3	Versioning Information

We’ve seen that a key aspect of pricing information is to use value-based pricing: sell your product at different prices to different consumers, according to how much they are willing to pay for it. We looked at two approaches to value-based pricing in Chapter 2: personalized pricing and group pricing.

Personalized pricing requires knowledge about individual customers. The best intelligence about customers comes directly from them, as when customers communicate their needs and indicate the products they would like to see or the categories of information of interest to them.

We certainly encourage companies to develop and exploit two-way communications with customers. However, you can still get valuable data about customers without customer-provided profiles, without expensive marketing data, and even without consumers’ active involvement. How? Answer: You can learn a great deal about your customers by offering them a menu of products and seeing which one they choose.

For example, if you offer a product line with one product targeted for professional users and one product for amateur users, you can simply observe your sales figures to see how your market splits. We call this strategy “versioning.” It means offering your information product in different versions for different market segments.

In this chapter we show you how to design your “product line” to capture the greatest profit from the information you are selling. Your profits will depend on both the total value you create for your customers and the fraction of that value which you are able to extract through the fees you charge for the information. To maximize your profits, you want to make the total value created as large as possible and then extract as much of that value as you can. This observation leads to the two basic principles for designing a product line of information goods.

• Offer versions tailored to the needs of different customers. A full line of information products will maximize the total value of the information you are providing.
• Design these versions to accentuate the needs of different groups of customers. Emphasizing customer differences allows you to extract more of the value you have created, as each customer selects the version that best meets his or her needs.

Economists call the second principle self-selection. You don’t have to figure out what value the customer puts on your information product, because the customer reveals that value through the version that he or she selects.

Consider the Quicken example in Chapter 2. How did Intuit actually solve its pricing problem? It created two versions of the software, Basic Quicken, which sells for about $20, and Quicken Deluxe, which sells for about $60. The Deluxe version has a variety of extra features that appeal to the power users but aren’t that attractive to the occasional user.

Let’s see how the two above principles have been applied to one of the oldest forms of mass-market information provision: the book. How can a publisher such as Viking make the most money selling the newest Stephen King novel? Viking would like to sell the novel at a high price to the avid fans who will pay a lot for their favorite author’s most recent book. But a high price would no doubt discourage purchases by those who are less enthusiastic readers of Stephen King. Ideally, the publisher would like to sell every copy of the book at a different price—that is, engage in the kind of personalized pricing that we described in the previous chapter.

The problem is, the publisher has no way to tell what any given individual is willing to pay for the book. Politely asking those customers who place the highest value on the latest Stephen King book to pay extra because they like it so much will not do the trick for obvious reasons. (Even if Viking or its distributors could keep track of readers who had rushed out to buy prior Stephen King books, attempting to charge more to this group would only encourage them to hide their identity or buy the book through another channel.) So it appears that the best the publisher can do is to differentiate the price by groups: sell at one price to the book club members, say, and at another price to retail book stores.

In fact, the publisher can do much better by applying our second principle: designing versions to emphasize customer differences. Here, high-value customers are impatient to get the book, while lower-value customers can more easily wait. The main difference here involves patience. Thus, the key to versioning books is to delay before offering less expensive versions. This is precisely what publishers do. After selling hardback copies to the intense fans, libraries, and book clubs, Stephen King’s publisher then releases the book in paperback—so all those other fans can purchase it. And finally, after a few years, the book might even be remaindered and sold at an even lower price to those who scrounge around on the bargain tables. The book example is no doubt familiar to you. But our extraction principle applies widely to the sale of information of all types.

When you think about it, releasing different versions over time is a pervasive strategy for selling information. Movie producers initially release their productions in first-run theaters. After a few months they move to the hotel and airline market. A few months after that, they sell to the home video market. All those young, impatient people go to the movies. Parents with small children and empty nesters stay home and watch the videos a few months later.

DESIGNING YOUR PRODUCT LINE

So how can you use versions of your information in a way that induces self-selection? The key is to identify dimensions of your information product, such as timeliness, that are highly valued by some customers yet of little importance to others. Then offer versions that differ noticeably in ways designed to appeal selectively to each type of customer. The fact that different types of customers value these dimensions differently is what provides the basis for successful sorting.

Delay

Information is like an oyster: it usually has the greatest value when it is fresh. This is especially true of “strategic” information, such as information about stock market or interest-rate movements, where individuals possessing the information have a strategic advantage over those lacking it. But the principle applies more broadly, since we all like to think of ourselves as being up-to-date.

The fact that your information customers want the latest information means they will pay more for fresh information, making it worth your while to acquire and deliver information in a timely fashion. What does it say about versioning? Following the principle of looking for ways in which consumers differ, the key point is that consumers differ widely according to how eager they are for various types of information. This observation underlies the common versioning tactic of delay.

Delay is a tried and true tactic for companies selling various services, not just information. Federal Express, not known for “delay,” offers two classes of service, a premium class that promises delivery before 10 A.M. and a “next day” service that only promises delivery some time the next day. To encourage the senders to self-select, Fed Ex will make two trips to an address rather than deliver nonpremium packages before 10 A.M. They realize, quite correctly, that providing premium service for “ordinary” packages would reduce the value of premium service. Similarly, it has been claimed that the U.S. Postal Service has slowed down first-class service to make more money off of its premium overnight delivery product, Express Mail.

