13 min read

Welcome to People Analytics, Are You Sure You Want This Job?

“Men* wanted for hazardous journey. Low wages, bitter cold, long hours of complete darkness. Safe return doubtful. Honour and recognition in event of success.”

Serious question. If you stumbled on the employment ad above, would you respond?

The ad is purported to have been posted by Ernest Shackleton to recruit people (“Men*”) for his Endurance expedition to the South Pole. There is some debate whether the ad was actually written by Shackleton, never-the-less whoever wrote it could get credit for compelling ad copy, as well as possibly the first example of a realistic job preview.

*This was around 1914, and apparently whoever actually wrote this was unaware of woman’s interest in suffering in the workplace for 25% less than men! This reminds me of a pernicious glass ceiling and wage disparity problem – and now suddenly taking a little boat down to the South Pole doesn’t actually sound so challenging to me. Lots more I can say about the journey into problems of diversity and gender, but we will circle back to HR data and diversity another time. For now, let’s just continue with our juxtaposition of the cold, icy cold, Antarctic and HR Analytics, in general.

A Brief Summary of The Expedition (Wikipedia): 

Endurance became beset in the ice of the Weddell Sea before reaching Vahsel Bay, and despite efforts to free it, drifted northward, held in the pack ice, throughout the Antarctic winter of 1915. Eventually the ship was crushed and sank, stranding its 28-man complement on the ice. After months spent in makeshift camps as the ice continued its northwards drift, the party took to the lifeboats to reach the inhospitable, uninhabited Elephant Island. Shackleton and five others then made an 800-mile (1,287 km) open-boat journey in the James Caird to reach South Georgia. From there, Shackleton was eventually able to mount a rescue of the men waiting on Elephant Island and bring them home. On the other side of the continent, the Ross Sea party overcame great hardships to fulfill its sub-mission. Aurora was blown from her moorings during a gale and was unable to return, leaving the shore party marooned without proper supplies or equipment. Nevertheless, the depots were laid, but three lives were lost in the process.

Clearly the expedition failed to accomplish all of its objectives – yet it is recognized as an epic feat of leadership, endurance and one of the last of “great expeditions”.  

Some time ago I came across the ad, which tickles my sick sense of humor and my imagination. Who would take this job.  I wonder - who wouldn’t take this job? In my mind I notice subtle similarities to the choices I have made in my career – which circles on the question how to change Human Resources, if not the direction of work as a whole, with data. Despite difficulty and a seeming complete lack of possibility for fame in this work, I am excited by what I do and I love every minute of it. I probably would have responded to the ad.

I am also intrigued by others who take an extended interest in my strange field -  partly because historically extended interest has been rare and partly because it is far from what most would consider a "rewarding career".  I have been tracking new roles in HR Analytics, Talent Analytics, Workforce Analytics (what I call People Analytics) through Job Postings off and on for a long time. More recently, I have been searching for these people on LinkedIn. It is a question that begs to be answered – who are these people, what is their story, what do they care about, what do they want to achieve, how are they received within their HR organizations, what are they working on, what are their struggles?   Those questions are a work in progress.  For now, the fresh faces and backgrounds I see in these roles in pictures and Linkedin headlines is already truly inspiring to me. They surpass my imagination in magnitude and breadth and to me represent "pure energy". I think the people who are drawn to these roles are special and amazing. I have no doubt that as a collective they/we will ultimately work through the difficulties to reach our destination (Destination? This too is a great question). “Honour and recognition” hopefully forthcoming. In real life who would respond to an ad like Shackleton’s? I think a high proportion of those people today would be these People Analysts. 

The Vast Expanse of Human Resources

The real HR is vastly underestimated and mostly lost on most people without direct exposure to a leadership role in HR in a large modern organization. By large I don't mean 500, I mean 50,000. HR doesn't really come into the spotlight until you reach 5,000 employees. At this point, you start to realize the inherent value in getting HR right - that is, if it is not too late for you! If you reach 50,000 employees you probably have figured HR out and now you preach it, never-the-less you are in a world of much bigger HR issues. If you are in a leadership position in HR for one of these companies few people know your precise struggles. It is a lonely world.  

If you could peer inside the little windows you would see that HR in large modern organizations is a complex network of technology, policy, process and people within discrete areas of specialization. HR can be grouped broadly in categories of Staffing (Sourcing, Recruiting, Onboarding), Total Rewards (Compensation & Benefits), Diversity, Talent Management and Organization Effectiveness (Performance Management, Succession Planning, Organization Design), Training and Development, Employee and Labor Relations, and HR Law & Govt. Compliance.It is a veritable alphabet soup of things to learn, with an entire language of its own. Sounds crazy, but yes, it is a diverse world of very real and very different things, all in this thing we call "HR".  If you want to dive deeper you can find more granular silos of responsibility, but I will refrain from the arcane details and just leave it at the high level.  I always forget - how many ways can eskimos describe snow?, and why?  I don't know the answer, however I can surmise that they must spend a lot of time with the stuff.

