Did you know that regression (or any GLM model) assumes that the Y value observations (outcome or dependent variable) are independent? If they’re not, your coefficients can be inflated and your predictions inaccurate.

For example, say you’re trying to find out what drives performance, engagement or retention in your teams; in other words, your dependent variable is performance, engagement or retention.

Now ask yourself a question: are the engagement scores of employees within a team independent of each other? Of course not: people in a team influence each other. For example, if Lao is unhappy about some aspect of work and disengaged, there’s a reasonable chance that other members of that team will feel the same and therefore their scores are not independent as required by GLM regression.

Or consider another example: if a team leader’s behaviour affects one team member’s performance, chances are this behaviour is influencing performance across the whole team. Therefore the team members’ ratings are not independent and you are violating a key regression assumption which will result in inaccurate predictions and poor recommendations to the company.

I discovered this vital fact when researching the drivers of couple romantic relationship satisfaction (a team with n=2) and quickly learned that if one partner is dissatisfied with the relationship, there’s a good chance the other will be dissatisfied as well. Thus the scores are not independent and regression will deliver inaccurate predictions. How did I fix it?

The solution is to use a technique called Multilevel Modelling (MLM) which doesn’t assume that outcome variables are independent; it’s certainly the best way to get accurate results and make good recommendations if you’re working with team data. In fact, I won’t do any #PeopleAnalytics team data project without it.

MLM has been in the news a lot in the past few years because political analysts used it to successfully predict Trump’s election wins and Brexit. Before this, they’d been using regression and getting their forecasts wrong.

If you want to avoid the errors made by political forecasters and make accurate predictions and valid recommendations about what drives team performance in your organization, move away from GLM regression to MLM.

More information available on request.

People analytics – when implemented in the right way – can be a real fortune-maker for businesses. The problem is that a lot of ‘analytics’ styles out there will not reap the rewards that businesses expect. Explore the four styles outlined below to see if your analytics approach is missing a money-making trick…

One of the perks of being a consultant is that it allows you to compare a variety of business approaches used by different companies. Take people analytics for example: I’m constantly amazed at how some businesses make a fortune from their investments in this area, while others struggle to make ends meet and even to recoup their initial, potentially substantial, investments.

What separates winners and losers in the game of people analytics?

There are a variety of people analytics styles to choose from – and the approach a business decides to adopt can determine whether they fail or thrive in their implementation of people analytics.. The four I’m going to consider here are:

1. Infrastructure Obsessives
2. Reactive Data Waiters
3. Data Miners
4. Proactive Business Analysts

1. Infrastructure Obsessives

You know those people who never seem to get any work done because they spend all their time rewriting to-do lists and playing with new time-management methodologies? Infrastructure Obsessive people analytics functions do pretty much the same. Instead of just knuckling down and doing some people analytics, they spend most of their time (and money) on…

• Governance: Setting up interminable governance structures for projects they’ll probably never run
• Stakeholders: Meeting ‘empathetically’ with stakeholders whose business they don’t understand and with whom they’ll probably never engage with again
• Data Privacy: Enriching their lawyers by planning for data privacy contingencies, most of which would require – in insurance parlance – 20 consecutive Acts of God in a 24-hour period
• Data integration: Spending months (sometimes years) sucking useless data from all their global databases and spreadsheets into one place; then discovering that it’s mostly out of date; then cleaning it for a sum that even would make even Giorgio Armani blush; and then eventually using maybe only one tenth of the data
• Technology: Evaluating technology after technology before finally investing an amount equivalent to the GDP of a small state on the ‘perfect platform’ – and again, only ever to use one tenth of its capability
• Conferences: Attending endless conferences and then meeting with swarms of ever-hopeful consultants to discuss dozens of methodologies that they don’t even understand, let alone ever use
• Consultancy: Spending more time with that high-end management consultancy than the average Fortune 500 company devising a people analytics vision, mission and objectives, which they’ll never end up implementing

And finally (but only if there’s any time left)…

• People Analytics: Waxing on ad nauseum about the value they could generate if only the above activities left them some time to do some people analytics

In other words, Infrastructure Obsessives spend so much time and money building an infrastructure that there’s seldom any time left for doing any decent people analytics, let alone profiting from it; and, in some cases, their bosses eventually curtail their people analytics investments altogether because of the high resource consumption compared with the low returns.

Of course, no one is suggesting that infrastructure is a bad thing.But people analytics – like any organisational activity – is about balancing risk and reward. If you go overboard with risk, you never end up collecting a reward; nor will you make real money from your investments.

There are many possible reasons why people become Infrastructure Obsessives. For starters, they are often the victims of consultancies that emphasise people analytics infrastructure but who themselves lack the experience to deliver meaningful analytics.

Another cause of infrastructure obsession is CHROs whose people analytics skills are limited, but who realise that ‘one needs to be visible in people analytics nowadays’.

For these individuals, infrastructure obsession is a useful deflective measure because it makes them appear to be busy with people analytics while in reality they’re not budging out of their comfort zones.

Finally, infrastructure obsession is sometimes caused by an excessively risk-averse corporate culture. In these cases, people analytics is seldom the only function to be affected by this aversion to meaningful activity.

