How to succeed in advertising

"If you're makin' me wait, if you're leadin' me on, I need to know."
--Tom Petty, "I Need to Know"

Over the next few years, there is likely to be a huge rise in performance-based advertising on the Internet. Historically, Internet ads have been sold on an impression-based metric known as "CPM." CPM, which stands for cost per thousand, represents how much an advertiser will pay to get in front of 1,000 sets of eyeballs. Some have suggested that price per click-through is a better model, but why stop there?

The proper model is performance-based advertising. In this world, an advertiser will pay only for an ad if and when the referral leads to a customer sale. Not impressions, not click-throughs, but honest-to-goodness revenues.

To be sure, this idea is not new. Businesses already accustomed to "origination fees" have flourished on the Web. This includes everything from long distance service and credit cards to mortgages. In each of these cases, content-based sites help customers identify a personally appropriate credit card or mortgage, and the matchmaker is then paid a certain bounty for finding the consumer. GetSmart is a great example of a business built around this concept. Affiliate programs are another form of success-based advertising. Amazon.com has no fewer than 35,000 affiliate sites that receive about 15 percent of the proceeds from book sales directed to Amazon.

Another form of success-based ad models are pay-for-performance banner ads. Internet advertising service DoubleClick believes in this market and has launched an interesting initiative known as DoubleClick Direct. However, new media in general have shied away from this market, with many executives dismissing the likelihood of adoption. I think the problem is that success-based programs are way too close to direct marketing on the marketing continuum. For most media types, comparing traditional impression-based advertising to direct marketing is like comparing Paris to Las Vegas. Performance-based advertising would seemingly jeopardize the essence of the experience.

Despite this snobbish reluctance, I see four reasons why success-based ads will flourish on the Web. The first and primary reason is that customers want it. With this model, the advertiser can assure that his or her ad programs are economically sound. You simply add the "bounty" fee to the cost of goods sold and you can predict the margin on each sale.

This is simply not the case with traditional advertising. The model shifts the burden of evaluating the quality of each ad program to the content company. If an ad doesn't have good responsiveness, the seller loses nothing, although the content company has used up inventory. Therefore, it is up to the content company to determine exactly which ads are likely to produce results. This shift of responsibility is quite palatable to the advertisers, and let us not forget the adage: The customer is always right.

The second reason to believe in this emergence can be found in the evolution of the last media revolution--cable television. If you watch one of the less popular shows on cable you will begIn to notice an overwhelming amount of 1-800 ads. This is because these programs have excess inventory, and the best way to maximize the revenue is success-based ad programs. Want a Topsy Tail? Need a citrus slicer? Want to add Super Hits of the '70s to your music collection? Ever notice the plethora of 1-800 ads on the financial news network CNBC? To highlight the issue of ad choice responsibility, have you noticed how often CNBC plugs Evita Nelson's MoneyLetter? That thing must be selling like hotcakes!

The third reason these programs will flourish is proliferation of unsold ad space on the Internet. In its most recent quarter, Internet bellwether Yahoo reported record revenues and earnings. (See related story) Interestingly enough, page views grew faster than revenues. This results in lower and lower average revenue per page view, which is also known as the "effective CPM." Internet advertising continues to grow at a healthy rate. However, and an increasing number of pages remain unused with respect to advertising.

Some suggest that this inventory glut will lead to price erosion. I think that there is another cure in performance-based advertising. What better way to bring economic equilibrium to the ad market than to fill the marginal page view with a performance-based ad?

One unique aspect of success-based advertising is the opportunity for marketers and financiers to conduct a more thorough quantitative analysis. How much should you pay for a referenced customer? You might think this is an easy question. Simply treat the bounty fee as an outsourced marketing expense. In this case, you would be willing to pay the same percentage as your sales and marketing expense ratio. If you normally expect to spend 12 percent of sales on marketing, then you would be willing to pay a 12 percent "sales charge" to anyone who brings you a guaranteed sale. However, life on the Internet is not this simple, and many retailers are willing to pay much more than this for a lead.

