“Data-Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Marc Goldberg, CRO at Method Media Intelligence.
Over the past several thousands of years, the trail of human civilization has been blazed, quite literally, by its advances in wayfinding. Prior to a map and compass, our ancestors relied on the natural landscape, the rising and setting of the sun and the stars in the sky to guide their way.
With the modern technology of GPS literally at our fingertips, we can assess where we are at any moment and weigh our options on the best route to our next destination. We can choose our own adventure and select a path with fewer turns or fewer tolls along the way.
But such transparency of data is lacking in our “modern” ad tech ecosystem – and not all of the data a vendor supplies is created equal.
So, if you’re not familiar with the landscape (and unsure of where exactly you want to go), it’s easy to lose your bearings when presented with information that can prolong your journey – and deplete your budget – as you trundle toward your intended destination.
During the review stage of a recent discovery test we did for potential clients, I received what has become a common question about invalid traffic: “Is this impression classified as SIVT (sophisticated invalid traffic) or GIVT (general invalid traffic)?”
Here’s the problem with that question: The answer is irrelevant.
GIVT vs. SIVT doesn’t matter
Most media buyers believe SIVT is worse. They assume, given its name, that it’s the more sophisticated type of fraud. This is untrue. “Sophisticated” refers to the sophistication of the detection, not the traffic.
And, anyway, sophisticated or not, if traffic is invalid, why even pay for it? Determining SIVT versus GIVT is a moot point if there is no human on the other side of the impression.
Furthermore, knowing whether traffic is SIVT or GIVT will not provide a road map for optimization. Classification of the type of invalid detection is not helpful in making improvements to targeting or delivery without also having proper supply path mapping in place.
Mapping is really important if you want to improve your buy. Regardless of a GIVT or SIVT label, the only thing that matters is that it’s invalid traffic. The information your vendor supplies should amount to clear directions on how to improve your buys – and it really helps to know the right questions to ask.
Ask lots of questions
If you look at the reporting interface provided by some of your vendors, you will likely find reference to your “proxy server (RVI) ad rate.” I challenge you to ask your vendor what this means and what you’re supposed to do with this information. And if it’s bad, how can you reduce this number?
Do the same for all of your vendor’s reporting, and then ask: How can I make any of this actionable? Classification is a nice-to-have, but not the only thing to have.
Because here’s what you need to remember: Reporting may be robust in some of these systems, but not necessarily helpful. More information is not helpful when you don’t know what it means and you don’t know what to do with it. A data output may look impressive, but be warned that it can also be deceptive.
The data they’ll give you is similar to a compass, as in it’s directional. The data you really need access to should be more like a GPS and gives directions. You should be able to understand what the data says and be able to do something with it. The supply map cannot be as simple and obtuse as “go north.”
Go deeper than verification
Data scientists and analysts will often talk about data integrity, and this is important. But there is also a common misconception that “verification” implies integrity. Reporting that focuses on classification infers that the data set is complete, and this is simply not the case. Rather, it’s the root of most of the problems in our industry today.
Sampled data sets have always had flaws, but the industry has been conditioned to believe it to be the currency of value. Granted, the TV world can’t get to 100% measurement – and how could they, given the methodology? – so sampling became acceptable.
But in digital, all media, including ads, is served directly to consumers, and therefore it can be measured in full. As a buyer, you need to ask why your digital verification vendor is still sampling.
You may think they’re not, but they are. Just ask them about their decision rate. If you haven’t asked this before, I don’t fault you. But moving forward, the decision rate will be a required disclosure in the MRC’s guidelines. If you cared so much about GIVT/SIVT, you’ll definitely care about this.
I’ve had to break the news to many buyers that some of their vendors may be firing on every impression, but they’re not making a decision on every impression. This is a subtle nuance, but “fire on” and “decide on” are two vastly different things.
Let that sink in.
Transparency and reporting for the sake of reporting isn’t valuable. But the repercussions are worse than just a lack of value. Surface-level reporting provides a false sense of security and direction.
Where in your reports did you see that all of your data is measured? Once you have all of the necessary data, though, you can start using your GPS properly to better understand the best and worst (supply) paths to take.
Start asking the hard questions. Because heading north on Easy Street won’t do anything for you other than get you nowhere fast.
Follow Marc Goldberg (@MarcJGoldberg) and AdExchanger (@adexchanger) on Twitter.
