A Snarketing post by Ron Shevlin, Director of Research at Cornerstone Advisors
Compete recently ran article claiming that Mobile Twitter Users Are the Ideal Audience for Advertisers. In it, Compete reports that, compared to other Twitter users, mobile Twitter users in the U.S. are 86% more likely to be on Twitter several times a day and 57% less likely to use Twitter on a desktop computer.
Compete didn’t stop there. A graphic shows a number of other differences:
My take: The term “more likely” is, in all likelihood (pun intended), inappropriate here. In addition, the term is quite possibly the most misused term in marketing.
It’s possible that my analysis — in THIS case — is wrong, but I know for a fact that it happens in the reporting of many other studies. So, if I’m wrong here, my apologies to Compete. But the explanation will go to show why so many others go wrong.
How did Compete determine that mobile Twitter users are THREE times more likely to use Twitter when commuting?
In all probability (the puns don’t stop, do they), they asked consumers: “Do you use Twitter when commuting?” If 30% of mobile Twitter users and 10% of other Twitter users said “Yes”, then Compete would have concluded that mobile users are “3X more likely to use when commuting.”
If 20% of non-mobile Twitter users use the service at work or school, and and 52% of mobile users do so, then mobile users are “160% more likely to use Twitter at work or school” (32% is 160% of 20%, so added on to 20% equals 52%).
The problem with all this is that none of it actually captures likelihood of doing something.
How much “more likely” are mobile Twitter users to use Twitter on a mobile device than other Twitter users? Since 100% of mobile Twitter users use a mobile device and 0% of other users do, the answer isn’t calculable.
These statistics (and the underlying questions) don’t capture “likelihood.”
If the question had been “How likely are you to check Twitter before going to sleep tonight?” and 100% of one group said “100% chance” and 50% of the second group said “100% chance” then maybe you could say the first group is twice as likely as the second.
But it seems doubtful to me that that’s how the questions were asked.
Regardless of whether or not I’m correct in my interpretation in this instance, there’s no denying that this happens all the time when marketers report out the results of their studies.
[Hell, I do it myself from time to time, and thankfully I have a colleague who is great at catching it and making me change it.]
Marketers may overuse the terms “disruptive” and “transform” and whatever, but I’m throwing “more likely” into the hat as the most misused term in marketing.
There is another issue with the Compete that bears mentioning, by the way.
It’s bad enough that the term “more likely” is (most likely) being misused.
But the underlying contention that all these “more likely” statistics add to make mobile Twitter users worth focusing on is missing one key element: How big is this segment?
You actually have to click over to the Twitter blog post on this study to find out. There, Twitter says that “60%of our 200 million active users log in via a mobile device at least once every month.”
Hmmm. Those 200 million users span how many different countries? And you, Mr. or Ms. Marketers, are serving consumers in how many of them?
The key thing to understand here, as a marketer, is what percentage of your customers and prospects are mobile Twitter users, not how many of Twitter’s users are mobile Twitter users.
It’s conceivable that your customers and prospects that are mobile Twitter users aren’t representative of the total pool of mobile Twitter users.
So go ahead, Compete, and hate me for tearing apart your research. Join the club.