Augustine Fou wrote an article for Forbes titled Current Digital Marketing Is Sports-Watching, Rather Than Marketing.
« More data doesn’t mean better data. More data could often mean just more crap. And the insights derived from it are just as crappy too. »
« If you’ve read any of my previous writing on bot activity and ad fraud, you will remember that large numbers of ad impressions, clicks, and traffic can be generated by bots. All of these activities are faithfully recorded by analytics. If marketers don’t realize this and correct for it, the insights they draw from the analytics will be wrong and they will make incorrect business decisions. For example, fake or fraudulent sites often exhibit much higher click through rates on ads than real sites with real human audiences. That’s because the fake sites use all bot traffic; humans don’t click on ads much (that’s assuming they even saw the ad in the first place and were not using ad blockers). Marketers have fallen victim to these “too-good-to-be-true” numbers and unwittingly allocated even more budget to fake sites (exhibiting high CTRs) and away from real sites (exhibiting lower CTRs). »
« Bigger numbers in dashboards and excel spreadsheets are just as mesmerizing as sports scores, but it’s not actual marketing. »
« But having 5,000 audience segments to choose from is simply an “illusion of choice” when targeting most audience segments is unnecessary and doesn’t yield incremental business outcomes. That’s because those audience segments were made from crappy input data. »
« There’s also the illusion of control. This is where marketers think that by having fraud detection in place, they are avoiding bots and fraud. But what if the bots are able to trick the detection to avoid getting caught, and other forms of fraud (like pixel stuffing, ad slot refreshing, popunders, etc.) are not detected by the tech? And you can never check for yourself because the verification is “blackbox” — i.e they don’t explain how they measured it. Marketers also think that paying for brand safety tech helps keep their ads off of bad sites. That assumes the BS tech actually works. (Hint: it doesn’t work, as previously documented). »
« Digital marketing works… but not with the big numbers many marketers are used to seeing these days… »
« Marketers should also make sure to focus on incremental business outcomes… Did the marketing activity drive more sales than would have happened even without the marketing? See a few examples here: When Big Brands Stopped Spending On Digital Ads, Nothing Happened. Why? »
Shain Kahan commented on Twitter:
« The ratio of “good” data to “bad” data is always decreasing, to asymptotically approach zero! So by necessity, if not by definition, more untreated data always means more bad data, modulo your luck. »
See also: Data Driven by Tom Redman.
“It costs ten times as much to complete a unit of work when the input data are defective… as it does when the input data are perfect.” Redman stresses the importance of preventing erroneous data at the point of creation. His data lake analogy illustrates this point. Imagine a lake of polluted data fed by polluted streams (business processes). Cleaning the lake is futile unless you first address the sources polluting the streams.