 This is a good place to pause – are you still with me?

Here is a more down to earth human story for you. Several times I have spoken to recruiters (an actual real life role in HR), and as I attempted to describe the complexity of tieing out data from many HR sources to some meaningful business conclusion and what that actually requires I have been interrupted with a reply something like - “It is nice you have had exposure to all those other things, but to be clear in this role you will deal exclusively with HR data. Are you o.k. with that?” With HR data? After only just having described a variety of HR data sources, not really even making the point about how these sources should be tied to business data, I am perplexed at who and what I am dealing with here. I am left to assume that by HR, they mean Staffing, a subset of HR and the rest of it is lost on them.  Clearly, per their guidance it will also be lost from me and the organization if I take this job under these conditions. Do I correct them, or proceed? I am sorry to be blazingly disrespectful, but I have been tempted to stop right there and say, “Hello, Hello, are you still with me?" Should I proceed ahead or turn back?  

O.k., for now I will spare everyone the graduate level course in Human Resources but if you would like one they offer specialized programs in this stuff at the University of Minnesota, Cornell, Rutgers, University of Illinois, USC, somewhere in Michigan, and a few other schools!  Some of the world's most respected CHROs of the largest organizations of the world came from those programs - I’d be happy to make introductions. By now you are rolling your eyes at me. No problem, I’m used to this. I am not sure what other people were doing when Shackleton was hacking away at the ice that was destroying the hull of his ship - I am sure his choices in life made a few eyes roll as well. 

Leadership and decision-making for HR sub-functions is distributed among members of the HR Leadership Team (HRLT). In large organizations the HR Leadership Team generally consists of 7-10 Sr. Directors reporting to a “Chief Human Resources Officer”(CHRO, SVP of HR, etc.). The Chief Human Resource Officer may have leadership, decision-making authority and budgetary control over the HR function, but in some cases the big budgetary decisions may go all the way up to the CEO and/or the board.

Staff/Employees/Humans typically represent a spend of 50-85% of most modern organizations revenue. Usually over 75%.  Stop - check it out in the annual reports, these days available on the internet. These are real numbers. Where is the profit? It is the little bit that is left over after all those people are done with their work and paid. The shareholders are buying a share in that little bitty slice.  "Our most important asset".  Have we looked at how we do HR with data - not much, not really - we know it is somewhere down there - it's a line item.

In most advanced organizations the head of HR will report to the CEO, but often they report to the head of Operations or Finance – this varies by company. It is usually a sign post – a little piece of ice floating – be sure to keep an eye on it, something big may actually be under this ice. Regardless, wherever they report, it is contingent upon the CHRO to drive alignment of HR goals with “the business” and alignment between HR sub-functions on a central strategy and operational architecture for Human Resources.

Seen as a “support function” annual HR goals and budgets typically FOLLOW development of business strategy and goals. HR goals are assembled in a last minute hurried manner sometime AFTER all the other business functions have had their shot at the podium. If those others strategies and goals are not yet fully clear, or the HR/People consequences cannot yet be ascertained with the information on hand, or the HR leadership team fails to communicate the people implications of the business goals, then HR will follow the business in a misaligned and disheveled path. In my experience, these are accurate actual descriptions of HR goals. Whether or not the HR team is able to formulate a coherent plan HR has last shot at organizational resources - consequently HR efforts are often unclear, underfunded and unrealistic (relative to the magnitude and difficulty of the goals). 

Examples of strategic HR objectives include: “Improve Employee Morale”, “Improve Employee Engagement”, “Reduce Employee Turnover”, “Change our Company Culture”... Hey, a big challenge excites me, but FYI these population level averages are resistant to dramatic changes. I’m not saying they don’t matter or can’t ever change – the polar icecaps are melting as we speak - I am saying moving these things in a different direction than they want to go requires serious attention to detail, resources, teamwork and commitment. It starts with data – let’s try to have a look at it together.

Before Data, HR Systems

Historically HR has vacillated between a single system ERP (Enterprise Resource Planning System) or HRIS (Human Resource Information System) that is average at everything or a series of “best of breed” applications that are better at a single purpose functions (Applicant Tracking Systems, Performance Management Systems, Compensation Systems, Learning Management Systems, Time and Labor Systems, etc…) Since HR CHROs and/or critical stakeholders in HR change over every 2-4 years you can guarantee exchange of favored technology to solve whatever shortfalls exist. Meanwhile, within a few years of implementing a solution we have new leadership and who prefer a newer, more optimistic path.

