I'm at the Economist's Talent Management Summit today.  It's been trailed, and I've trailed it, as being focused on the mobile, agile workforce.  I have to say that I've not seen much about it this far (and it's now lunch time).

We did have an absolutely great kick off from Will Hutton, but that was about the new mobile, agile business - not the workforce.  And after that, most of the focus has actually been pretty transactional.  We need to engage our people, yes, but we need to go beyond this too.

So I was pleased to listen to a short input from the Economist's Kenneth Cukier on big data.  And that's without being a huge fan of big data, but I certainly accept it's part of the move to a mobile and agile future, whereas I think most of the rest of this morning has been mainly about what's already been going on today.

Kenneth started by supporting Andy Albon's (Head of HR at Birmingham City Council's) message that big data analytics are about knowing which way to jump when our assumptions provide wrong.  He then provided a summary of some of the emerging big data practices:

 1.   In the US, 60% of employees are hourly wage (call centre) workers.  50% of these will change jobs over the course of a year.  So it's a big problem.  One organisation looked at 3m data points based on 30,000 employees and found that when people apply for these jobs, if they apply using a browser they have installed (eg Firefox or Chrome) they are 15% more likely to stay longer than if they apply from a browser that came bundled in with a computer (Explorer / Safari).  You can't use this as the only signal but it's still interesting.

2.   Another organisation found that selecting people with a criminal record can lead to improved performance and likelihood to stay.  But the best predictor of whether they stay in a job is if their friends work there, or better, if they like their supervisor .  So it's more valuable to ensure that supervisors are better trained and give better reviews that paying the employees more money.

3.  A large international brokerage pays £10m to a couple of thousand highly paid employees.  Everything these employees do is measured and they're fine with this.  The company scores each research report whether it's opened, printed out etc.  And it turns out there's a correlation between if this is not a solo report but has ben co-authored with someone from a different area and whether it is likely to be seen as more valuable. Employees are now measured and rewarded on this.  It may not be the way for all employees as it's too intrusive? 

4.   Finding diamonds in the rough through social media, eg using git hub (a repository of software networkers) - spying on them - finding out who are the developers who contribute most, who's code is downloaded most.

 Kenneth and the organisation have no idea why these things works like this - perhaps something about the gumption of those who take an extra effort to improve their computers being the same sort who take a bit extra care in selecting their next role.

But does that mean you can then recruit on these measures?  Kenneth seems to think it does.  Eg, if you could use credit scores to raise the quality of your recruit wouldn't you do it?   Well, I wouldn't, no.  That's not a moral issue but a business performance and effectiveness one.  Even if there is a real and meaningful correlation here, it's not a causation.  Changing your browser does not make you a better employee and neither does a higher credit score.  Something else is causing both factors.  Find out what that something else is and then recruit for that.

That's my first worry about big data.  No matter how large the data set there'll always be other stuff outside it.  And if it's that other stuff that's truly causal, not just predictive or correlated then you're not going to know and you're likely to be led towards recruiting for completely random reasons.  Especially as the more important aspects of HR are often the hardest to capture in quantitive data form.  And are therefore less likely to be included in your data set.

My second worry is about how much all of this is being over-hyped.  Yes, data, big and small, and analytical insight into this data is important.  But does it deserve to be on every conference, every magazine issue, ever blog?  For me, the answer's no.  This isn't the future (thought it is probably a part of the future) of our profession.

Actually I thought it was interesting that most of Kenneth's examples were from The Economist's recent article on robot recruiters.  Is this the extent of the interesting case studies?  And OK, this could just be because it's early days on the journey to a more mobile and agile workforce.  But I still suspect it's something else.