In the same way, information providers can offer early delivery of information at a premium. For example, PAWWS Financial Network charges $8.95 per month for a portfolio accounting system that measures stock values using twenty-minute delayed quotes. For $50 per month they will use real-time quotes. We don’t know how they buy these quotes, but it would make sense for them to purchase real-time quotes, which they immediately forward to the high-paying customers, then delay the release of those same quotes for the other customers. PAWWS is willing to incur the extra costs of delay to get customers to self-select, just as Federal Express does when making two visits to an establishment rather than one.

User Interface

Another possibility is to provide high-paying customers with more powerful search capabilities. It often makes sense to offer different search interfaces to experienced and inexperienced users. In many cases, experienced users tend to be users with high willingness to pay; they are the customers who first signed on to purchase the information and generally use it most intensively. Allowing high-paying users more elaborate search capabilities makes sense in this case, even though there is little or no incremental cost associated with a more elaborate interface.

Smart design of user interfaces supports the idea of a more elaborate interface for more experienced users. Casual users typically welcome a stripped-down interface, while advanced users can handle additional capabilities. This makes the search interface an ideal candidate for versioning. (Also, as we’ll see in Chapter 5, a simple user interface makes it easy for customers to start using your product, while later on a more involved, proprietary interface can make it more difficult for these same customers to drop your product for that of a rival.)

The Knight-Ridder company Dialog uses this strategy in its Web-accessible databases. One product, DialogWeb, is offered to “information professionals, on-line searchers, researchers, and other professionals.” Another much cheaper and less powerful product, DataStar, offers a subset of the full Dialog database, with a much simplified user interface. DataStar advertises that “no training is required,” which is attractive to nonprofessional searchers. But DataStar lacks the power of the full Dialog, making it unattractive to professionals. By versioning its product with different user interfaces, Knight-Ridder can simultaneously maximize the value of its database to customers and capture a large fraction of that value for itself.

Convenience

A versioning strategy that is closely related to delay is control of convenience by restricting the time or place at which an information service is used. Before the Web became popular, Dialog used to offer Dialog after Dark, a database searching service that was restricted to use after 5 P.M. Video rental stores now offer overnight, two-day rentals, and five-day rentals. Divx, which we will discuss in more detail in Chapter 4, offers DVDs that can be viewed only during a particular forty-eight-hour period.

Companies can also restrict access by location rather than time. For example, some on-line database providers have offered libraries licenses that allow unlimited use by patrons within the library but restrict use by off-site patrons.

Image Resolution

You can also use visual resolution to discriminate between users. For example, PhotoDisk has a library of photographs on the Web. Professional users want high-resolution images that can be printed in commercial journals; non-professionals want medium- or low-resolution images for newsletters. PhotoDisk sells different size images for different prices; at the time this chapter was written, it sold 600K images for $19.95 and 10Mb images for $49.95. Its on-line catalog offers small, thumbnail images called “comping images” that potential purchasers can examine. Once users choose the image they want using the low-resolution version as a guide, they can then download a medium- or high-resolution version, depending on their needs.

Speed of Operation

When selling software, a common strategy is to sell versions with different capabilities. Wolfram Research sells Mathematica, a computer program that does symbolic, graphical, and numerical mathematics. At one time, in the student version of Mathematica, the floating-point coprocessor was disabled, slowing down mathematical and graphical calculations. To implement this strategy, Wolfram had to add a floating-point library to the package at additional cost to itself, even though the software package with the floating-point library sold for a much cheaper price.

This same strategy shows up in hardware. The IBM LaserPrinter Series E was functionally identical to the standard LaserPrinter, but printed five pages per minute rather than ten pages per minute. A leading consumer testing lab for computer equipment found that the difference in speed was due to a chip that inserted wait states to slow down the printer! Why did IBM deliberately degrade the performance of its printer? Company managers realized that if they made the performance of the Series E too good, it would cut into the sales of their standard model. By versioning their product, they were able to sell to the home-office market at an attractive price without cannibalizing the sales of their professional model.

Intel followed much the same strategy with its 486SX chip, designing the chip with an integrated mathematical coprocessor that was then disabled. That allowed Intel to sell a low-priced chip to those who didn’t need floating-point calculations while still maintaining a relatively high price for the math-enabled CPU.

Flexibility of Use

Another important dimension of information that can form the basis for versioning is the ability to store, duplicate, or print the information. Back in the days of copy-protected software, some software companies (such as Borland) sold two versions of their software—a low-priced version that could not be copied and a high-priced version without the copy protection. Nowadays, Lexis/Nexis imposes charges on some users for printing or downloading information. If customers differ significantly in their willingness to pay for storing, copying, or transferring information to other media, this, too, can form the basis for profitable versioning.

Capability

Table 3.1 summarizes the product line of Kurzweil, a software producer of voice recognition products. The products are distinguished by the total size of the vocabulary included and by the addition of vocabulary appropriate to specific professions. Note the dramatic differences in prices: the high-end version for surgeons is a hundred times more expensive than the entry-level software! Kurzweil has correctly recognized that different market segments have different needs—and that the high-end will pay handsomely for the enhanced capability.

Table 3.1. Kurzweil’s Effective Versioning by Capability

Product	Price	Description
VoicePad Pro	79 $	Vocabulary of 20,000 words
Personal	295	Vocabulary of 30,000 words
Professional	595	Vocabulary of 50,000 words
Office Talk	795	General office staff
Law Talk	1,195	Legal vocabulary
Voice Med	6,000	Medical offices
Voice Ortho	8,000	Special purpose medical vocabulary

Features and Functions

Intuit’s versioning of Quicken, discussed at the beginning of the chapter, is an example of how to use the feature set of a product to segment a market. The Quicken Deluxe version offers a mutual fund finder, a mortgage calculator, an insurance needs estimator, and other features valued by high-powered users. The basic version of the software offers only the core checkbook software. Intuit has pursued the same strategy with TurboTax, selling both a stripped-down and a deluxe version.