Never wanting to get trapped in the ice, there is constant pull towards fragmented systems stemming from HR sub-function implementation of “best of breed” applications that align with HR sub-function operational objectives. If I am the head of Staffing and my team is tasked with improving Staffing, why would I settle for anything less than the best Applicant Tracking System? If I am the head of Compensation and my team is tasked with getting a grip on Total Rewards, why would I settle for anything less than the best Compensation Management System? So on and so forth. There is nothing wrong with this, however it results in a lack of technology optimization across sub-function silos and some important data consequences that must be addressed before or during reporting and analytics.

Larger, established HRIS systems (PeopleSoft, Oracle, SAP, Lawson, and recently Workday), have more dependencies to worry about and thus are inherently slower to innovate than the collective of single purpose systems. The core systems will always lag in one or more sub-function operational areas. For example, even 10-15 years ago you could facilitate the Performance Management on SAP or Oracle HRIS, however Success Factors came along and offered a solution that was designed for this and consequently much better at it. So for 5-10 years Success Factors took the HR market for Performance Management, creating a hundred million dollar market and adding yet a new system category for HR to manage and integrate. Success Factors was later acquired by SAP (for 3.4 Billion) , and to this day can still be purchased by your organization as a stand-alone Performance Management application, with or without SAPs HRIS product.

To this day most of the organizations I have worked with or for have NOT fully integrated Performance reporting between Success Factors and their HRIS. They prepare for the process to begin by manually setting and loading a file. There is someone deep within the bowels of the organization doing several spins on this. Probably also with a smart phone going off widely in the middle of dinner with the family - if this person even gets dinner with the family for a  month or two of their life. Getting it out and rejoined to a changing organization and broader analytical purposes is another thing entirely.  You will then discover, you can't export all the data you want directly from the same report, and there may not even be a common key!  You might ask, “You mean to tell me we have no efficient method to join and report this data that is central to HR and to the business? Are you kidding me?" No.

We have seen the same in Staffing, Learning Management, and Compensation. Compensation is especially bad - I like to call Compensation Planning (an annual event at every large company) a "planned emergency".  99.9% of the problems is embed in systems, 100% of it is caused by choices made by humans.  By virtue of the many simultaneously moving fronts of sub-function innovation HR will undergo constant fragmentation and change in systems. Innovation in HR systems is good for us, but it is also very disruptive.

Let’s Talk About HR Data

Contrary to widespread belief HR actually has a lot of data, and a lot of good data. It is just locked away in systems not designed at all for reporting or analysis. Universally, HR systems are designed to facilitate operational processes and while they can perform their intended operational functions well (by design), they often do so at the expense of reporting and analysis.  I will take a bucket of ice water over the head if someone shows me a single HR system that can perform a chi-square or binary logistic regression out of the box.   How about any statistics beyond addition, subtraction, multiplication and division . We are waiting. Still waiting.

Frankly, I like to keep my statistics and data visualization software options open – I don’t want my HR system to do everything for me, but I do need to be able to get the data out of my HR systems. Often even this seemingly basic task (get the data out) is difficult. Apparently, nobody thought of about getting the data out. If you are laughing, stop laughing. I’m not kidding.

We can’t get deep insights or complete picture of the story until we get the data out of multiple systems, join it, and have a look at it through applications designed to work with data for reporting and analysis. Most HR “Analysts” are cobbling data together inefficiently with manual, undeveloped or broken data process. Some 80% of the effort of HR Analyst efforts are attributed to manual augmentation of non existent or inefficient data workflow. They spend very little time on actual science, statistical analysis and presentation of analysis. I have been an HR Analyst in one form or another for over 15 years, I speak to HR Analysts weekly, and I have seen the surveys. 

HR data are/is complex because people and organizations are complex and the sub-functions to support HR (as described at high level in above) are varied. There is hardly any similarity in reporting between Compensation and Benefits, let alone Staffing, Training or Employee Relations. Who owns Turnover reporting? Who owns the Employee Survey? These come from completely different systems, with different data structures and different metrics. The desired metrics can escalate into the hundreds and with variations in the thousands. Most HR metrics are compound and with dozens of potentially relevant dimensions to monitor. 

Let's take a look at a very commonly produced and seemingly simple HR metric: “Employee Attrition/Turnover”. This metric is formulated as a compound calculation (Time Period Exits / Time Period Average Headcount). Now before you shout "Eureka! - I have it - you just divide this number by this other number" keep in mind that you will need to calculate this by segments along multiple dimensions: location, division, business unit, tenure, grade/pay group, performance group, age, ethnicity, gender, etc.… You may have 50 locations. Some locations have 5 people and some have 15,000 people. What if you want division by location by gender. Also keep in mind that in annual form the denominator (average headcount) requires 12-13 or more data points for each of subset of each dimension, and these subset counts must all align in definition and time period consistently with the numerator.

