up quite aggressively for others, such as perhaps the ones who spend a lot of time on the business pages. And if you price discriminate by giving away discount codes, few people object at all.

For companies which aren’t as high-profile as the NYT, price discrimination is a no-brainer. Amazon was doing it as long ago as 2000, and Uber does it every day, by charging extremely high headline fees, and then giving away various discount coupons like confetti, to carefully-targeted audiences.

The internet is a zone where companies sell products with zero marginal cost, and with a lot of information about exactly who their audiences are. In that world, it would be weird if they didn’t try to charge different prices to different customers. We’re used to the freemium model, which is very basic price discrimination. In future, expect that model to become a lot more sophisticated.

Comments
11 comments so far

Don’t forget that coupon issuance is a form of dynamic price discrimination, practiced most effectively by supermarkets and retailers like Tesco, which have collected and stored personal data using loyalty cards for decades.

One of the challenges for publisher websites is that much of the most valuable data is associated with a cookie that can be easily deleted (and often are, at rates that can vary drastically by user). Therefore it’s often not possible, and indeed can be highly misleading, to analyze user data over e.g. a 30 day period in the way that you describe.

As a result, only the handful of publisher websites that require logins to view any content have a decent idea what users do on their site over the course of a week or month. Even fewer would be able or willing (as in incented)to do any serious price elasticity modeling. Since Amazon et al require (or come close to requiring) login and authentication, they are able to do much, much more.

Theoretically, things should improve here for publishers with native apps, since more persistent methods of user identification can be used.

So where publishers need to get smart is to look for proxies, just as the airlines did, using some other data source, that predict future subscription purchase behavior via controlled testing. For example, users who regularly arrive at the site on its homepage vs. a deep link could have a higher probability of paying more for a subscription… You get the drift.

(p.s. I used to do this stuff professionally — great post, as usual)

Posted by David4321 | Report as abusive

Uber doesn’t give out coupons ‘like confetti’. You pretty much only get a discount if you’re a first-time rider or if some payments app or some other startup is paying for it.

Posted by absinthe | Report as abusive

My sense is that we will see a movement against this kind of price discrimination at least in some product/sectors. Eg fin services firms are now prohibited from using gender to price products like car insurance in the EU. OK, this is linked to equality legislations, but I suspect if companies start charging different prices based on things like postcodes then some sort of troubles await. This is the sort of issue that can kick off a boycottretty quickly once many people become aware of it – think Starbucks tax avoidance in the UK kind of boycott.

Posted by fxtrader7 | Report as abusive

Re “Which logically means that access to the website itself is worthless”. Very flawed logic there; if that were logical then it would never make sense to buy anything offered as a “buy one get one free” offer. Because the second one is worthless and since it’s the same as the first one that’s worthless too and who in their right mind would pay for two worthless things?

Not Spockworthy.

Posted by anthrosciguy | Report as abusive

“the more that a vendor can discriminate according to willingness to pay, the more value that vendor can add”

I think you have the wrong verb there, Felix. It isn’t “add” value, it’s “extract” value. Unless you’re saying that the only value that counts is what accrues to the sell side.

If that’s the case, though, then haven’t we strayed pretty far from the idea that willing buyer and willing seller mutually agree on prices in open markets?

Posted by Altoid | Report as abusive

Here’s the thing about price discrimination: it also reduces the enjoyment that people get from products. That’s a more fundamental reason for the visceral hate, not a vague sense of unfairness.

As an aside, the perfect illustration of this right now is the video game market, where price discrimination has become pervasive. In video games, though, you’re generally paying for additional stuff, instead of paying different amounts for the same product. Nonetheless, the sales techniques used to price discriminate are completely loathed by most consumers: http://www.417am.com/2013/09/free-to-pla y-games-are-awful-and.html

The link elaborates a bit in layman’s terms, but people with some econ background should have an easier time conceptualizing the problem. Welfare economics acknowledges that what you pay for a product is instrumental in your enjoyment of that product. Consumer surplus isn’t some empty abstraction, but a measure of the value perceived by a purchaser. A product with no consumer surplus is a product a consumer doesn’t care whether they buy.

Price discrimination is the practice of minimizing consumer surplus. See how that causes a problem?

The closer you get to effectively price discriminating, the more you are diminishing consumers’ overall welfare. While they may keep buying in the short term, it’s hard to believe that the real, quantifiable reduction of the value they are receiving won’t eventually have deleterious effects on the market.

Posted by WHS | Report as abusive

Also, that Rory Sutherland article seems off-base.

I can think of one circumstance in which a private producer would use profit-generating customers to subsidize loss-generating customers. That’s when the thing that attracts the profit-generating customers is capacity.

Imagine, for instance, an airline that flies three routes. It knows that being able to provide all three routes is essential in attracting profit-generating customers, but on any given day, only one will generate a profit. It doesn’t know in advance which one. So it makes sense to sell tickets on the other routes at a loss–it can’t get rid of them, and there’s no reason to waste the excess capacity.

But this reasoning breaks down in the instances Sutherland describes. That’s because, when you have first-class and second-class categories, you know in advance which ones will be profit-generating. In that instance, there’s no reason to maintain excess capacity in the loss-generating categories. Just get rid of them!

Posted by WHS | Report as abusive

Also, that Rory Sutherland article seems off-base.

I can think of one circumstance in which a private producer would use profit-generating customers to subsidize loss-generating customers. That’s when the thing that attracts the profit-generating customers is capacity.

Imagine, for instance, an airline that flies three routes. It knows that being able to provide all three routes is essential in attracting profit-generating customers, but on any given day, only one will generate a profit. It doesn’t know in advance which one. So it makes sense to sell tickets on the other routes at a loss–it can’t get rid of them, and there’s no reason to waste the excess capacity.

But this reasoning breaks down in the instances Sutherland describes. That’s because, when you have first-class and second-class categories, you know in advance which ones will be profit-generating. In that instance, there’s no reason to maintain excess capacity in the loss-generating categories. Just get rid of them!

Posted by WHS | Report as abusive

Extracting every last bit of consumer surplus is a way for vendors to maximize revenue in a given set of transactions, but in repeated transactions how do you feel about vendors that manipulate their prices like that?

Posted by chazbet | Report as abusive

Extracting every last bit of consumer surplus is a way for vendors to maximize revenue in a given set of transactions, but in repeated transactions how do customers come to feel about vendors that manipulate their prices like that?

Posted by chazbet | Report as abusive

You forget to mention that McKinsey is advising NYT on the new products. So it is McKinsey which is valuing NYT at zero and access to different devices as valuable.

Posted by 2paisay | Report as abusive
Post Your Comment

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/