"Waiting for [insert ad server here]..." shouldn't be holding up the delivery of your web content. Too often, though, ads are placed before content in the HTML of a page, so a failure to deliver an ad will often result in your not seeing the content you were looking for.
I feel like we got to a good place when we had only a few ad servers to worry about. Publishers used their own ad servers to rotate ads and manage inventory. Advertisers and agencies used their own ad servers to control creative rotation, and to pull reports at any point during the campaign. The worst that might happen was a latency issue that could be easily traced and addressed.
Ad queues for any given placement are a lot more complicated these days, though. In addition to farming ad impressions out to advertisers and agencies, publisher-side ad servers also have networks and exchanges to deal with. As creative as the solutions have been for minimizing latency on the technology side, there's no denying one simple fact: There's a limit to how many parties can get a look at a given ad impression, and that's without considering how many can reliably decide whether or not to take it or kick it back to the publisher within a reasonable length of time.
That limit will vary, of course, depending on a wide variety of bandwidth and end user-dependent variables. But there's a limit.
Also consider that once the multiple ad-delivery systems develop a real-time consensus on which ad to serve, the floodgates really open when it comes to server calls. Rich media ads can call multiple servers to deliver ad assets, particularly when streaming video or interactive content gets involved. There are calls for click-tags and destination URLs, calls to verification services, and third-party surveys.
There are plenty of calls that happen outside the queue that relate more to marketing and measurement than content delivery -- calls to analytics services and measurement companies are among them. But let's put those aside for a moment and just look at the ad queue for a moment.
If you were to graph the average number of ad calls made per ad delivered and look at that over a span of time -- say 1996 to present day -- there is no argument that would make me believe we wouldn't see a hockey stick at the end.
Which makes me wonder... From an industry perspective, who is looking at how advertising holds up content delivery from a 30,000-foot level? Who is calling for a better understanding of how the increasing number of calls to external servers is wreaking havoc on our ad delivery infrastructure? How much more can it hold before it fails? How much longer are end users willing to wait for the content they asked for to be rendered in their browser?
These are all issues that digital advertising folks should be thinking about as exchanges, verification services, and other such things become more popular. If we are overtaxed, as I suspect, then we're going to need to reprioritize and reset expectations about what the ad delivery infrastructure can be reasonably expected to do.
Like a crack in the foundation of a house, this is something that would be tremendously easy to ignore until it becomes a really serious problem. Ask your friendly neighborhood ad-ops person for his or her perspective. Let's not sweep this one under the carpet.
Tom Hespos is the chairman and president of Underscore Marketing and blogs at Hespos.com.
On Twitter? Follow Tom at @THespos1 or @_MarketingLLC. Follow iMedia Connection at @iMediaTweet.
