Traditional media verification, like brand safety, IVT, and viewability, paint a glowing picture of improvement year after year. How can we explain such progress? Simple, said Integral Ad Science, the industry focused on it.

Want to know what happens when we don’t focus on key sources of media waste, such as when we just assume data is accurate without verifying it? Jake Moskowitz from the Emodo Institute explains how negligence has left us with a data accuracy problem that’s larger than all the above metrics IAS measures — combined — and how it’s the largest source of media waste.

One way the industry typically measures the accuracy of data is through Nielsen Digital Ad Ratings, which measures the age and gender of people whose devices are served ads digitally, and what percentage of users who were served ads were in the actual age and gender segment the ad was trying to reach.

What we see, consistently from Nielsen benchmarks, is that this data is about half wrong, sometimes even worse.

Location data, while one of the most valuable and useful marketing data sources, is another significant source of media waste. One reason is the false sense of trust people get out of the precise nature of the data. Lat/long data often goes out to six decimal places, which translates to knowing where a phone is within three inches. And while we know that location data on phones is not that accurate, our industry blindly trusts this data to be true.

Looking at cell tower data from major carriers in the US, Emodo is uniquely able to see what tower a given device was connected to right before and after a data point was collected, and know that if those towers are 100 miles from where a device was supposedly located at the time, that data point can’t be true. In more than 100 studies we’ve run, we found that only 39% of location data was accurate to within one mile.

You can’t do a responsible job of building a segment, nor make key decisions based on an attribution study, if 61% of data is wrong by more than one mile. It’s time to take the same level of energy we put into traditional verification metrics and put it toward driving improvement in data accuracy. We’re bound to see the same type of improvement.