Your ad server decides whether to serve an ad. Why then is a spreadsheet handling the prediction? Your ad server knows how many ads it has displayed. Why is that information being re-keyed into a billing system? These are blockers to efficiency and success. They are not insurmountable. The industry will move in that direction.
The technology behind managing online display advertisements can be incredibly complex. The business behind it, the technology, is even more difficult to comprehend. Then there’s the poor publisher behind who buys the ad-server technology to manage their own increasingly complex business. These are the people I work with day in, day out. Publishers are trying to get a grip on technology when, truthfully, they don’t care about technology. Generally, they care about delivering the advertising campaigns that somebody is paying them to deliver. The tech should do that. Seamlessly.
Sadly, much of the technology in this space is a black box. Few people in an organisation understand how it determines how/when/why ads appear. While it’s not ideal, I don’t consider it too surprising, given that the online display market is only a teenager and is still maturing. New approaches to selling online media result in new tech features, which in turn need to be understood and explained to the end business. Here, I am focusing on the publisher ad-serving business because the agency world presents different problems.
However, there is one part of the business that I think we’re missing in the advertising tech game for online content owners. We need to be the dashboard for a publisher or content owner’s advertising & marketing departments. And we’re not. We are one of many systems, from inventory prediction spreadsheets to unrelated billing.
Although we shouldn’t need to be every component of the system, we do need to unify the data. Your ad server decides whether to serve an ad. Why then is a spreadsheet handling the prediction? Your ad server knows how many ads it has displayed. Why is that information being re-keyed into a billing system? (OK, not everywhere, but in many places). These are blockers to efficiency and success. They are not insurmountable. The industry will move in that direction.
However, there’s one area where we have a disconnect. Advertising technology systems don’t tend to integrate well with those web analytics systems. This, to my mind, is the biggest disconnect we have in the industry.
I do know all the arguments that explain the disconnect; I’ve used them myself. The systems are trying to achieve different things, and the stats from the analytics systems provide lots of useful data beyond the kind of advertising data our industry needs. Designers, content producers, systems admins, and network people all need the data web analytics software delivers. Often, it’s great and detailed data.
But web analytics systems report the number of users on websites. In any business, these numbers are usually passed along to management and the board. They’re used on marketing materials to shout, “Hey, look how great we are. We have a big number here, and it’s getting bigger by the month. My graph groweth towards the sky”.
People love good numbers and a good story. So, the board in the company gets the marketing team to tell the good number story far and wide. And they really don’t have to go that far until they’re telling it to the ad salespeople. Who, in turn, pick up the phone and tell the news to their customers. And before you know it, everybody wants to buy ads against these lovely big numbers because they have these big numbers.
Unfortunately, the ad server disagrees with these big numbers. There are lots of reasons why. Some valid reasons for different data, some valid differences between counting approaches, some just tech differences. Occasionally, these differences can add up to a big number.
Some years ago, a major broadcaster had a site that generated big numbers. But their ad technology didn’t say the same number. Remember, people like big numbers. So the management didn’t like the smaller numbers from the ad server. Smaller numbers meant a smaller revenue. And yours truly was asked to look into it.
After being provided with two data sets, I started looking at them and, quite quickly, found a number of clear, obvious discrepancies. The nice analytics numbers counted all the pages, even the ones without ads (and there were a good number of them). The analytics numbers counted non-human traffic from search engine bots and other robots. You can’t serve ads to them. And if you do, people don’t like it.
I could go on, but this is already long enough.
Let me be clear. This is not a problem with the analytics industry or the ad technology business. Analytics companies could easily provide reports that match by filtering, in the same way ad servers do. That’s just a report to them. But the issue is with the end client. They don’t want to know the technicalities of the differences. And why should they? But equally, they should not be allowing sales and marketing folks to build and sell advertising models based on numbers they have no chance of delivering.
My appeal to the ad-server business is this. Develop an interface that marries analytics data with ad-serving data. For my purposes, let’s assume that’s easy. Create a single view of the data for the end business so they buy, sell and predict from the same numbers. Make sure predictions and post-delivery reports are based on the same data,
I think it’s the only way we can make the industry work without the distrust that happens when different systems purport to report on the same things but provide different reports.
And don’t get me started on differences between ad servers!
