Affiliate Programs

by Jakob Nielsen on July 12, 1999

Sidebar to the section on Web marketing in Jakob Nielsen 's column on Web research .

See Also: Sidebar on reverse-affiliate programs (a new Web business model)

Many readers have asked me to define "affiliates programs," so apparently they are not as well-known as I thought. This could be one more reason they are not used widely enough, even though they are one of the most effective Web marketing tools.

An affiliate program is a way for sites to pay for incoming traffic: when site A links to site B, there is some mechanism for site B to pay a referral fee to site A , depending on who follows the link and how valuable they are to site B. Site A is called an "affiliate" of site B, because it is associated with site B and uses it to provide a service to its users.

The most famous affiliates program was introduced by Amazon.com with the launch of the Amazon Associates in July 1996. Amazon currently has about a quarter million sites in their affiliates program. In the case of Amazon, the referral fees work as follows:

  • fees are only paid for users who buy books at Amazon during the visit that results after having followed a link from the affiliated site
  • the fee is usually 5% of the sales, though it can be more for certain books if the site linked directly to the product page for that book

So, for example, if you follow this link to Amazon.com and buy a book there, I will get 5% of the price of the book.

In the case of Amazon, the affiliated site does not get anything if the user doesn't buy during the first visit but decides to return and shop later. Other affiliates programs are more advanced and pay based on the life-time value of the referred customer and not just based on his or her initial purchase. It obviously requires more programming to keep track of users over time, but all marketing theory certainly tells us that life-time value is more important than individual sales, so hopefully we will get more affiliates programs to take the long view in the future.

Other examples: Autoweb.com (about 5,000 affiliate sites) pays $5 for every referred user who ends up posting a used-car ad and CarPrices.com (about 9,000 affiliate sites) pays $3 for every referred user who asks for a price quote on a new car (whether or not they end up buying it — which does not seem to be tracked yet).

Strengths of Affiliates Programs

Affiliates programs are a natural for the Web because they provide a way to let value flow along the links .

Value always flows forward along a link since users follow links and thus provide additional business to the destination site. But how about the referring site? Without an affiliates program, the only way one gets compensated for providing an outbound link is very indirect: a well-chosen link is a service to the users who will certainly appreciate being pointed to something useful. Thus, a site that links well will have a higher chance of being re-visited in the future because the users will see it as a valuable resource and give it some percentage of the credit for the benefits they derived from jumping to the destination site. The success of Yahoo proves this theory: in the beginning, Yahoo was nothing but a collection of links and users kept coming back.

Affiliate programs make the backwards value of linking explicit and thus compensate sites for contributing to a more cross-linked Web. They make the Web richer and more useful since even small sites can link to additional services that they do not have the resources to provide themselves. For example, I certainly don't want to stock books in my garage and yet I can help my readers by recommending the books I like in my areas of expertise. And not only can I recommend the books, the user can buy them with a click. And I get paid 5-15%.

From the perspective of the destination site, affiliates programs work for many reasons:

  • They provide qualified leads since the users who follow links from other sites are among the kind of people who are likely customers (assuming that the referring site has targeted content), they have a reason for coming, and they have been told what to expect (assuming that then referring site writes good explanatory links)
  • It is only necessary to pay for performance . Nothing needs to be paid if there is no click-through. Even better, the referral fee is typically based on the value derived from the visiting user, so nothing needs to be paid for a user who doesn't buy anything.

Because affiliates programs are based on pay-for-performance, it makes sense for the destination site to sign up many small sites as affiliates. After all, if the site doesn't send any traffic, there is no cost. And the computer keeps track of the referrals, so there is no need to manually negotiate complex deals or have other overhead that normally makes it expensive to have small partners.

Weaknesses of Affiliates Programs

Some of the strengths can turn into weaknesses: because sites now get paid to link, there is a temptation to add too many links to the Web. Every extra user interface element is one more thing for users to look at, so links should only be included when they in fact add value to the user.

