Tucana People Analytics World 2017 ~ Strategic Human Capital Management (HCM) Blog

Friday, 12 May 2017

Tucana People Analytics World 2017

I’m late in providing my summary of Tucana’s People Analytics World 2017 and have struggled to find a topic that others have not already written about. However I think one useful perspective may be the level of variety within analytics at the moment.

This applies firstly to analytical approaches. Most of the presentations, not unnaturally, focused on quantitative analytical approaches. My favourite from these case examples was Microsoft’s use of social traffic metadata to suggest best practice behaviours of the most effective managers. Becky Thielen and Jeremy Tyers explained that this has shown, for example, that effective managers maintain larger networks than their reports:

Employees reporting to managers with large networks are 7% more engaged and have networks up to 85% larger
Employees with networks 110% larger than their managers are 50% more likely to be disengaged.

However, I also enjoyed the qualitative examples, eg Adidas’s analysis of employee experience during key moments that matter, following the same approach used for the company’s customers, and applying across HR / management activities, workplace and the digital workspace. This involved an an automated employee response system providing a Net Promoter Score, but also open text comments, and textual analytics to help see summaries and patterns across the business. Stefan Hierl suggested that to really need to understand what’s going on for people you need to read this stuff and hence gave leaders access to the comments of their people. He often talks to leaders about the numbers but you also need an equal focus on the qualitative side - what’s in the minds of people as this provides a better feeling for what’s going on. At Adidas, listening to people is a core competence of HR - we should not outsource this.

Similarly, some of the presentations focused on very data oriented approaches. I liked Experian’s example of flight risk modelling, taking data on over 200 employee attributes from multiple sources and plugging it into Experian’s business modelling tools. This has delivered circa 20 key statistically significant attributes.

But what I really liked about this particular conference was its heavy focus on a strategy based approach. This started with Alec Levenson’s session noting that the tasks and behaviors that do not improve strategy execution vastly outnumber the ones that do. Therefore analytics needs to start with strategy not data. The point was at least partly supported by Peter Howes noting that incremental improvement on traditional data reporting tends not to work that well.

Even more important than this level of variety about analytics was the variety about people management itself. This includes different views about the direction of causality between culture and behaviour, and engagement and performance. Peter Howes asked us whether we are making decisions about people with the same intellectual rigour as our decisions about other business factors: Finance, Sales, Production, Supply Chain. Whilst we don’t even fully understand the role of something as central as engagement this struck me and at least some other delegates at the conference as a rather ludicrous suggestion.

I think the amount of variety I have described above has a couple of important consequences.

1. There cannot be a single maturity curve which all organisations need to follow. I much prefer Deloitte’s and 3n Strategy’s approach of identifying different aspects of analytical maturity each of which can be strong or weak independently of each other (eg Deloitte’s categories of strategy, people, process, technology and data).

However, for many organisations, the best approach may be a combination of the various approaches. For example, Eden Britt at HSBC presented a heavily quantitative, data oriented approach, involving, for example, several smart analytical approaches for providing insight on attrition:

Clustering - which are the areas of high attrition?
Machine learning - who will leave?
Text mining - where will they leave to?
Markov models - when will they leave?

But Britt also emphasised the importance of deciding upon:

What is the question you are trying to answer?
What is your hypothesis?
What data do you have?

His suggestion when requested to provide yet more data is to always ask: “How’s that data going to help you make a better decision or to do something in a different way?”

2. I think the variety in our approaches also raise broader challenges to emerging or prevailing wisdom:

It means there is limited use for standardised taxonomies. These may be OK for whether someone has a driving licence but not for eg employee engagement, or for analytics itself. These may help keep the lights on cheaper but they also interfere in the search for competitive advantage.
It may also mean there is limited use for benchmarking models and tools. For example, when comparing cultures, is hierarchical opposite to delegating?, or professional the opposite to social? - they’re not in my view of the world.
It means that setting up a professional body for people analytics would be largely unhelpful. At this point in the evolution of the people analytics field we need to encourage conversation on different perspectives rather than to see these constrained.