But the article is wildly exaggerating the "nobody knows" gap of superbowl ads. There's an entire industry specializing in measuring the effect of "write only" ad channels and all the big TV advertisers spend a considerable fraction of their ad budget on commissioning those measurements.
Now the immediate, built-in metrics of online advertising have been eroding the business of that industry for quite a while. Those metrics are far more detailed, they boast low SNR and absolute precision as opposed to the murkyness of extrapolated trendlines in polls. It's very tempting to rely only on the built-in metrics then, and that's exactly where the problems the article is exploring are kicking in. Those old-school pollster methods that start with zero data and only measure the continual development of brand awareness plus the occasional campaign trace have something big going for them, which is that the little data they yield is universally relevant, whereas that ocean of numbers that is produced by built-in metrics can easily be a useless distraction or at worst actively misleading. Perhaps our would be wise for big ad spenders to employ some of that "zero data" brand awareness observation that could measure TV ads even when all the ad-spending is going online, into the realm of built-in metrics.
Now the immediate, built-in metrics of online advertising have been eroding the business of that industry for quite a while. Those metrics are far more detailed, they boast low SNR and absolute precision as opposed to the murkyness of extrapolated trendlines in polls. It's very tempting to rely only on the built-in metrics then, and that's exactly where the problems the article is exploring are kicking in. Those old-school pollster methods that start with zero data and only measure the continual development of brand awareness plus the occasional campaign trace have something big going for them, which is that the little data they yield is universally relevant, whereas that ocean of numbers that is produced by built-in metrics can easily be a useless distraction or at worst actively misleading. Perhaps our would be wise for big ad spenders to employ some of that "zero data" brand awareness observation that could measure TV ads even when all the ad-spending is going online, into the realm of built-in metrics.