Digital advertising has a renewed zeal for transparency and accountability -- and for good reason. For too long, advertisers have paid for ads that consumers never saw, either because they were blocked, viewed by bots, or not viewable in the first place. However, as trade bodies like the IAB and MRC finally introduce reasonable standards to address these issues, we are seeing such efforts come up against a surprising barrier: the structural inefficiencies of the network that delivers ads in the first place.
When the science falters
Thanks to the science at hand, it takes about one second to deliver an ad. But a whole lot happens in that second. Most of it is spent waiting: waiting for the DNS lookup, waiting to establish a connection, waiting for the request to send, waiting for redirects, waiting for tracking and viewability pixels to fire. Even small delays in this process can cause major problems that contort and obscure our efforts to account for the ad's delivery and performance. Server latencies -- and the sagging load times that they create -- are at the root of many sticking points for ongoing efforts to combat ad blocking, ensure viewability, and resolve bid discrepancies.
Advertising is a physical if not mechanical process, a flow of information traveling along network fiber to the concrete servers, spaced by real geography. All mechanisms can falter. Ultimately, our ability to deliver on a fully transparent process depends on understanding this extensive physical infrastructure that makes the entire industry possible, and how issues with that infrastructure play out in the execution of advertising's core value proposition.
Bad load times = murky viewability guarantees
Bad load times impact viewability in an obvious way: if a page takes too long to load, fewer viewable ads will serve. A lot of publishers will say that non-viewable ads were simply below the fold. The truth is, they are often the result of the ads taking so long to load that no viewability pixel could fire. Or worse, a user gets fed up and clicks away from the site before the ads -- or the content, for that matter, ever had a chance.
In a tragic twist of fate, poor load times are often a direct manifestation of publishers' attempts to make more ads viewable by loading them all at the top of the page. It's an issue that no amount of lazy-loading can fix. IAS and others measure viewability by firing a dedicated pixel along with a digital ad. If a page is taking a long time to load, that pixel can fail to fire even when the ad has billed.
The triggers of blocking
Sluggish ad load times are also a big reason why users block ads. Recently, The New York Times found that a majority of the page's data is devoted to loading ads versus loading base editorial. Ads that consumers don't want to see are using more than half of their bandwidth. That glut of bandwidth represents inefficiencies in the networking infrastructure.
Recent discussions about how the advertising ecosystem is going to respond to the thread of ad blocking have centered on ad quality. Ads should be better; they should be less interruptive, more targeted, and so forth. But they should also be leaner. They shouldn't weigh so heavily on bandwidth and budgets. That means addressing the complex system of how those ads are served.
Server latencies also cause discrepancies in reporting of impressions, usually between 5-10 percent of budget outlay. Basically, the lag between a booked request and the actual delivery of viewable creative causes differences in accounting. Advertisers have grown accustomed to these inefficiencies. They've adapted by developing a motley array of stopgap measures to make up for the difference, from overbooking or "padding" their impressions, to planning altogether separate "makegood" campaigns. But the structural problem isn't human error -- it's caused by delays in the networking itself.
Go under the hood
What this means for advertisers and their technology partners is that they should take their networking infrastructure very seriously. There are a few different ways to go about solving for bad load times, by testing more frequently and streamlining operations to ensure fewer ad servers operating on each page. But it all begins with a deeper understanding and more robust scrutiny of the networking infrastructure by all parts of the advertising ecosystem; both supply and demand sides, all the way down to creatives and content producers, should go under the hood and understand what's clogging the pipes. In an atmosphere of heightened awareness around transparency and verification, the stakes are certainly high enough.