Nearly 34 million people visit personals sites each year. The industry is dominated by two major players. Match.com is the biggest site with nearly 6 million visitors per month while Yahoo, at 3.4 million, is the second most popular site, leveraging its large population of web directory and email users. Following well behind these two are a dozen or so fairly evenly trafficked sites (see chart: Major Players in Online Dating).
The business models for most of these sites are some combination of free and paid access. For all the sites browsing the ads is free. However users have to register and pay between $20 and $25 in order to post ads or respond to ads or both. Over 93% of online dating subscription prices are in the $5-50 range. Both Match.com and Yahoo charge to post an ad and respond to ads. Lavalife (the 10th most trafficked site) is free to post ads but requires users buy credits in order to contact other users. Overall 98% of purchases at online dating sites are subscriptions. 54% of those subscriptions are monthly, 22% annual.
The user population of online dating sites is about 60% male, 40% female. The ages of users range from 18 to 80 with the largest percentage (32%) between 35 and 44 years of age. Followed by 24% for 25-35 year olds, 22% for 45-54 year olds and 11% each amongst 18 to 24 year olds and those 55 and over (see graph: age distribution). The fastest growing segment appears to be the 18-30 space. In the first quarter of 2002, fully half of the new users of Match.com have been under 30. This suggests that dating sites are no longer just for the desperate, those over 35 and still single, but have become another source for those with time and looks on their side to accelerate their dating or improve their dating efficiency.
Online Dating Growth
While spending on online content as a whole has increased, no category has increased more than online dating (see graph: Quarterly Growth of Consumer Spending by Category of Online Content). While the other categories such as business content and research have experienced steady linear growth, the growth curve for online dating is an exponential curve. Not only is it growing but the rate of growth quarter over quarter is also increasing. At its current rate by the time numbers are available for the rest of 2002 it will exceed business spending as the largest category of online spending. It is also the only category of online content that has significant consumption externalities. When purchasing any of business content, research, single player games, news, credit help, sports or online greeting cards, the number of other users of the product does not play more than a minor role in the purchasing decision.
There is also plenty of room for the industry to grow in both revenue and users. Soon these services will have video chat capabilities. People will overcome a lot of their negative perceptions of the practice as the number of users increase and many of these primarily US-based services will begin to expand internationally.
Network Characteristics of Dating Sites
It is already clear that online dating sites are showing the sort of rapid growth typical of networked industries, but the service also displays many of the other characteristics of these markets. Oz Shy in The Economics of Networked Industries identifies a few main characteristics of markets that behave as networks. These are:
- Required complementary products
- Compatibility and standards
- Consumption externalities
- Switching costs
- Significant Economies of scale
Required Complementary Products
Many products are not useful without a complementary product. Portable CD players are useless without headphones, and DVD players require DVDs to play. This is one of the factors that separate network goods from goods such as an apple which can be consumed by itself. Arguably online dating services could be the required complementary product to the Internet. With an Internet connection one needs content to look at, and certainly users of online dating services are required to have an Internet connection, so there is some complementarity here but this connection is somewhat weak. While CDs are needed for a CD player to be useful at all, dating services are not a requirement for the Internet to be useful. Further, even with the rapid growth in online personals, it still lags Internet growth and coverage so significantly that for at least the next little while the overall growth of the Internet will have a negligible effect on online dating when compared to the growth it will experience from existing Internet users.
