Programmatic advertising is for many companies THE way to do advertising. It uses computer systems to automate bidding, making split second decisions about whether to bid and how much based on a particular user’s interests and browsing history. To a large extent, this is the only way that advertisers can possibly handle Internet marketing since there’s no way that a human being could deal with all of the decisions required to determine when to bid and how much, much less in the amount of time given before a website loads.
There are some major issues with programmatic advertising, and although there’s really no way to get around using it in many circumstances, advertisers should at least be aware of them. Two in particular are the high costs of this type of advertising and the fact that the choices that computers make can be questionable at best.
AdAge recently took a look at why programmatic advertising is so expensive and highlighted the fact that it carries enormous overhead costs. With standard, non-Internet advertising, the manpower required to set up an advertising campaign is pretty low, but a large number of people are needed to keep programmatic advertising going.
Five full-time employees are needed to spend a $100 million national broadcast budget, one media agency executive said, while the same number would be needed for a $5 million programmatic buy.
It takes a significant number of highly paid people to keep programmatic advertising going because it relies on computer systems and a number of large databases in addition to some very complex and math-y algorithms. The technical aptitude required to keep these systems running is high, and mistakes can be extremely expensive. According to the VP of integrated marketing at Kimberly-Clark, about $2 million in mistakes is made each year, with a single error costing around $30,000.
As a result, commissions for handling these projects is much higher than with non-Internet advertising, and some companies may be looking at paying multiple commissions if several advertising systems are involved. It is estimated that about $.50 on the dollar goes to the actual cost of an ad.
Advertisers may think that these costs are well worth the price of doing business due to the high levels of targeting available to them through programmatic advertising, but the accuracy is fairly questionable. The data that agencies have about a particular person may not paint an accurate picture of a user, and computer systems are required to make inferences about a user that may completely off base.
For example, I occasionally am served advertisements in YouTube that are in Spanish. The only time I ever come close to speaking the language is when I mispronounce my order at Taco Bell. I have no idea why Google thinks I can even understand these ads, but they hold the firm belief that I can.
Part of this may be due to the fact that a computer is required to take signals from a user and make a guess about what it means. Someone looking up a recipe for sugar free cookies could either be diabetic or on a diet. A person looking for items for newborn babies could be expecting or attending a baby shower. Computers may have a vast amount of information about a user, but without an actual background on the information, programmatic advertising is essentially based on hopeful guesswork in many cases, and expensive guesswork at that.