While we applaud the recent MediaPost column about the next efficient frontier of display advertising, given our namesake and recent entry into the display market, it makes a few assumptions that deserve clarification.
Assumption 1: Transparency of Information
By way of background, the efficient frontier is the outcome of the Modern Portfolio Theory (MPT). This theory, developed by Harry Markowitz, assumes an efficient marketplace i.e. all investors have access to the same information at the same time. No marketplace is purely efficient but many come close. The U.S. stock market is one example and another is search marketing (although quality scores definitely do make it less efficient). The current display industry is a highly inefficient one. Most advertisers do not even know where all their ads are shown or have access to reporting tools to find them. Ad Exchanges promise to change all of this, with a public auctioning system to ensure both efficient pricing and a complete tracking solution. In this case, the issue of ads being unknowingly shown on porn sites is moot. If your ad is being shown on a porn sight you will know this right away and will have an efficient mechanism to stop it.
Assumption 2: All advertisers have the same goals.
There are brand advertisers, advertisers who focus purely on ROI, aggregators, lead generators who rely on the strength of other brands etc. Their goals are different. A brand advertiser might not want her ad seen in an adult site but what about an adult dating service? The issue then is not that ads are seen in inappropriate sites but more that the ads are seen in websites that are inappropriate for the advertisers goals. However, in a scenario where the entire funnel is tracked, one can optimize to a goal (Example: revenue) or a blend of goals (example: Impressions+ Revenue).
The key here is that, the ad exchanges have to provide easy control to advertisers to manage their campaigns to specific goals. We expect this to be the trend, as exchanges such as Adx and RMX have begun providing API access to marketers.
Assumption 3: There are multiple frontiers and multiple markets.
Just like there are stock markets and commodity markets which operate on different exchanges, there are multiple markets in online advertising. There is a display market and a search market. Within the display market there is and will be a branding marketplace on premium sites like NYT and Yahoo! Finance and non-premium sites like weatherbug. It does not mean that you cannot apply portfolio theory. In fact, you have to apply portfolio theory in different markets just like we select stocks within the asset allocation model. For example, an advertiser may have to determine how to efficiently spend their advertising dollars between Yahoo! Finance and NYT just as one picks stocks among large cap U.S. companies.
One other piece that I think deserves further clarification is the issue of retargeting. The idea is that a consumer who visited your website would be more valuable then one who saw the ad for the first time. There is more to it. If you have the capacity to track the user on display, search, mobile and organic advertising then further optimization is possible. In this case, the efficient frontiers would combine to become one curve that would answer the question "What would be the best way to allocate all my online advertising dollars and what type of returns can I expect from it?". Tell me one CMO wouldn't like the answer to that question!
At Efficient Frontier, we believe that the success of this approach depends on several factors.
We say, bring it on!
Dr. Siddharth Shah
Assumption 1: Transparency of Information
By way of background, the efficient frontier is the outcome of the Modern Portfolio Theory (MPT). This theory, developed by Harry Markowitz, assumes an efficient marketplace i.e. all investors have access to the same information at the same time. No marketplace is purely efficient but many come close. The U.S. stock market is one example and another is search marketing (although quality scores definitely do make it less efficient). The current display industry is a highly inefficient one. Most advertisers do not even know where all their ads are shown or have access to reporting tools to find them. Ad Exchanges promise to change all of this, with a public auctioning system to ensure both efficient pricing and a complete tracking solution. In this case, the issue of ads being unknowingly shown on porn sites is moot. If your ad is being shown on a porn sight you will know this right away and will have an efficient mechanism to stop it.
Assumption 2: All advertisers have the same goals.
There are brand advertisers, advertisers who focus purely on ROI, aggregators, lead generators who rely on the strength of other brands etc. Their goals are different. A brand advertiser might not want her ad seen in an adult site but what about an adult dating service? The issue then is not that ads are seen in inappropriate sites but more that the ads are seen in websites that are inappropriate for the advertisers goals. However, in a scenario where the entire funnel is tracked, one can optimize to a goal (Example: revenue) or a blend of goals (example: Impressions+ Revenue).
The key here is that, the ad exchanges have to provide easy control to advertisers to manage their campaigns to specific goals. We expect this to be the trend, as exchanges such as Adx and RMX have begun providing API access to marketers.
Assumption 3: There are multiple frontiers and multiple markets.
Just like there are stock markets and commodity markets which operate on different exchanges, there are multiple markets in online advertising. There is a display market and a search market. Within the display market there is and will be a branding marketplace on premium sites like NYT and Yahoo! Finance and non-premium sites like weatherbug. It does not mean that you cannot apply portfolio theory. In fact, you have to apply portfolio theory in different markets just like we select stocks within the asset allocation model. For example, an advertiser may have to determine how to efficiently spend their advertising dollars between Yahoo! Finance and NYT just as one picks stocks among large cap U.S. companies.
One other piece that I think deserves further clarification is the issue of retargeting. The idea is that a consumer who visited your website would be more valuable then one who saw the ad for the first time. There is more to it. If you have the capacity to track the user on display, search, mobile and organic advertising then further optimization is possible. In this case, the efficient frontiers would combine to become one curve that would answer the question "What would be the best way to allocate all my online advertising dollars and what type of returns can I expect from it?". Tell me one CMO wouldn't like the answer to that question!
At Efficient Frontier, we believe that the success of this approach depends on several factors.
- People know what they are doing: There will be advertisers who succeed with this approach and some who will fail. Its not because the approach is poor but because of their lack of expertise using it.
- The availability of technology to adopt this approach: From the advertiser side, the portfolio approach is highly automation and technology intensive. To truly leverage the long tail, one needs to be able to dynamically change bids on the ad exchanges over thousands of sites based on ROI and other pre-conversion metrics. This places huge demands on technology. In search you have one knob to turn i.e. the bid for a keyword, in display you have 10 or more including the bid, the site, time of day, the number of times to retarget etc.
- The ease of use provided by the ad exchanges: At present display exchanges are a dime a dozen. There will be maturation and consolidation in this space with the largest exchanges providing advertisers with the best controls for a more efficient marketplace winning out. In the end, the advertiser will win.
We say, bring it on!
Dr. Siddharth Shah
Comments