Compatibility and Standards
With the requirement many networked goods have for complementary products comes the need for compatibility and standards. The benefits of making goods compatible with others are clear. Consumers can benefit by being able to choose from a number of headphone choices knowing that for the most part (except on airplanes) these headphones will plug into their CD player, stereo, and computer. From a headphone manufacturer’s perspective it is far cheaper to build for a standard jack than specifically for every device. While there is not a single standard that all online dating services are based on to the point where a profile or other stored information from one service can be seamlessly transferred to any other, several competing platforms are starting to emerge. Nerve.com, after the rapid growth of the personals section of its website, spun the service off into a separate company, Springstreet Networks, which provides a standardized shared match making service for other content providers such as salon.com, bust, boston.com and others. Match.com’s database is shared with licensees including msn.com, nytimes.com and villagevoice.com to name a few. Relationship Exchange is another network that powers the personals services behind.cupidjunction.com, and personals.canada.com amongst others. Even some of the smaller dating sites like people2people.com are licensing their databases to online newspaper sites like sfgate.com. The competition is not necessarily amongst the dating sites themselves but the networks they supply and pull from. However, Yahoo Personals is independent and not licensed to other parties. It is banking on its existing position as the top website on the Internet and its large pool of users of its other tools to provide a large enough pool of users. Lavalife also has its own proprietary system and is banking on its unique features and credit based rather than subscription based system. While Yahoo and Lavalife are staunchly independent, the advantages to smaller sites are obvious.
A personals site is only as valuable as its pool of profiles that match the user. If the user is unable to find people with the same interests, geography and whatever other characteristics determine a match, the user will not subscribe to the site for very long, nor be enticed to subscribe in order to contact a matching user. The larger the pool of people, the better the selection of people that may match, the more likely the user is to subscribe or purchase credits.
Consumption Externalities
The increase in value with an increase in users is the fourth characteristic of networked industries: consumption externalities. If there is one heterosexual person of each gender on one of these sites the possible number of connections is only one. However with five people of each gender the number of possible connections is 25. With 100 males and 100 females the number of possible connections is 10,000. Thus the total value of the network, as measured by the number of possible connections, increases by Vn = nm x nf where n is the number of users and m and f indicate male and female respectively.
While other networks such as fax machines increase at a faster rate: Vn = n(n-1)/2, these networks are homogenous in that every fax machine can connect to every other fax machine, where as on a heterosexual dating site each new user can only connect with the opposite population. However, even at this rate it is easy to see how the adoption rates for network products can be slow at first and gain tremendous momentum. This factor has been confirmed by increases in how long the average user now spends on personals sites. When the pool of users was smaller one year ago the average subscriber spent 9 hours per month on dating sites. According to a study by Jupiter-Media Metrix that number as of Oct 2002 is up to 13 hours a month. Site activity, and likely implied in this the value of the service, has increased with increased site traffic and registered users.
Switching Costs
A fourth characteristic of networked industries is a high switching cost or lock-in. Shapiro and Varian have separated these costs into five categories:
- Contracts
- Data Conversion
- Training and Learning
- Search Cost
- Loyalty Cost
With 54% of users on monthly contracts, and significant discounts for users who have subscribed annually, the economic loss from contract lock-in on dating sites is relatively minor. For instance, subscribing yearly to Yahoo.com is equal to the price of 4.5 months at the monthly rate; therefore any loss associated with leaving the service relative to the more flexible monthly option only exists for the first third of the year.
Re-entering data on a new site is a relatively minor switching cost. Filling out an initial profile on any of these sites is about 0.5 to 3 hours work depending on the time spent perfecting your answers. However, repeating this same process on another site will likely be less time consuming. While there is not a complete overlap, some of the questions are the same or similar. Even if it is completely different a loss of 3 hours is not a substantial cost compared to the over 150 hours a year the average user spends on these sites. Further, once a relationship has been established, it is carried out outside of the system, through phone, email, instant messenger or in person. None of these established relationships will have to change. However, some potential relationships may be lost in a move, if the relationship has not progressed to the level where it moves outside of the system and out of anonymity.
While data entry and contract lock-in costs are relatively small other switching costs are almost non-existent. In terms of learning costs, most of the major systems function in the same fashion or can be picked up relatively easily. The systems are all web-based and for the most part take advantage of standard web design and usability practices making the learning curves small. None of the major systems have any loyalty or rewards programs so these items are not lost. Finding out about the major other dating services is also not difficult. A simple web search on yahoo (powered by google.com) will provide a list of all the major competitors to yahoo personals (see Appendix A: yahoo search results). Further, the web is inundated with advertisements from dating services. Match.com alone is expected to spend $20 million in advertising this year. These costs are not any more than would exist with a non-networked good such as a toaster which would have comparable or even higher search, loyalty point loss, or training costs.
