The future of AI in marketing
May 23rd, 2019
“Humans don’t make very good decisions about people” – Tom Davenport
That was the attention-grabbing headline to an interview in the HR industry publication, People Management, in its July 2014 issue. Given that Tom Davenport is one of the most respected authorities on how businesses use Big Data, we were curious, to say the least, on what his thoughts were on the role of data in the hiring process.
The author of Keeping Up With the Quants and Big Data@Work, Davenport is well respected for his level-headed analysis of the use of metrics in the work place. In the interview, Davenport notes that despite the availability of useful data, HR departments tend not to use that information to derive conclusions and make forward-looking decisions. Instead, data is used merely for reporting, which Davenport terms ‘backward-looking.’
Davenport argues that data should be leveraged to assign values and inform business objectives. For example, he says, if your company has an attrition problem, you can mine data to make connections, identifying the issues linking departing employees, and thus perhaps right the ship going forward. This line of thinking makes perfect sense; it’s not a far cry from what ERP systems have been offering HR departments for years – a set of transactional data to capture employee data.
These metrics can be enormously useful, of course – but they’re less effective at getting at the underlying reasons why you might have an attrition problem. Operational issues are just as likely to stem from dynamics, personalities and emotions as anything that could be captured in an ERP system. And we can’t see how the role of HR could ever be replaced there. It’s about building trust with people to get them to talk openly, actually looking at people and their environment to spot problems and working out the right time and channel to communicate.
So the claim that humans don’t make good judgements about other people surprised us. To clarify, Davenport says in the interview that humans “tend to seek out and hire those who are much like themselves” and can be “unfair in evaluation processes.” Certainly there’s a grain of truth here. Human judgement is fallible and there are times when our own biases or agendas can cloud the hiring process in a negative way.
But on an enterprise level, we believe the human element is invaluable, crucial, irreplaceable. Leveraging Big Data to find specific skills sets or qualifications overlooks the critical element of ‘fit’ – is this employee right for your organisation? By trusting an algorithm to predict the best hires for your company, you miss out on the intangible values that the human element adds – the ability to assess future potential versus current qualifications and technical skills, for example, or the perception to decide whether a candidate really matches your corporate values.
It all comes down to the age-old question of whether it’s better to hire for specific skills to meet a short-term operational need, or to find someone who has the potential to be a true asset to the enterprise in future. We’ve seen many examples of someone being hired because they have expert knowledge of a particular software, only to find they’re simply not a good fit for the organisation. We believe in sticking with the old adage ‘Hire the attitude, not the skills’ – and humans are simply better at assessing that.
Which is not to say that humans don’t make mistakes in the recruitment, retention and engagement processes, of course. But they’re not problems necessarily resolved by the use of Big Data. Indeed, by the end of the interview, Davenport concedes that data “will never replace the need for some level of interaction before you hire someone.” We couldn’t agree more.
We’re curious to hear your thoughts on the topic. To what extent do you use metrics in the HR function? Would you ever hire someone identified by an algorithm without meeting them first? Please share your comments below.