When you run online media programs do you know what you are really getting or paying for? Over the years behavioral targeting or “BT” as it is often referred to in the industry has become an extremely popular method of targeting these programs.
There are many diferent definitions of “BT” and what media vendors call BT. Some are as simple as tracking people who come to a website and do or do not perform a specific action. When the person visits a site within the ad network, that person is shown banners for the site that is running the media campaign. A company might want to follow people around the web who have been to the site and not checked out with the shopping cart, they might choose to (re)target all consumers who have been to a website or even specific pages of a website.
Some more advanced types of behavioral targeting may pick up demographics of an individual including age, sex, marrital status and income among other things who have been to certain sites or categories of sites and have (or not) performed certain actions online and target those consumers with ads for a particular product or service.
If you have a home computer and your wife shops at Lane Bryant, when you log onto the computer and get shown Lane Bryant ads on all the sports sites you visit, it is because one or more ad networks are tracking and recording the activities perfomed by the Internet browser on your computer.
Typical click-through rates on banner ads these days are about as close to zero as you can get in most cases. This is especially true for banners on general content sites. Even banners that are behavioraly targeted often get click-through rates just that are just a fraction of 1%.
In traditional media where nothing (or very little) is tracked and it is primarily demo targeted impresions that are purchased, there is no expectation of any direct response to the ads since they are simply broadcast to consumers.
With all the tracking capabilities available on the Internet it is possible to show the return associated with banner advertising. Whether it is a good investement or no depends on how it is measured. When you run a PPC search campaign on Google, Yahoo!, MSN, Ask.com or one of the other 3rd tier engines, the success of the campaign is generally measured by the amount of sales revenue generated in relation to the amount spent on advertising. Sales are only tracked when people click-through the ad and subsequently make a purchase, fill out a lead form or perform some other activity that takes them a step or two closer to completing a transaction.
In order to justify the cost of media campaigns when nobody actually clicks on the ad, the “view through” metric was concocted by the ad networks and media companies to attribute credit to the media campaign. If someone merely loads a page with an ad on it (whether they view it or not in many cases), the ad server sets a cookie and if a person subsequently makes a purchase within the duration of the cookie window (typicall 30 to 90 days), then the media campaign is credited with the sale.
This brings up the question – if the consumer has already been to the website and not made a purchase, would they have purchased regardless of whether or not they were subsequently exposed to the banner ad? It is likely that there is some lift, but it it anywhere near enough of a lift to justify all the costs associated with a media campaign?
While there are tons of “studies” that say banner/online media is effective (all put out by the the ad networks and companies that stand to profit when you buy online media), it doesn’t seem than any neutral party has evaluated whether onine media really has a significant effect on consumer behavior.
If anyone really wanted to test out the effectiveness of online media, at least from a retargeting perspective, they could run a test in the following manner.
Pick a site with significant site traffic such that a large retargeting pool would be available.
Setup the campaign as one normally would with all the different actions and behaviors that result in a consumer falling into the retargeting pool
Serve the banner to half of the people in the retargeting pool
Set the cookie that would normall be set when a person views a banner but not show the banner to the other half of the retargeting pool
Look at the difference in behavior between the people who were exposed to the banner and those who were cookied as if they say the banner
If there is a significant lift in sales or other conversion events associated with the portion of the retargeting pool who were served the banner and not just cookied, then we can infer that the media campaign had a positive effect on consumer behavior and quantify the value. If there is no statistically significant difference between the two groups of consumers, then we conclude that the media campaign had no positive effect on consumer behavior.
Other variables including the creative, the number of impressions served per person and the frequency with which they were served and the length of the campaign would obviously affect the results, however the campaign could be let to run long enough to determine at what threshold of campaign frequency and duration is required to see some noticable difference. Is there an ad network out there that would run an experiment like this and agree to publish the results regardless of what they were? Probably not since it could rather conclusive prove that either onine media does not work or that the size and scope of the campaign would need to be so big that all but the largest brands with multimillion dollar budgets would be wise not to invest in an online media campaign.