Significant Economies of Scale
Finally, network goods display significant economies of scale. Particularly with new technologies or informational goods, the cost of producing the first product. The first personal digital assistant (pda) or the first copy of a book is significant, as the amount of labor going into the research, development and revision process for each of these two goods can be quite large. While the first PDA may cost in the millions, the second can be reproduced for under $200. Likewise the second copy of a book might only cost $1 to produce. The marginal costs are quite low compared to the sunk costs to create the original product or service. The average cost curve thus declines sharply. With such goods the ideal production number is virtually infinite and thus a competitive equilibrium does not exist. Typically in these situations a dominant firm captures the majority of the market. This characteristic appears to hold for dating sites as well. The major costs for these sites are developing the code, database design and site flow: the software development, as well as attracting the initial customer to the site through advertising and marketing costs. System development costs can range from $1 to 9 million. However, after the site is built and the user has subscribed, the additional cost for each user of the system is relatively minor. Each additional user who averages 13 hours on the site per month, downloading 4 pages per minute at 30kb per page uses under 100mb in bandwidth. At a bandwidth cost of $3/gb this is a monthly cost of $0.30 for a user paying $20 per month. While there are other incremental data costs, such as server space and processing power, these are equally minor. Some customers may require assistance initially but much of this support comes through automatic or self-help customer service systems such as email auto-responders, documentation, frequently asked questions lists which involve a high initial development costs and a very low per use cost. Developing these can be considered part of the initial system development and only increase the steepness of the average cost curve. The low cost to manage these services is apparent at Nerve.com’s personals site which required just 7 people to run at a time when they had 350,000 users.
Across the five major characteristics of networked industries the dating industry strongly shares three of these characteristics but does not display a particular complementary product (besides the Internet itself) nor does it display high switching costs or lock-in. However, the industry has high consumption externalities, is beginning to standardize and build databases compatibility across many websites and displays significant economies of scale. All typical of networked industries.
Decision to Date Online
Even with the tremendous growth in online dating services it still has a reputation as being for those who can’t get a date by any normal means. The Wall Street Journal reports that couples who meet online tend to lie to their friends and relatives about where they met. Further, Cindy Hennessy, president of Match.com reports "Our greatest competition is not other online competitors, it is people's willingness to discuss their participation in online dating more openly". The number of people in their 20s joining online dating services is on the rise, it’s hard to believe that people in their 20s are becoming more desperate, so the attitudes towards online dating must be changing. However, the alternative case is simply one of network externalities. The more people who join, the better the pool of potential dates, the greater the value, the more people who join, and so on. This circular cycle of increasing value and users suggests that the main issue is not the attitudes towards online dating but the attitudes towards the people who date online. Online dating is lame because the people on these sites are lame. Akerlof, through his work on conformity and social decision making has provided a structure that can be adapted for this particular scenario. Assuming that all people would generally prefer to meet the people they form a relationship with in an off-line environment, the utility of dating can be expressed as:
U = aqoff – poff + d → Meets people offline (school, work, bars, events)
aqon – pon → Meets people online (dating sites, chat rooms, etc)
0 → Does not date
Where a is the added value to each method gained with each additional user. This can be due to both an increase in the pool of potential dates and a decrease in the stigma attached to that particular method of dating. In the above definition the assumption is that this is the same for all potential users. The number of people, or quantity using a particular method, is expressed by q and the total cost associated with any one service, or the price, is denoted by p. Since we’ve said that most people prefer to find partners offline, we express this added value by d. This is the fixed portion of people’s preference for offline dating irrespective of how many users each has or how attractive the pool. This could be due to the story associated with how they met their significant other, or other utility gains due to the randomness, fate, or luck associated with this method. If this were the same value for all users then there could only be two potential states. Either everyone dates online or everyone does not. However, in reality, it is highly unlikely that everyone will date online so d will be different for each person. The number of people at each level of d does not matter except that this will affect the rate of adoption growth. Each user will choose the service that yields the highest utility. As more people choose online dating, its value increases. This increase could overcome the weight of some other people’s d and thus cause more users to switch and the cycle continues.
