Advertising vs. Sales - Why Free is Very Expensive

In the years 2006-2008, Google earned, respectively, $10.5BN, $16.4BN and $21.1BN in advertising revenues. With net income of $3-4BN each of those three years, it is hardly surprising that "free supported by advertising" became the model of choice for many an aspiring startup, as well as many non-startup businesses. During the latest mini-bubble of venture investing, it was nearly impossible for an online company to get an investor's meeting if your business plan was not based on the presumption that your services would be free and paid for by advertising.

In the 12-18 months since the investing bubble collapsed, and investors went into a deep freeze not that different from what occurred in 2001-2002, the value of the advertising model has been severely questioned. Whether using CPC (cost per click) or CPM (cost per thousand impressions), the rate for each of these has gone down. In some cases it actually has gone down; in others, the assumptions of CPMs and impressions (or CPC and click-rates) was dramatically overstated. In January, AdAge reported that a study from the Interactive Advertising Bureau and Bain Capital was showing CPM rates on ad networks of $0.60 to $1.10, with their own direct sales of $6-10. The report indicated that (a) those 10x multiples for direct sales were likely overstated and (b) online publishers were only able to unload around 1/3 of their ad space inventory directly anyways.

Assuming the high CPM of even $1.00, an online services company that wants, say, a small $1MM in revenue, needs to have 1BN impressions per year ($1.00 CPM means 1,000 impressions makes $1.00, times 1MM to make 1BN impressions). Over 12 months, that is 83MM impressions per month. If we assume that each unique visitor visits once per week, and sees 10 pages, and each page has 3 ads, then each unique visitor sees 120 impressions. For 83MM impressions per month, this company needs nearly 700,000 unique visitors each and every month, just to clear $1MM per year in sales. Of course, these numbers change if:

• Like most sites, visitors see fewer pages per visit, reducing revenue.
• Visitors visit more or less frequently than weekly.
• The site has some strong unique niche, like ClubMom, that can command higher CPMs.
• The site does not have a strong unique niche, and thus gets much lower CPMs.
• The site does not have 700,000 unique visitors per month, a fairly high number.

Essentially, just to clear $1MM in annual revenues, the site needs to have many regular, loyal, high-usage visitors.

Conversely, if the site managed to sell to each of these unique visitors at just $5 per month (the model used by backup services, like Mozy), then it would generate $3.5MM per month in sales with the same number of visitors. Those kinds of numbers could even attract professional investors. Clearly, if you can sell instead of using a three-way (service provider, user, advertiser) model, it is far more revenue-intense and, likely, profitable. Of course, paying customers demand a whole different level of service, but $42MM per year can pay for a lot of customer service.

Two interesting case studies came out in the last several months that illustrate this concept.

• YouTube vs. iPhone vs. Kindle: On June 15, 2009, Business Insider published a report by Goldman Sachs analyst James Mitchell. In it, he showed how YouTube revenues were essentially on par with iPhone App Store and Amazon Kindle eBook sales in 2008, while eBooks should double YouTube revenues by 2010, while iPhone App Store revenues will be quadruple YouTube revenues in the same year. It is important to note that YouTube has 71MM unique monthly visitors, according to its own information, and has been around for over 4 years. The iPhone, by contrast, has been around for 2 years, with the App Store somewhat less, and certainly nowhere near open enough to encourage heavy usage until into 2008. The Kindle launched a few months after the iPhone, and is thus even younger. Essentially, one of the hottest properties on the Internet, with an enormous number of visitors, cannot match direct e-sales products half its age. Interestingly, both iPhone Apps and Kindle eBooks are of the same essential format as YouTube videos: 3rd-party content. In the case of YouTube, it is user-generated videos; in the case of the App Store, it is overwhelmingly 3rd-party developers; in the case of the Kindle, it is 100% publisher-author provided books. If we compare users, as stated, YouTube has 71MM unique monthly visitors, while the total number of iPhones sold by Apple is slightly over 20MM, and the number of Kindles is almost certainly below 1MM. With fewer than 1/3 the users, and half the life, iPhone App Store and Kindle eBooks are handily putting YouTube in its place. Finally, notice that we are discussing revenues, not profits. Video is notoriously expensive. It requires a lot of bandwidth, even with advanced encoding such as mp4, and a lot of processing and memory. There is no doubt that the fixed and variable costs for YouTube far exceed those of the App Store and Kindle eBooks.
• iPhone App advertising: A great company, based in New York, is Pinchmedia. Pinchmedia is one of several companies (another excellent one is Medialets) that provide in-application advertising and metrics (akin to analytics such as Google's Analytics) for iPhone applications. Earlier this year, Greg Yardley, the founder and CEO of Pinchmedia released a fascinating study based on the analytics his service has provided in 30MM downloaded applications. The report is available here, and is worth reading. It is a masterpiece of data analysis. The net result of his analysis is the following. Free applications, on average, are used 6.6 times more than paid ones. Since the minimum price is $0.99, and Apple keeps 30%, at minimum a paid application returns $0.70 per user. But since free applications are used at most 80 times per user, based on Pinchmedia's analytics, then to make advertising worth it, 80 usages would have to generate at least $0.70 per user. $0.70 in 80 sessions is an equivalent of $8.75 CPM, if one ad is shown per session. Yet, Pinchmedia is seeing CPMs more on the order of $0.50 to, at the most, $2.00. Essentially, unless you have a very high usage application, one in which there are lots of advertising opportunities, i.e. users stay in the application a lot and return to the application far in excess of the average 80 usages, or, if the application has a much stronger than 6.6:1 ratio of free:paid, because of the inherent nature of the application, you are unlikely to recoup anything near the revenue of selling the application.

Summarizing, selling directly to the end-user beneficiary is far normally much better for your revenue than advertising, Google's search business notwithstanding. There are instances in which advertising-supported free services do, indeed, make a lot of sense, and need to be examined on a case-by-case basis. This brings back to mind the late 90s Internet bubble, when revenues did not matter, all that investors and managers looked at were "new economy" metrics, primarily "eyeballs." Once again, the market has reasserted itself. There are businesses wherein advertising-based works on the Internet, just as it does in the real world. Clearly, however, in most cases, free can be very expensive indeed.