Comprehensiveness

In some cases, comprehensiveness is a crucial dimension: some customers will pay a big premium for more complete information. Information completeness varies a great deal, depending on the context. Consider how people use Dialog. Public affairs specialists and journalists like the fact that they can now search newspapers around the country or around the world. Scholars and students writing in-depth articles will place great value on historical depth. For marketing purposes, managers often value information that is broken down by customer or offers lots of details about historical purchasing patterns. The difference between DialogWeb and DataStar rests partly on these distinctions, which are a natural dimension along which any database provider can base different versions.

Annoyance

A prime example of this is “nagware,” a form of shareware that is distributed freely but displays a screen at the start or end of the session encouraging you to pay a registration fee. Public television stations use this strategy in their fundraising drives. During one recent campaign, our local PBS station announced that it would stop breaking into the musical performances if users would just donate another $10,000 to meet the station’s goal!

Support

The final dimension that we consider is technical support. Netscape originally made its browser available for free in a download over the Internet and for a price on a CD that came with a manual and access to technical support. Of course, by offering a downloadable version for free, Netscape gets around the “experience good” problem we described in Chapter 1: anyone can try the product with little or no risk to see if they like it.

McAfee Associates, which we discuss in more detail in Chapter 4, offers its virus detection software in a free, shareware version or as part of a subscription service that provides professional advice, notification, and technical support.

This strategy is somewhat dangerous for two reasons. First, if your customers really need technical support, they may decide your product is of low quality. Second, technical support is very costly to provide. Promises to offer support that are not delivered can be disastrous in terms of public relations.

In Table 3.2 we list the various dimensions we’ve discussed alongside a list of users or uses for which these dimensions have meaning. This list is not meant to be complete, and the examples should only be taken as illustrative. There are as many dimensions on which to version as there are dimensions to your product. Versioning is thus very product-specific.

ADJUSTING PRICE AND QUALITY

Your goal in versioning your information product is to sell to different market segments at different prices. By creating low-end and high-end versions of your product, you can sell the same thing to customers with significantly different levels of willingness to pay.

Table 3.2. Product Dimensions Susceptible to Versioning and Their Likely Users/Uses

Product Dimension Delay	Likely Users/Uses Patient/impatient users
User interface	Casual/experienced users
Convenience	Business/home users
Image resolution	Newsletter/glossy uses
Speed of operation	Student/professional users
Format	On-screen/printed uses
Capability	General/specific uses
Features	Occasional/frequent users
Comprehensiveness	Lay/professional users
Annoyance	High-time-value/low-time-value users
Support	Casual/intensive users

If your premium-price, high-end product attracts some low-end customers, that’s great: you’re getting more revenue from them than if they had stuck to the low-end product. So, it pays to make your high-end product as attractive as possible. The problem arises at the other end of the product line: if your low-end version is too attractive, it may attract some customers who would otherwise pay a premium price for the high-end version.

There are two ways to avoid this cannibalization. First, reduce the price of the high-end product to make it relatively more attractive. Second, reduce the quality of the low-end product to make it relatively less attractive.

Discounting Your High-End Product

When you create low-end information products, you may have to cut the price of your high-end product to keep your high-value customers happy. You should think about this choice the same way you think about pricing to meet the competition. Do your high-end products really offer sufficient value to your customers? If you discount the high-end price, will the increase in sales compensate you for the price reduction? Don’t lose track of the fact that high-end sales lost to your low-end product still contribute to your revenues. What’s important is the difference in the revenue you get from the high and low ends of your product line.

Value-Subtracted Versions

As we’ve indicated, versioning works for all kinds of goods. But versioning information has some special features.

For physical goods, it is usually more expensive to produce an extra unit of the high-quality versions. A Lexus costs more to build than a Camry, and a nineteen-inch TV is more costly to build than a fifteen-inch TV. But with information, it generally costs just about as much to distribute the fancy version as the plain version. In many cases, in fact, production of the low-quality version incurs additional costs, since it is often a degraded form of the high-quality version.

Think about delay. A financial service firm that offers real-time and delayed stock prices needs added storage capacity to offer the delayed service. Or resolution: the images have to be scanned using a high resolution and then degraded to produce the low resolution. Or speed: Wolfram Research had to build or purchase a floating-point emulation library in order to produce the student version of its software.

With information you usually produce the high-quality version first, and then subtract value from it to get to the low-quality version. This isn’t universally true: versioning based on technical support costs more. But it is true often enough to formulate a basic design principle: if you add a fancy new feature to your software or information product, make sure there is some way to turn it off! Once you’ve got your high-value, professional product, you often want to eliminate features to create a lower-value, mass-market product.

PITFALLS—AND HOW TO AVOID THEM

Although customers may not like some of the practices we have suggested, it is important to remember that the low willingness-to-pay market often would not be served at all unless producers can “degrade” the product in a way that discourages high-willingness-to-pay consumers from purchasing it. Without the ability to offer distinct versions, your best strategy may be to offer the high-end product only, and offer it at a premium price. In dealing with customers, you should emphasize that the cheaper versions enhance consumer choice; indeed, customers picking them are revealing that they value the option to buy a low-end version of the product at a discount.

The key issue in designing an information product line is to pick the right dimensions on which to adjust the quality and to make sure that the quality adjustment cannot be undone by clever consumers or intermediaries.