Also, there is a temptation to link to things that you cannot fully recommend simply because you get paid to do so. A higher-level analysis says that the user's trust is the most valuable long-term proposition for a website and that it would be stupid to jeopardize the long-time survival of the site for a short-term profit. You can fool some of the people some of the time (a few users may buy a bad product because of your link), but you can't fool all of the people all of the time (pretty soon people stop taking your recommendations seriously; they may even stop using the site altogether). But even though strategic considerations dictate prudent use of linking, even for paid links, the person writing any individual page may still be led into temptation.

Some categories of products and services lend themselves more to affiliates programs than others. Books are the prime example because there are books about all special interest areas. Thus every site will have some books to recommend within its area of expertise. And there is value-add from the recommendations since the specialized site will know more about the books in its field than the more generic advice possible at a general site. Want a tourist guide to London? You can't find the best one at the bookstore site, but a travel site could have a shortlist of the best 3-4 guidebooks for different kinds of travelers.

Unfortunately, most categories would only work with a more targeted set of affiliates. Take, for example, plumbing supplies. I would have no credibility if I started recommending a source of plumbing supplies and the users of my site would not constitute qualified leads for a site selling plumbing supplies. But specialized sites for professional plumbers or for do-it-yourself types could join such an affiliates program. In general, there would only be a few sites that sold the supplies but many sites that covered relevant topics and that could turn into affiliates.

Thus, appropriate use of affiliates programs limits them to sites that actually have some form of affinity. Current practice on the Web is not quite so clean, unfortunately.

International affiliates programs have problems due to currency conversion (it's no good to receive a check for ten Australian dollars if your bank charges twenty dollars to convert it into your own currency). It is also common for users to prefer to receive service from local sites instead of overseas ones. For example, I receive much email from European readers saying that they would like to buy through my links to Amazon but have the books shipped from Amazon.co.uk (one of Amazon's European sites). Currently, the only way to do this would be for the referring site to register for separate affiliates programs with each of the regional suppliers and have separate links to each of them (click here to buy this in the United States, but click here to buy it in Europe or here to buy it in Asia - and sorry, I was too lazy to support Latin America). Obviously not a good solution: clutters up the UI and adds a substantial burden to the referring site. The destination site needs to be able to handle international customers and transition transparently between its different services, ensuring appropriate referral fees no matter which one the originating site links to.

A final weakness is the lack of transparency and difficulty in finding and setting up affiliates programs. It is currently difficult to track down which sites offer affiliates programs, and many sites still don't have one. It is also difficult to judge the terms and conditions and to assess which programs are the most fair to the referring site. There are no standards for referral fees or easy ways of comparing different programs.

There is still substantial overhead in signing up for affiliates programs. This is supposed to be a fully computerized process, but people still have to fill in forms manually every time they want to register for a new program. There is no centralized way of joining and managing affiliates programs. The overhead makes it infeasible for a site to join an affiliates program for a single link or a small deal. The destination site also has administrative overhead in mailing out many small checks since each destination site has to pay each referring site on its own.

Future of Affiliates Programs

Less administrative overhead : we will get a central site that manages affiliates programs. This will make it easier for destination sites to establish a program and easier for referring sites to join. It will be feasible to join for a single link since you can simply use your universal affiliates identifier. And the payments will be centralized so that payments of a single cent become feasible (because you don't have to cut checks).

Affiliates programs become integrated with micropayment schemes . Thus, they will be used not just for the sales of products but also for the sales of content and services. If, for example, a newspaper charges 5 cents to read a certain article with an in-depth analysis of some issue, then the referring site may get a cent each time it links a reader to that article.

Advanced programs will support multi-level referral fees . Currently, only the site that made the final link to the money-generating transaction gets any referral fee. But how about the site that guided the user to the referring site? It gets nothing. Example: I like a book and recommend it. If I link directly to the bookstore site, I get a referral fee if you buy it. But let's say that I am a good Web designer and want to link you to additional views and reviews regarding the book and not just present my own review. Many readers would follow the links to the other reviews, each of which obviously contains a "buy this book" link. Some readers will be good Netizens and return back to the original site that got them interested in the book and buy it from there. But many others will either not understand the nature of affiliates programs or be lazy or might not even remember which review they saw first. So they will simply buy the book from the last review they read, thus giving it the entire referral fee. A better approach is to generalize the idea of backward-flowing link value and proportion some of the referral fee back to the site that linked to the site that linked to the sale.