Also keep in mind that organizations can be expressed in different ways (people reporting relationships or cost center reporting relationship, which do not typically match), and that the only constant is that organizations are constantly changing.

Imagine a data set that changes simultaneously along multiple dimensions over time and you are trying to report on this consistently over time to demonstrate a "trend" or "story" - how exactly does this work?

If you are flabbergasted, don't worry, most of us People Analysts can work this problem in our sleep.  This isn't even the hard stuff. I just described the calculation of one metric, now multiple this effort by 20 more metrics with data sitting across different systems and try to derive some meaning from that. Shouldn't we be analyzing this stuff together to tell a story, not independently?  

Most of the people doing good work on employee turnover have long ago moved on to more advanced ways of looking at exits (logistic regression, survival analysis, hazard curves, predictive models, etc.) and incorporate data from many many more data sources. I can ask employees three questions and based on their answer tell you if their chance of leaving in the next year is "about average, 2x average or 3x average" without knowing anything else about them. Give me their job and salary and tenure and a few other details and we can clean up on this problem.  Let me join it to performance and compensation and now you can decide carefully how you want to distribute the limited budget you have to work with.  Why not?  If you are not working on this, what exactly are you working on?

Speaking of clean up. For most organizations, you will get started on a seemingly simple problem, and bang into an inconsistency or perceived data quality problem, which requires people sitting in another sub-function you have no authority over to make a decision, fix a process, or change how they do something. This is not simple any longer – this is really difficult. You can circle on the same issues for years. Welcome to People Analytics, Are You Sure You Really Want This Job? 

Then There Were People

HR professionals are not selected for a background in technology, math or science and so depend on people outside of HR for augmenting support in these capacities. Since HR is seen by “the business” as lower in priority than other business functions (software engineering, finance, marketing, sales, etc.), and therefore not a very prestigious appointment for anyone - IT and data science support suffers. The head of HR is on the phone about something they want and the head of Sales, which call should I take? If HR is already underfunded for its goals, HR IT is even worse, not receiving adequate talent pool or funding. Budgets are divided among the heads of functional HR silos and so no unifying technology solution can be reached. There are simply too many different jobs to be done in too many different systems. Data architecture for this is an after-thought.

Each Sr. Director will compete for the time, attention and resources of whatever HRIS or HR Analytics professionals exist to try to achieve their objectives first (at any and all cost to others –their performance rating is on the line!). If they cannot get that attention they want from IT or from their HR Analysts they will try to go outside the organization for the support.  Talk about being crushed in the ice.  You will be loved and despised, sometimes consecutively, sometimes simultaneously. How this can happen is one of the great mysteries of the universe - outside of light being possible to be described both as a  particle and wave.  It is sort of like the possibility of dying of thirst while standing on water (ice). Input fire and maybe drown. Now I am being dramatic. This is just to say that I think Shackleton and company did a pretty good job, given the odds.

I have had opportunity to be in these data oriented HR roles for variety of very well respected organizations (Merck, PetSmart, Google, one of the best Children’s Hospitals in the country to name a few) - believe me when I say these HRIS and HR Analytics folks have many more “bosses”, a combined list of objectives much longer, technically more complicated, and with less resources than anyone else. There high turnover among HRIS and HR Analytics professionals for a reason. You got one. Great. I hope you have a backup plan!

Other People Analytics posts by Mike West


Who is Mike West? Mike has 10+ years of experience building People Analytics from the ground up at companies such as Google, Merck, PetSmart, Children's Medical, Jawbone and other places.

Mike's passion is to develop thought leadership in HR and to cross pollinate the frameworks and processes he helped develop and pioneer as an employee at these places. 

Mike spends most of his time teaching, coaching and writing on all things People Analytics.  

Does People Analytics Work? #PAQA

7 min read

Does People Analytics Work? #PAQA

“Half the money I spend on advertising is wasted; the trouble is, I don't know which half.” Henry Ford, Lord Lever, John Wanamaker... People often...

Read More
The Work of Freedom is Not Finished : People Analytics and Diversity.

5 min read

The Work of Freedom is Not Finished : People Analytics and Diversity.

January 18th, 2016 In the 53 years that has passed since Dr. Martin Luther King Jr. delivered his I Have a Dream speech America has clearly come a...

Read More
Why you should consider joining a people analytics peer group.

Why you should consider joining a people analytics peer group.

Earlier this year i joined one of The Learning Forum's workforce analytics peer groups and i wanted to share my experience in attendence and why i...

Read More