For example, Microsoft offers two versions of its Windows NT software: the Windows NT Workstation, which sells for about $260, and the Window NT Server, which sells for $730–$1,080, depending on configuration. Workstation NT can run a Web server but accepts only ten simultaneous sessions; the server version will accept any number of simultaneous sessions. According to an analysis by O’Reilly Software, the two operating systems are essentially the same. In fact, the kernel (the core component of the operating system) is identical in the two products and relatively minor tuning can turn Workstation NT into Server NT. In response to O’Reilly’s analysis, Microsoft claimed that the two operating systems differ on more than 700 counts. According to one reporter:

“While the Big ‘M’ folks in Redmond maintain the products are vastly different, critics allege Workstation can be switched into the Server version with a few easy tweaks. An official Microsoft marketer suggests that’s like arguing the only difference between men and women is a Y chromosome. We think it’s more akin to discovering your date is in drag.”¹

Microsoft’s marketing strategy made sense. The problem was that some sophisticated consumers were able to turn the low-cost version into the high-cost version. The danger for Microsoft was that system administrators could easily upgrade Workstation NT into Server NT, thus defeating Microsoft’s strategy.

ON-LINE AND OFF-LINE VERSIONS

Our discussion of market segmentation brings up an interesting point about on-line information: it is often also available off-line. In many cases, off-line information is actually more convenient to use. For example, many readers feel that it is easier to read text on paper than on a screen. Similarly, music probably sounds better on your stereo than on the tiny speakers that came with your computer.

This quality difference cuts two ways: on the one hand, the fact that text quality is poorer on the screen than on paper means that you can sometimes offer documents on-screen for free and then sell them on paper. Ed Kroll’s famous book, The Whole Internet, could be downloaded on-line at no cost at the same time it was sold in bookstores for $24.95.

In a similar vein, as mentioned earlier, Netscape Navigator can be downloaded for free or purchased on disk. Many users are willing to pay for the disk version since they find it more convenient to use, not to mention the fact that it comes with printed documentation and user support. We are told that Netscape was quite surprised by the success of the retail product since the people in the company all had high-speed Internet connections and disdained printed documentation. They didn’t appreciate the position of the home dial-up user with a 4,800 bps modem. But once Netscape realized a market was out there, the company was more than happy to sell into it.

The difference between on-line delivery and off-line delivery cuts the other way, too. It is much cheaper to sell information on-line since there are no production or distribution costs. If you want consumers to buy the on-line product, you should try to figure out ways to make it more attractive to the consumer than the off-line version.

Esther Dyson offers a useful way to think about this. She suggests that you treat your on-line content as if it were free. This way, you focus your mind on ways to add value to your product. Dyson’s Dictum is great advice, since it makes you think about information provision as a service rather than a good.

A practice of the National Academy of Sciences Press is a good illustration of this principle. It offers both on-line and printed versions of its books. Because the on-line version of a book is great for browsing and the printed copy is great for actual reading, the on-line version adds value because it gives the reader a way to browse without cannibalizing the sales of the hard copy.

In many ways, selling information on-line and off-line is like selling physical products through two separate channels of distribution. In deciding which “channel” to promote, and how to price into each channel, you need to consider not only the costs associated with that channel and the character of demand through it but also the extent of channel spillover or cannibalization. If an on-line download of information displaces a hard-copy sale, revenue may well be lost. On the other hand, if today’s download enhances demand tomorrow for both on-line and off-line information, sacrificing some current revenue to make more in the future may make good business sense.

The key question to ask yourself is whether the on-line version is a complement or a substitute for the offline version. If it is a substitute for the off-line version, then you’ll want to charge for it, recovering costs through fees or advertising, or version it so that it doesn’t directly compete with your offline version. If it complements your offline version, then you want to promote it as aggressively as possible, since it adds to the bottom line by encouraging sales of the off-line product.

Often, providing information on-line enables the supplier to add value in ways that would not be possible off-line. This in turn creates opportunities for new versions. Perhaps most obvious and important fact is that on-line information can be searched, sorted, or filtered electronically. On-line information can also provide cross-references through hyperlinks to further information. West Publishing sells CDs containing legal reference materials with hyperlinks to its on-line subscription service. These CD sales thus promote West’s subscription revenues.

Merely posting something that is available in print on-line doesn’t add value to it, so you won’t be able to sell it for a premium price (although you may reach more customers). And, even more important, if you don’t come up with ways to add value to your on-line content, your competitors will surely come up with a way to add value to their content.

When you get right down to it, it is very rare to find someone who has truly unique (versus merely differential) content. AP, UPI, and Reuters all sell newswire stories. Reuters managed to gain a competitive advantage by using the strategy described earlier: bundling news stories into packages targeted at specific industries. Reuters saw the wisdom in Dyson’s Dictum—treat your content as if it were free. The company’s strategy focused on adding value to its on-line services, not just providing the same content on-line as off.

HOW MANY VERSIONS?

When you start to think about versioning your software or information services, the first issue that comes up is how many versions you should offer. The answer is highly context dependent, but we can offer some guidelines.

First, one version is too few, for just the reasons we have described earlier in this chapter. Everyone who sells information should think about what they might do to segment their market. On the other hand, you can have too many versions in your product line. On the supply side, there are costs to maintaining several different products. On the demand side, you run the risk of creating user confusion. You must make it crystal clear to your users which version you think is appropriate for them.

Kurzweil’s menu, shown in Table 3.1, is a good model: customers know their own line of business so there is little confusion between the medical and legal versions. There is no reason not to create additional versions targeted toward finance, agriculture, and so on.