In markets like telecommunications, cable, and credit cards--where customers sign up to subscription services and churn is measurable--many companies treat such outlays as "customer acquisition costs." These expenditures, which are sometimes called origination fees, are compared not to the imminent sale price, but rather the lifetime value of the customer. That lifetime value is equal to the future cash flows (not revenue) expected over the life cycle of the customer, discounted back by a reasonable rate. What we are really doing is treating the customer acquisition as an investment and the lifetime cash flows of the customer as the yield on that investment.

If you expect the customer to stay forever, you can simply treat the cash flow as perpetuity and divide this periodic influx by the discount rate to arrive at customer value. So let's say that a telephone company expects to earn $40 a month in revenue from a customer and $6 a month in cash flow. If they can keep this customer for life, the value of that customer will be equal to $72 divided by the discount rate. If we use 15 percent, this equals $480. Therefore, this telco might be interested in spending up to $480 to acquire this customer, as any amount below that level would be considered a value-creating investment.

Many Internet retailers are using this formula to justify expenditures on the Web. This may make sense, but you must be very careful with your math. In the above example, we assumed zero churn, which may not make sense for a CD store on the Web. You must also account for churn in the following way:

Let's say you expect to lose 5 percent of your customer base a year. With this model, the average life of a customer equals 1/0.05, or 20 years. Here, instead of using a perpetuity model, you discount the life costs back over 20 years. For very low churns, this has little effect (at 5 percent, the $480 above only falls to $450). However, the value of the customer decreases exponentially as churn increases. At 10 percent churn, this figure falls to $361 and at 25 percent churn all the way to $205. Internet retailers with little track record will have a hard time calculating churn, but a mistake could be quite costly.

Another problem presented to Internet retailers is that of repeat customers. At present, most affiliate programs pay for all customer leads, yet only new leads really make sense with the above model. To compensate for this, it is important to discount the approximated customer value by the percentage of new customers vis-à-vis old ones. For instance, a $300 customer is only worth $210 if 30 percent of the referrals are from repeat buyers. As the Web matures, this percentage is likely to rise, which should push down the amount people are willing to pay for customer acquisition. One last concern may be efficiency of the Web itself. Price comparisons are very easy on the Web, which could lead to increased competition and lower profitability.

I have heard stories that CD retailers are paying as much as $40 for a customer lead. If we assume that the average consumer spends $100 a year on CDs (at a given retailer), has a modest 10 percent churn, and produces a 10 percent cash flow margin, then this customer is worth about $50. Of course, this assumes that there are no repeat buyers. Throw in 25 percent repeat buyers, and this falls to $37.50--and below the magical $40. So while the current success-based advertising expenditures may not be totally out of line, they are dangerously close to the edge.

One could even proclaim these fees as acceptable. If an Internet retailer can attract enough viewers, then they themselves may be able to become bounty hunters. Consider Amazon.com: With 2.5 million users and a firm understanding of their interests, the bookseller is in a good position to produce leads for others. Looking for a book on real (not virtual) surfing? Don't be surprised if Amazon offers to send you to a site that will sell you a surfboard or book a trip to Hawaii. Now what if Amazon can generate 5 percent of sales for each trip and surfboard sold? In this case, it might be willing to treat books as a loss leader and make it up on 100 percent gross margin referrals. Think about what effect this could have on the book business.

There is no question in my mind that success-based models will eventually rule on the Web. This is not to say that impression ads will go away, but this remarkably efficient advertising will serve an increasingly larger portion of this powerful new medium. Every industry from product sales to doctors, lawyers, and other complicated service providers will eventually use ad programs such as this. What's more, these programs could have very interesting effects on the competitiveness of each industry and potentially neighboring ones.

Of course, if you extrapolate this vision, you see a world that looks more like a flea market than Hollywood. Have you ever stepped onto a public beach outside a Mexican vacation resort? How do you say no dinero in Webspeak?

J. William Gurley 1997-8. All rights reserved. The information contained herein has been obtained from sources believed to be reliable but is not necessarily complete, and its accuracy cannot be guaranteed. Any opinions expressed herein are subject to change without notice. The author is a general partner of Hummer Winblad Venture Partners (HWVP). HWVP and its affiliated companies and/or individuals may, from time to time, have positions in the securities discussed herein. Above the Crowd is a service mark of J. William Gurley.