Analytics is a funny critter. Sort of like The Blind Men and the Elephant (Note: while the story is ancient I'm partial to this poetical version) wherein truths can be limited. Even (deliberately) misconstrued.
A month or so ago I posted Fool Me Once... Fool Me Twice... noting the Road to Damascus moment of an enterprise CMO and the dirty little ad impression 'secret'.
Now Google itself has released a study, The Importance of Being Seen, Viewability Insights for Digital Marketers and Publishers, (pdf file, 496KB), November-2014, that agrees to 56.1% of their ads being served not being seen - but still counted as impressions.
There is a nice dollap of information provided BUT the twists and turns and exceptions are very telling and on which I'll be commenting; go read it for the good stuff, there is some.
The biggest laugh - if one is an advertiser - is that Google uses the Media Rating Council and IAB standard of a viewable impression as: 50% of an ads pixels are on screen for one second. Yup, that's a 'viewed' ad, folks!
The most obvious twist is Google's always following the 56.1% unseen number with - and highlighting - average publisher viewability is 50.2%. Yup, the percentage is skewed by a 'small' aka 10% number (20% by my interpretation) of sites. It is interesting to note that Google, while noting this, says nothing about addressing it.
Their turn (following the twist) is the graph showing that vertical ads are seen more... given that it is more likely that 50% of a vertical ad will be seen at or below the fold this is not exactly surprising.
And then here comes the exception: study based on display (but not including video) ads rendered in browsers only.
Me thinks, that while there is a lot of useful information provided (go read), the manipulation and benchmark shenanigans lower the percentages substantially. That doing so still can't get below 50% overall is hysterically troubling. For advertisers.
Lies, damned lies and statistics... in a similar vein to the above is how the shale gas reserves are being calculated (included for our resident geologist) by the US Energy Information Administration (EIA) and a new university study. Natural gas: The fracking fallacy by Mason Inman, Nature, 03-December-2014.
The main difference between the Texas and EIA forecasts may come down to how fine-grained each assessment is. The EIA breaks up each shale play by county, calculating an average well productivity for that area. But counties often cover more than 1,000 square kilometres, large enough to hold thousands of horizontal fracked wells. The Texas team, by contrast, splits each play into blocks of one square mile (2.6 square kilometres) — a resolution at least 20 times finer than the EIA's.
Resolution matters because each play has sweet spots that yield a lot of gas, and large areas where wells are less productive. Companies try to target the sweet spots first, so wells drilled in the future may be less productive than current ones. The EIA's model so far has assumed that future wells will be at least as productive as past wells in the same county.
What it says on the box may not be what is inside.