Dialog’s segmentation into casual and professional users is natural, but the names (DialogWeb and DataStar) are not particularly descriptive. However, trying to subdivide this market further runs the risk of user confusion, especially if Dialog doesn’t come up with better names.

But what dimensions should you vary to construct different versions? There are two general strategies: analyze your market and analyze your product.

Analyze Your Market

Think about whether your market naturally subdivides into different categories of consumers and whether their behaviors are sufficiently different that they want (or are willing to tolerate) different quality classes of product.

Airlines recognized early on that there were basically two classes of travelers: those who traveled for pleasure and those who traveled on business. Tourists normally planned in advance and stayed for several days, typically including a weekend. Business travelers had quite different patterns of behavior: their trips were short and during the week. They also often had to change routes on short notice. These key differences allowed the airlines to segment their markets by offering two fare classes: advanced purchases, with Saturday-night stayover and penalties for changing the fare, and ordinary Y-class travel.

The business/tourist distinction is a natural one for airlines, and it is a good place to start thinking about your market. Are there professional and amateur users? If so, what distinguishes them? Your low-end information product should be lacking the key attributes that high-end customers uniquely crave. If you understand your market well enough, you will be able to come up with versions that both give value to your customers and raise revenues.

Analyze Your Product

You should take a hard look at your product and identify its key attributes, with an eye on segmenting the market according to one or more of these attributes. Look at the list in Table 3.2. Can you use delay, user interface, resolution, speed, format, capability, or features to segment your market? A good starting point is to consider offering a high-end and a low-end version for each key attribute for which there are clear differences in customer value.

As we suggested earlier, a common strategy is to produce the highend product and then degrade it in some way to get the low-end version. You should think carefully about how this approach might apply in your market.

Look at Table 3.2 and see if these dimensions apply to your product. Choose the resolution of your images for your professional art market, build your search capabilities for the most sophisticated group of users, design the speed of your downloads for your most demanding users, and so on. Add features until the incremental value of those features to your most demanding customers just equals the incremental development cost.

Then, when you are ready to develop the product for the lower-end markets, just start turning features off. Take the high-resolution images and produce low-resolution versions. Put wait states in your program to slow it down. Remove the buffering. Do whatever it takes to make the product relatively unattractive to the high willingness-to-pay users but still attractive to the next group down.

Designing your product from the top down offers two advantages. First, it’s easy to meet the competition if it arises. Suppose your “premium” version uses high resolution and your “standard” version uses medium resolution. If your competition comes out with a high-resolution product at massmarket prices, you can respond by repositioning your high-end product for the mass market. Since your premium product is already available, this will usually just involve some new packaging and promotion.

The hard part is coming up with new features for the high-end version. But this is what your R&D group is supposed to be doing—designing new features for your most demanding customers. These should go into your high-end product first, and then diffuse down to the lower-end products as competitive upgrades.

The second advantage is that you can use the low-end version of your product as a way to “advertise” the high-end product. Just as 72 dpi comp images on the Web offer a sample of the high-resolution images that PhotoDisk can provide, the student version of Mathematica serves as an advertisement for the professional version. You should make sure that your low-end users know how much better or faster the high-end version will work.

PAWWS, mentioned earlier in the chapter, is facing competition from other firms offering portfolio analyses, among them RiskView, a joint venture involving Dow Jones, IBM, and Infinity Financial Technology. A press release describing the system explains the motivation of these firms:

By offering free access to its databases, Dow Jones said it hopes to create wider demand for its indexes from the investment community and academics. Infinity believes the new product will stimulate demand for risk analysis from brokerage firms, prompting them to turn to companies like itself to design more sophisticated systems. And IBM gets to demonstrate its Internet capability in the financial risk and management arena. IBM can also link the new product to other services it provides that give investors additional information.²

Each of these companies indicates that it is willing to give away the services of RiskView to encourage individuals and firms to make use of other services for which they charge: Dow Jones wants people to buy customized data, while IBM and Infinity want people to buy more sophisticated services from them. This is potentially formidable competition for PAWWS. On the other hand, PAWWS has a significant advantage in being first to market and having a knowledge base of experience in dealing with its customers.

GOLDILOCKS PRICING

If you can identify many different constituencies for the information you sell, and there is little likelihood of user confusion, there is no reason not to offer many different versions. Consider the industry news feeds offered by Reuters and other on-line services. There is little risk of confusing an airplane manufacturer with a fast-food business, so why not divide the market as finely as possible? Indeed, in many cases, less information can be more valuable: by filtering and sorting the information, so the airplane manufacturer does not need to flip past pages describing fast-food franchising practices, the information service becomes more valuable.

On the other hand, mass-market software is often offered in just one or two versions. There are two reasons for this. One is the network effects mentioned in Chapter 2. Users want to be able to exchange electronic documents, and it is much more convenient if there is only one version of the product. Look at all the flak Microsoft got by changing the file formats for Office 97. (Of course, Microsoft’s strategy of one-way compatibility probably accelerated the adoption of Office 97; we’ll talk more about this in Chapter 7.)

The other reason is that naive users often have trouble identifying which product is appropriate for them. However, this problem can be turned around and even used to your advantage. For example, if you buy a new digital camera, you are likely to get a stripped-down version of Adobe’s Photoshop software called PhotoDeluxe bundled with your camera. As a first-time purchaser of the camera, you are likely to be a new user of digital photography and can use the PhotoDeluxe out of the box. As you become more sophisticated, there is a good chance you will upgrade to Photoshop, the professional version of Adobe’s software.