2. Reactive Data Waiters

Data Waiter people analytics functions start out similarly to Infrastructure Obsessives, but they at least make some money by actually using their infrastructures to do some ‘analytics’. Their issue is that the analytics they deliver are primarily just reactions to user requests for simple data reports which could be taken care of with a little self-service and end-user training.

However, even in this reactive scenario, Reactive Data Waiters could add some value by asking users why they need the data in the first place, rather than simply just handing it over. Typically, user responses are:

1. “My manager asked me to get the report but I don’t know why she wants it”
2. “We’re concerned about engagement, retention, absence, cost of recruitment.” In other words, they require the data to address a people process or workforce capability issue (see Levels 3 & 4 in figure 1 below).

In both of these cases, people analytics professionals could adopt a more value-added approach by trying to determine whether there is an associated business issue (of the kind found in Levels 1 & 2 of figure 1) on the basis that there’s not much point in trying to address Level 3 & 4 people issues if they aren’t causing any business problems.

A Data Waiter mentality is usually more commonly found in people analytics professionals who understand the HR language of Levels 3 & 4, but are uncomfortable with the business language of Levels 1 & 2. The fix is usually a commercial education programme.

While Reactive Data Waiter people analytics functions probably manage to keep the wolf from the door, there is a huge opportunity cost because they don’t come close to the potential of what might be achieved if they adopted a more proactive people analytics approach as described below.

Figure 1: Human Capital Value Profiler Framework

3. Data Miners

Data Miners focus on the technology components of people analytics infrastructure and – in particular – on bringing all their corporate data and any big data they can find into single virtual cloud. This is so that they can systematically trawl through it looking for spurious relationships (correlations), which may (or more likely may not) exist.

Sophisticated Data Miners use specialised tools such as RapidMiner or SAS, while amateurs bring to bear the skills they gained at that one day visualisation training course (you know the one) to use Tableau, Business Objects or Cognos to create a gallery’s worth of graphs with artistic imagination that even the Louvre would be proud of.

While visually compelling and imaginatively eye-catching, they add approximately zero value to their employer’s human capital decision-making process.

This is why professional data analysts refer to data mining as data fishing – because even if Data Miners do happen to stumble across significant relationships in their large data collections, they’re more likely to be meaningless coincidences rather than real Level 1 or 2 business issues.

To illustrate this point, here are some examples of ‘relationships’ discovered by Data Miners, which include gems such as employees who use Firefox or Chrome are better employees…

As far as I can tell, there are only two kinds of company that ever make money out of data mining:

1. Companies that have solved all their business problems and happen to have spare data mining resources (there can’t be many of those around)
2. Companies whose business model revolves around data and advertising such as Google and Facebook

4. Proactive HR Business Analysts

Finally we come to the rather small group of companies who get real value from their people analytics investments.They do this by proactively seeking out high-value Level 1 & 2 KPD and business opportunities (see Figure 1) before even thinking about building a people analytics infrastructure. In other words, they go straight for the money with Level 1 & 2 projects like:

1. Productivity improvement: Identification and automation of low-level repetitive tasks based on optimisation modelling – determining their company’s workforce productivity drivers and then using employee-centric training to deliver these
2. Enhancing customer depth and share of wallet: Scientific profiling and recruitment of high potential customer-facing personnel and salespeople; and again, the delivery of training based on analytically determined competencies
3. Innovation enhancement: Application of people analytics to identify where their company’s corporate culture is inhibiting competitive advantage and then designing interventions to remedy these issues

This is not to say that Proactive HR Business Analysts don’t fulfil also a Data Waiter function providing simple user reports and data as needed: it’s just that they view reactive data-waitering as a people analytics hygiene factor rather than a core activity.

There are three other points worth noting about the Proactive HR Business Analysts:

First, they view Level 3 & 4 workforce capabilities and people practices as vehicles for improving their company’s KPDs and business outcomes, rather than as ends in themselves. In other words, they don’t seek to increase engagement or retention for its own sake unless there is a concrete business issue attached to it.

Second, Proactive HR Business Analysts focus primarily on predictive analytics rather than simple reporting. This makes sense because profitable people analytics is the result of predicting which problematic Level 3 & 4 people processes and workforce capabilities are causing missed Level 1 & 2 targets. Once identified, action can be taken to repair the faulty Level 3 & 4 predictors.

Finally, Proactive HR Business Analysts are agile: they operate on a shoestring by favouring rapid desired business outcomes over an exaggerated need for people analytics infrastructure. They then put some of the resulting income towards developing a people analytics structure.

In other words, each successful people analytics project funds the next stage of the infrastructure. This is not to say that they don’t also carefully review all the legal requirements as they go along. But it does mean that, in general, their technology and data integration costs are minimal.

The money makers are in the minority

In summary, the only companies making real money out of people analytics are those that proactively seek out business issues and use predictive analytics to address them.

These people analytics functions have an intimate understanding of the trade-off between investments in people analytics infrastructure and proactive action and are experts at balancing this risk in order to earn a handsome dividend.