Not all affiliates programs should be based on the money economy . As long as hard cash is the only form of payment, it is only feasible to compensate for links that generate direct sales. But there are also benefits from links that promote user loyalty or repeat traffic. Let's say that I want to refer to a stock quote. Whatever stock quote service I pick to supply the data would get an enhanced presence on the Web and thus some small added prestige and brand recognition. But even with micropayments they might not be able to afford a cent every time I served up one of their quotes.

Instead, payments could be in the form of scrip of some kind, including status in a frequent-user program . Maybe if I link enough to the stock-quote service they will award me with the right to get real-time quotes instead of delayed quotes, or I will be allowed to access some of their archives that they don't make available to the general public.

We might see shop-bot affiliate programs that allow the referring site to unbundle the recommendations of products and suppliers. Currently, I have to recommend a book and a bookseller in the same link. But in the future, it should be possible to recommend the book and link to a shopbot that would add a recommendation of where to buy the book. The resulting sales commission should be split in some fashion between the recommender of the product and the recommender of the provider selling that product. An unbundling of product recommendation and supplier recommendation would also handle the problem with referring international users to a vendor in their own country.

A final trend would be increased focus on the life-time value of a referred customer instead of the narrow focus on immediate sales. For example, if I link a user to a site today but he or she doesn't buy anything until tomorrow, I should still get a referral fee. We need to think hard about ways of tracking and adjusting the value of a user over time. For example, it might be reasonable to have a higher referral rate for the users' first purchases and a lower rate for subsequent purchases. Quite similar to sales commissions in the real world.

History of Affiliates Programs

Amazon Associates may be the most famous, but they were not the first affiliates program. An early sports site, www.S2.com, used referral fees in the summer of 1994. Even if (in the spirit of those early days) they based their statistics on hit counts. Jeff Scott, now the Internet Architect for NETdelivery Corporation , tells the story:

Myself and a co-author built the first Aspen Skiing Company web site for 4 ski areas, an outdoor sandal company, 2 ski retailer sites and 4 ski manufacturer sites and a bunch of smaller ski industry vendors. This all in the spring and summer of 1994 .

We were working in the Small Works Lab of Bill Joy's in Aspen, Colorado. In this frenzy of development we also had been working on a venture going under the auspices of www.S2.com. S2 stood for SportSource. Our attention was promoted with the affiliate idea in mind to drive viewers respectively between all of these sites in some fashion. We had a raw attempt at banner placement and even a rotating ad selection script that drew graphics based on where the visitor had linked from. S2 was written up as the first and foremost Action Sports Site on the web by several authors that produced soft-cover books with lists of web sites. (Do you remember that? Books that had actual site content in them with screen snapshots and site information... Talk about out-of-date fast!)

Anyway, our model was a success for driving visitors between content of the involved and participating sites. The bonus for us was the "borrowed landscape" idea that, since we were developing the content for all these sports related sites, if we shared that content in a general area (S2.com) and it had a different appeal than the directness of a particular vendor, our revenue went up since all participating sites were paying for the click through (hits at that time), for the traffic. We had a great model for a while, site hosting, development, maintenance, internet consulting, ad development and click through revenue all from 2 dozen sites or more.

Wow, I write this and remember the excitement and frenzy of that summer... So much happening in the evolvement of the industry and now it is all second nature to so many.

As a side note, let me add that I too remember the frenzy of 1994. That's when the Web was truly exploding, growing almost 2,000% from 600 sites to 12,000 sites in a year. And we made the rules ourselves as we pushed projects through in weeks that now take several months.


Share this article: Twitter | LinkedIn | Google+ | Email