But what can you do if you can’t figure out what the “natural” user classes are? While lots of organizations decide to produce two versions—“professional” and “standard”—we think this is probably not the best choice. A better policy, we believe, is to produce a “standard,” a “professional,” and a “gold” version. That is, we suggest adding a highend package targeted toward users with very high value for the product.

The rationale for this suggestion derives from a psychological phenomenon known as “extremeness aversion.” Consumers normally try to avoid extreme choices—it leaves them out on a limb. It’s perceived as risky to go for the top or the bottom of the product line for most consumers, and much safer to choose something in the middle. Positioning a product so that it represents a compromise will end up getting you extra purchasers. Just like Goldilocks, most consumers don’t want to choose between “too big” or “too small.” They want the product that is “just right.”

Consider a fast-food restaurant like McDonald’s and imagine that it offers just two sizes of soft drink: small and large. Some users are sure of the size they want, but others will be uncertain. They will agonize over the choice, and some will come down on the side of the smaller, cheaper size, generating less revenue for the restaurant.

Now suppose that the restaurant offers three sizes of soft drink—small, medium, and large. Those who can’t make up their mind now have an easy out: choose the medium size. This will happen even if the medium size in the three-choice example is the same price and size as the large size in the two-choice example! By adding a jumbo size that almost no one consumes, the producer can end up selling more than he would with only two choices, in part because the median product looks attractive in comparison with the expensive, jumbo version.

This effect can be significant. Itamar Simonson and Amos Tversky describe a marketing experiment using microwave ovens.³ When the choice set consisted of a bargain-basement oven at $109.99 and a mid-range oven at $179.99, customers chose the midrange oven 45 percent of the time. When a high-end oven at $199.99 was added to the choice set, the same midrange oven was chosen 60 percent of the time! As Smith and Nagle point out: “Adding a premium product to the product line may not necessarily result in overwhelming sales of the premium product itself. It does, however, enhance buyers’ perceptions of lower-priced products in the product line and influences low-end buyers to trade up to higher-priced models.”⁴

Extremeness aversion is used all the time in marketing. Every restaurateur knows that the best selling wine is the one with the second-lowest price on the menu. A common practice is to offer an obviously low quality wine at the bottom end, and set the price of the next wine up to be only slightly higher. This makes it seem like a really good deal, virtually guaranteeing significant sales.

How can extremeness aversion be used for information goods? The important thing to recognize is that the product you really want to sell is the middle product—the high-end product is there only to push people toward the compromise choice. If you are selling a newsletter, consider offering an immediate notification service of news events. If you are selling images, offer a superhigh-resolution version that would exceed the needs of most users. If you are versioning based on different feature sets, add features that almost no one would use but that give the high-end product a distinct identity.

One important strategy is to offer premium quality technical support as the main differentiator of the “gold class.” This might be something like an “immediate response” line that connects users to technical support people without delay. This costs very little to offer. As long as too many people don’t choose the gold version, the cost of adding this kind of support will be small.

CUSTOMIZING BROWSER AND CONTENT

In Chapter 2, we argued that Java could be used to customize information you collect about user behavior, allowing you to assemble a much richer set of information. Java can also help you in versioning information, since it can be customized to display the particular type you are selling in an optimal way.

For example, if you are selling bit-mapped images of text pages, you can optimize the viewer for black-and-white textual material. If you want to display objects in 3D that allow users to choose different viewpoints, this is also relatively easy to accomplish.

You can exploit the characteristics of how people view these images in order to add value to your product. For example, if you are looking at page 17 of an on-line article, it is likely that the next thing you will want to view is page 18, so the Java-based viewer can download page 18 in the background.

There are dozens of other forms of customization that could be done. Users of MovieLink want to view their favorite theaters first. Users of financial information services might want to highlight certain stocks. This kind of personalization can be done on the server side, but it is much more scalable if done on the browser side. By using Java (or programs like Java) the producer of the information can optimize the browser to display that information in more useful and effective ways.

But more subtly, you can also use Java to version your information. If you have some nice feature that makes your information more valuable to the user, you can also turn that feature off for some classes of users. You can offer professional access to your information (with page buffering), then offer access to the same information with the buffering turned off. Users with high willingness to pay pick the system that displays more quickly; users with low willingness to pay make do with the other one. This trick allows you to segment the market in very creative ways. Java-based viewers allow you to vary the ways in which consumers can access your information and give you a new tool to induce self-selection.

BUNDLING

Bundling is a special form of versioning in which two or more distinct products are offered as a package at a single price. A prominent example in the software industry is Microsoft Office, a product that bundles together a word processor, a spreadsheet, a database, and a presentation tool. Each of these products is also offered separately. This is what distinguishes bundling from tying, in which the individual products are offered only in the package.

Microsoft Office has been phenomenally successful, capturing over 90 percent of the market for office suites. There are several reasons for its success. First, the products are “guaranteed” to work well together: material can be cut and pasted or linked from one document to another with some confidence. Furthermore, the component parts use shared libraries so that the Office applications take up less disk space and work together more effectively than would be the case if you installed separate versions of the applications.

Even without these benefits lowing from integrating the different pieces of the bundle, bundling can be attractive and profitable. Since the price of the bundle is usually less than the sum of the component prices, a bundle of two products is effectively a way of offering one to customers who would buy the other product at a smaller incremental price than the stand-alone price. If each of two components sells for $70, and the bundle goes for $100, the incremental price of the second component is $30, less than the stand-alone price of $70. Dun & Bradstreet follows precisely this approach in selling detailed information about the consumer purchases of branded products, information obtained from scanner machines at the supermarket checkout counters and other retail locations. Manufacturers purchasing Dun & Bradstreet data in one geographic area get a discount on data obtained in other areas.