Online dating is increasing so it is possible that this cycle is taking place. People are not changing their attitudes towards online dating (their d), however the value proposition due to the increasing pool of users is increasing to the point where it overcomes more and more people’s preference for offline dating.
Decision between Services
While it is important to understand the increases in online dating as a whole, another important question to answer is: which service will people choose and how will the competition for users affect consumer utility, price, the dating services’ profitability and societal welfare?
Shy’s model of duopoly competition selling differentiate brands to heterogeneous consumers in the computer hardware industry can be applied here. It starts out with a very similar definition of consumer utility to the one used above between online and offline dating methods. For this discussion, the assumption will be that there are two main competitors: Match.com and Yahoo, whose offerings are incompatible and differentiated. Given this a user who prefers Yahoo to Match.com will have a utility that looks like this:
U = aqyahoo – pyahoo → Subscribes to Yahoo Personals
aqmatch – pmatch – d → Subscribes to Match.com
Here again, q is the number of subscribers, p the price of a subscription and a the added utility with each additional subscriber. Different in this case is that d is the dis-utility associated with subscribing to Match.com if one prefers Yahoo. This could be because of differences in features, user interface or simply because it is tied to the same registration as Yahoo’s other features. Since Yahoo does have a fairly large user base the value of the differentiation between the two must be greater than the effect of the network size or d > an, where n is the total number of users of online dating services. However, Match.com could undercut Yahoo’s pricing and take the majority of its user base if it set its price is: pmatch < pyahoo – d – an. Essentially, if its price added together with the benefits of an increased network size if people switched is below Yahoo’s price by enough to overcome people’s preference for yahoo’s system. Both competitors will set their prices such that they will not be undercut. Shy goes into more details on the formula that determines this equilibrium but the end result is rather intuitive. The greater the differentiation (d) between the two services, the higher the price of each service, and the profitability of the service provider. Conversely, the more users value larger networks (pushing an higher), the lower each competitor will have to make their prices.
If Yahoo and Match.com chose to make their services compatible then this would lessen the race to build the larger user base in order to make their service the more valuable one. With this competitive strain removed either service is able to offer the same pool of users. Shy shows that this allows the services to charge more and earn higher profits. One of the reasons firms choose to make their goods compatible is the threat that the other firm will make theirs compatible. So why are these two services not compatible? If a user can go to Match.com and find all the users at Match.com and Yahoo personals, then Match.com is able to charge more. However, with web technologies, either firm can block the other firm from co-opting its database. The firms must agree to mutually share their databases. Many small sites do share databases or license other match making services. These sites do not have a large enough user base themselves to build a worthwhile service or simply find it faster or cheaper to license the services of others rather than build their own, trading the large fixed cost and low marginal cost for a flat average cost line. In Yahoo’s case, it has a large enough user base to make building a service that reaches critical mass possible so does not necessarily need to collude with Match.com. Further, since Match.com licenses its service to other sites including Yahoo arch-rival Microsoft, it may have reasons beyond the success of its dating service for staying independent. Yahoo and Microsoft both run portal sites (yahoo.com and msn.com), instant messengers, shopping sites, investment sites, and free email services to name just a few of the lines on which they compete. For Yahoo to share its database with Match.com could allow Microsoft to pluck these users for its other services.
As long as the major dating service sites compete using non-compatible services, prices will remain low and vary with the differentiation between the services (more differentiation making it less price competitive) and the value users place on network size (more value making it more price competitive). Further, due to the complicated competitive landscape, it is unlikely the largest services will become compatible with each other.
