Those of you who follow closely know that I’ve been asking about putting together back issues of my newsletter as a purchaseable eBook. One of the questions in the survey was about what price, if any, you would pay for such an eBook compilation. I also asked for your Twitter handle so I could follow you back and do some data analysis.

In that spirit, I took the respondents of the survey, 67 of them who provided Twitter handles, and ran Klout and PeerIndex on them with the simple question: does influence have any correlation to purchase intent? The reason I’m curious about this is that if there were a strong correlation, when we go targeting audiences for social media campaigns, that influence score could be used as a proxy for purchase intent and help us target better.

A few comments about the data: this data is not representative. It is self-selected respondents to a survey who are members of my audience. There’s a built-in bias to the survey – if you don’t like me, chances are you don’t follow me, and therefore wouldn’t have any purchase intent either. Conversely, everyone responding to this survey should have some built-in purchase intent as people who follow me and care enough to respond to a self-selected voluntary survey. Before you go post this on Mashable as the next great research study, know that it does not pass any tests for representation of population at large. (yes, as a bit of tongue in cheek, I labeled this post [NEW STUDY] even with these conditions)

So, without further ado, Klout:

SOFA Statistics Report 2012-03-13_08:03:02

And PeerIndex:

SOFA Statistics Report 2012-03-13_08:03:02

For the non-statistically minded, these are very weak correlations, so weak that they’re nearly meaningless on their own. (they are just barely statistically valid, too – 67 df and a two-tailed p of 0.287 for Pearson is just scraping the line)

Take a step back and consider that this is an audience with a built-in bias, so there’s already some latent purchase intent that wouldn’t be there in a representative sample of the population. I draw the reasonable conclusion that influence scores are no predictor whatsoever of purchase intent. Now, you might say, “Well duh, that’s perfectly logical, they’re two separate behaviors” and you’d be correct…

… except that there are marketers and agencies (I’ve talked to a few at conferences and events recently) who are using influence scores as the be-all and end-all of their data-driven marketing efforts. Everything they are currently doing is focused on those influence scores, which doesn’t make a whole lot of sense, especially in light of the fact that I can’t get more than a meager correlation out of a biased audience.

The bottom line is, influence scores appear to not say a whole lot about who will actually buy from you – so don’t use them for that end.


You might also enjoy:


Want to read more like this from ? Get updates here:

subscribe to my newsletter here


AI for Marketers Book
Get your copy of AI For Marketers (2019 Edition)

Analytics for Marketers Discussion Group
Join my Analytics for Marketers Slack Group!


Pin It on Pinterest

Shares
Share This