In considering bundling, you need to determine whether you would like to offer a targeted discount on one product to customers who would purchase the other product anyway. For example, if customers who value current-year information highly also are likely to value year-old information highly, it makes little sense to offer a discounted bundle containing information from both years. The on-line Wall Street Journal offers a discount to the subscribers of its paper version, since the people who already read the paper version get less value from the on-line version than nonsubscribers. But note that the Journal does not offer the paper subscribers a discount for the archives. They correctly realize that the on-line version is worth less to the paper subscribers, but the archives, if anything, are more attractive to paper readers, so there is no need to discount the price. Of course, you may be forced to offer such discounts if competitors do so as a way of attracting the most lucrative customers.

Dispersion in Customer Value

Bundling software applications can also allow you to significantly increase the value you extract from your customers when it reduces the dispersion in their willingness to pay. Let’s consider a simple example.

Table 3.3 illustrates Mark and Noah’s willingness to pay for two pieces of software. Mark works in the marketing department, where he uses a word processor most of the time and has occasional need for a spreadsheet. Noah works in accounting, where he mostly uses his spreadsheet but occasionally makes use of a word processor.

How should the software vendor price the word processor and spreadsheet to generate the most revenues? It is easy to see that there are only two sensible prices for their products: either $100 or $120. At $120 for each program, Mark will buy only the word processor, and Noah will buy only the spreadsheet. So, if each piece of software sells for $120 the vendor will earn total profits of $240. In contrast, if each program sells for $100, Mark and Noah will each buy both programs, and the software vendor makes $400. Clearly, pricing each product at $100 is the preferred strategy in this example.

But consider what happens if the software producer bundles the word processor and the spreadsheet together. Let’s make the conservative assumption that the willingness to pay for the bundle is just the sum of the willingnesses to pay for the components. In this case, Noah and Mark would each be willing to pay $220 for the “office suite,” resulting in a total revenue of $440 for the software vendor!

Bundling increases revenues in this example because the willingness to pay for the bundle is less dispersed than the willingness to pay for the components. This will happen when the consumers with a high willingness to pay for one component tend to have low willingness to pay for another component, that is, when there is a negative correlation across components in consumer value. Remember, if you set a flat price, you can only charge as much as the most reluctant purchaser is willing to pay—in our example, $100. So, if you do charge a flat price, techniques that reduce dispersion of willingness to pay will tend to increase revenues. Bundling can serve this function.

Table 3.3. Willingness to Pay for Software Applications

	Word Processor	Spreadsheet
Mark	$120	$100
Noah	$100	$120

Of course, if you can use differential pricing, you can charge users with high values high prices anyway, so dispersion is not as much of a concern. Dispersion only matters if you are forced to use flat prices.

Bundling can reduce dispersion even when consumer values are positively correlated simply because the sum of a large number of values will tend to be less dispersed than any single value. As long as values are not perfectly positively correlated, you will typically get some reduction in dispersion by bundling.

Other Reasons for Bundling

There are many other reasons to bundle information goods such as computer software. One important consideration is option value. A consumer may find Microsoft Office an attractive purchase even if she doesn’t currently use a spreadsheet, since she might use a spreadsheet in the future.

If the consumer does decide to use a spreadsheet in the future, she will naturally choose the one that is “free” in the Microsoft Office bundle. Of course, the spreadsheet really isn’t free—the consumer paid for it when she purchased the bundle—but it does have a zero incremental cost once the bundle has been purchased.

Microsoft has exploited this sort of pricing in another interesting way. Back in the days when Microsoft faced competition in the operating systems market, it licensed DOS to clone manufacturers using a sliding scale that depended on the number of machines that the manufacturer produced, whether or not DOS was installed on them. This was called a per-processor license, because Microsoft’s OEM customers paid royalties to Microsoft for their DOS license depending on how many processors (machines) they sold. Note that the pricing was based on the production of machines, not on the number of machines in which DOS was installed. This meant that when the manufacturers installed an OS on the machine prior to shipping, the natural choice was DOS, since it had already been paid for by virtue of the licensing policy. DOS had zero incremental cost of installation, making it very attractive relative to the competition. The Justice Department challenged this pricing structure in 1994, and Microsoft agreed to abandon it; see our discussion in Chapter 10.

Information Bundles

Information is commonly sold in bundles now: magazines are bundles of articles, and subscriptions are bundles of magazines. This makes good sense: there is often considerable variation in how much users would be willing to pay for different articles in a magazine. One reader of the Economist may read only the articles about America but get a lot of value out of them. Another reader might read only the articles about Europe and feel the same way about her choice. By bundling these articles together, the Economist reduces the dispersion in willingness to pay for the collection of articles it sells. If the publisher had to sell each article on a pay-per-read basis, it would likely get significantly less revenue.

The same thing holds true for subscriptions. Owing to lack of interest or lack of time, you probably don’t read every issue of every magazine to which you subscribe. But you may still be willing to pay for the subscriptions because there are some articles in some issues that are valuable to you.

Certainly there are other reasons to bundle articles together into issues. There are economies of scale in printing, binding, shipping, and marketing. But even if many of these economies of scale were reduced—as they are for electronic publications—it may still pay to sell articles bundled together into subscriptions for just the reason described above: bundling will generally reduce the dispersion in willingness to pay, thereby enhancing revenue.

Customized Bundles

Information technology allows for some interesting twists on bundling. Currently, pop music is sold on CDs, which are typically bundles of individual songs. This is in accord with the rationale described above: people have different favorites, and bundling the songs together reduces dispersion in willingness to pay.