Conclusions
Online dating services are a large and rapidly growing market displaying several characteristics of networked industries. These sites require that another complimentary product (the Internet) be in place, have some emerging standards and compatibilities between various services, have high network externalities, and significant economies of scale, however they do not have much in terms of switching costs.
Online personals can be analyzed using theories developed for other networked industries. With this in mind, it is clear that these services will continue to grow rapidly as more users increase the value of the service. Further, the fight for users is fiercer than in non-networked industries so prices will be suppressed as long as the services continue to be incompatible.
Bibliography / End Notes
Patricia Busa, “Date line,” Texas Monthly May 1999: 26+.
Dominic Walsh, “Date game,” The Times 13 Aug 1998: 27+.
“Online Paid Content: U.S. Market Spending Report,” Online Publishers Association, Aug 2002: 4.
Steve Smith, “Online personals: Viagara for content's bottom line?” EContent Dec 2001: 64.
Oz Shy, The Economics of Networked Industries (New York: Cambridge University Press: 2001) 4.
Warren St. John, “Young, Single and Dating at Hyperspeed,” New York Times 21 Apr 2002: 9.1.
Verne Kopytoff, “DIGITAL DATING / Web sites find that singles and subscriptions make a perfect
match,” San Francisco Chronicle 19 Aug 2002: E.1.
“Online Paid Content: U.S. Market Spending Report,” Online Publishers Association, Aug 2002: 4.
Verne Kopytoff, “DIGITAL DATING / Web sites find that singles and subscriptions make a perfect
match,” San Francisco Chronicle 19 Aug 2002: E.1.
McLean, “Online personals attract singles” USAToday.com 21 Oct 2002: D.01
Warren St. John, “Young, Single and Dating at Hyperspeed,” New York Times 21 Apr 2002: 9.1.
Steve Smith, “Online personals: Viagara for content's bottom line?” EContent Dec 2001: 65.
Oz Shy, The Economics of Networked Industries (New York: Cambridge University Press: 2001) 2-6.
Steve Smith, “Online personals: Viagara for content's bottom line?” EContent Dec 2001: 64.
Max A. Kalehoff, “Top Three U.S. Properties Break Through 100 Million Visitor Mark, Reports
comScore Media Metrix” comScore Networks 26 Nov 2002, 29 Nov 2002
<http://www.comscore.com/news/mm2_release_112602.htm>
C. Shapiro, H. Varian, Information Rules: A Strategic Guide to the Network Economy (Boston: Harvard
Business School Press: 1999).
Lloyd Andrews, Avid Online Personals User, Personal interview, 28 Nov 2002.
Jacob Neilson, “Do Interface Standards Stifle Design Creativity?” Alertbox 22 Aug 1999, 22 Nov 2002
<http://www.useit.com/alertbox/990822.html>
Rebecca Flass, “Match.com searches for a match,” Adweek, 9 Sep 2002: 4.
Al Ries, Jack Trout, The 22 Immutable Laws of Marketing: Violate Them at Your Own Risk
(HarperBusiness: 1994) 44.
Steve Smith, “Online personals: Viagara for content's bottom line?” EContent Dec 2001: 64.
“Dedicated Servers” Canaca Internet Solution 26 Nov 2002.
<http://www.canaca.com/dedicatedservers.php>
Steve Smith, “Online personals: Viagara for content's bottom line?” EContent Dec 2001: 64.
Jennifer Sananow, “How Did You Meet? If Answer Is 'Online,' Couples Tend to Lie,” Wall Street
Journal Online 6 Nov 2002, 29 Nov 2002.
<http://online.wsj.com/article_email/0,,SB1036533515435168988,00.html>
G. Akerlof, “Social Distance and Social Decisions,” Econometrica 65 1997: 1005-1027.
Shy, The Economics of Networked Industries (New York: Cambridge University Press: 2001) 27-36.