Technologies are now becoming available that will allow users to create their own CDs. MusicMaker allows you to choose from its database of 30,000 different tracks and create your own customized CD for less than $20. This is a great example of mass customization of information.

Another example is the so-called “personalized newspaper.” Here the user chooses a set of categories and a software agent assembles articles in those categories for delivery. This technology allows a user to create his or her own bundle of articles. Customized textbooks are also available.

How should such products be priced? To get a hint, look back at the example of Noah and Mark that we used to introduce the idea of bundling. Think of their willingness to pay for the software packages as willingness to pay for individual pieces of music that can be laid down on a personalized CD. (To get the right order of magnitude, think of the numbers as denoting cents rather than dollars.) If we priced each piece of music at $1, we would make $4 off of Noah and Mark. But suppose that we used nonlinear pricing and said that the first song you chose would cost $1.20 and each additional song would cost $1. In this case, both Noah and Mark would chose both songs, and we would end up with $4.40, just as if we bundled the products ourselves. This example shows that quantity discounts can play the same role as bundling. In fact, quantity discounts can be thought of as a generalized form of bundling and are useful for much the same reasons that bundling is useful. MusicMaker, the custom CD site mentioned above, uses just this form of quantity discounts: the minimum order is five songs for $9.95, with additional songs costing only $1 each.

PROMOTIONAL PRICING

Promotional pricing is a commonly used marketing strategy. Promotions take many forms: firms can use sales (limited-time reductions in price), coupons (which require the consumer to bring in pieces of paper that then allow them discounts), rebates (in which consumers must mail in a piece of paper to get some money back), and so on. All of these marketing techniques have one feature in common: they impose some inconvenience cost on the consumer.

In the case of sales, the consumer has to watch for the sales to occur. In the case of coupons, the consumer has to clip the coupon and remember to take it to the store. In the case of rebates, the consumer has to remember to fill out the rebate form and mail it in.

Between 80 and 90 percent of adults use coupons at one time or another, but only 2 percent of all coupons produced are ever redeemed. This suggests that people use coupons very selectively: some people use them for food, others for computer software. Clearly, coupons wouldn’t be a worthwhile marketing strategy if everybody used them. If everybody used them, the seller may as well cut the price and eliminate the cost of dealing with the coupons.

The coupons are worthwhile only if they segment the market. A coupon says, “I’m a price-sensitive consumer. You know that’s true since I went to all this trouble to collect the coupons.” Economists say that a coupon is a credible signal of willingness to pay. It is “credible” because only people who have a low willingness to pay tend to use coupons.

The same sort of thing goes for sales. The people who show up when you have a temporary price reduction are the people who find it worthwhile to watch for sales. These tend to be people who are price sensitive. People who buy even when your price is high aren’t very price sensitive, almost by definition. Sales and other forms of promotions are often ways to segment the market into price-sensitive and nonprice-sensitive components.

What does this have to do with information pricing? Well, suppose that information technology lowers search costs so that everyone can “costlessly” find the lowest price. This means that sales are no longer a very good way to segment the market. Or suppose that software agents can costlessly search the net for centsoff coupons. In this case, the coupons serve no useful function.

Promotions of this sort are useful only if they are costly to the consumer, because it is only by imposing costs that they can identify price-sensitive consumers. If the computer costlessly does the searching or coupon clipping, the marketing technique loses its function.

Bargain Finder is a case in point. Brian Krulwich, a researcher at Andersen Consulting, designed a little program that would search online CD stores for the best prices for music CDs. Bargain Finder was an immediate hit on the Web: it had more than 100,000 uses in the first two months it was available. But after a few months of use, three of the eight stores that Bargain Finder searched decided to prevent it from accessing their price lists.

Remember the first lesson in Chapter 2? Avoid commoditization. The on-line CD stores didn’t want to compete on price alone. They wanted to compete on service and value added. By allowing Bargain Finder to look only at one dimension of what the stores offered, they ended up commoditizing their product.

This sort of commoditization may be hard to avoid with Internet shopping. Services like PriceScan compile lists of advertised prices for computer equipment and consumer electronics. This is a great service for consumers, but it will make the retailing market even more cutthroat than it already is.

LESSONS

• Adjust the characteristics of your information products to emphasize differences in what customers value. You can offer different versions that have differential appeal to different groups, adjusting the price if necessary to sort the consumers.
• You can version your products along a variety of dimensions. Delay, user interfaces, image resolution, speed of operation, format, capability, features, comprehensiveness, annoyance, and support are some examples.
• Add value to on-line information to differentiate it from hard copy. Don’t just put text on-line—do something with it that you can’t do with the print version. At the very least, make it searchable and use links for cross-references.
• If your market segments naturally, design your information product line to match. For example, if there are professional and amateur users, offer versions that are designed and priced to appeal to each of these market segments.
• If your market does not segment naturally, choose three versions (just like Goldilocks). If you don’t know how many segments there are, three versions is a good default choice. Plan to make most of your money off the middle version.
• Control the browser. Controlling the browser by using Java or similar technologies helps you modify the way you display your information, which helps you version and price your market.
• Bundling makes sense if it reduces variation in willingness to pay. Combining complementary goods increases revenue if it decreases the variation across customers in their willingness to pay.
• Nonlinear pricing can also be used to let consumers build their own bundles. Quantity discounts can increase usage and revenues at the same time.
• Promotional pricing makes sense if it helps you segment the market. Design your promotions to elicit different responses from different types of customers. Such targeted promotions help support versioning.