Ten Counterintuitive but Unifying People Analytics Design Principles

It's easy to get lost in the forest when you focus on the trees. It's a danger you face with people analytics, only because people analytics by necessity looks close-up at data and tries to determine patterns and lessons that allow you to make a positive change that impacts your business. You must be careful to focus on the task before you — to create positive change. In this article, I share tidbits of wisdom, learned the hard way, about maintaining that focus. I list them here so that you may watch for these patterns and, when you get stuck, have a resource to check against to make sure you’re taking into account the big picture.

The Heart of People Analytics Isn't Metrics — It's a Problem Solving

It's evident that the analyses performed by people analytics require calculation, but the calculation is neither the first step of people analytics nor its goal. Calculation is just one step among many to an end — and that end is the solution to a problem that provides useful measurable improvement to a human system.

People analytics will inevitably require calculation, but the exact nature of those calculations is determined contextually; for example:

» Maybe you’re helping the company figure out which hires to bet on with a limited budget.

» Maybe you’re helping the company figure out the best ways to keep its highest-performing salespeople without giving away all the profit they produce.

» Maybe you’re helping the North Pole company get 20 percent more productivity out of its elves by carefully addressing elf needs.

Though some calculation needs are generalizable, I don't know what problems you need to work on now or the strengths, weaknesses, or unique proclivities of your specific elves that suggest an answer. Even if I did know the answer, these particular proclivities might change from year to year.

REMEMBER: The purpose of people analytics isn't to prepare the same calculations at different companies — one time or all the time. The purpose of people analytics is to help the company calculate the probable answer to dynamically changing people's problems.

Every Analysis Based On People Analytics Is Both Familiar and Unique

Every people-related problem you encounter in an organization has embedded within it some elemental issues that are analyzable with familiar tools — standard methods of HR data extraction as well as typical survey-based data collection and statistical tools. At the same time, each organization's people-related problems are unique. Did the company bring on this problem itself, or did a change among its talent competitors bring it on? Is it a "hiring-quality" problem, an issue of "what to do with the people once we get them inside," an "attrition" problem, some combination of all these, or something else? Though one company may be fighting its way into a new market and experiencing growing pains, as a result, another is trying to get the most out of what it has in a stable industry. In another example, results may be spoiled by a few bad managers, and somewhere else, the company as a whole isn't good at managing. The permutations are endless. It's okay to build a toolbox of familiar tools, but novel development and application of those tools are essential.

Inside Every Business Problem Is a People Problem Struggling to Get Out

If there’s one thing I’m sure of from my career in business, it’s that if you have a problem and ask why that problem exists enough times and in enough variations, you will eventually find people there waving back at you.

Here are some simple examples that don’t even require asking “Why?” more than once or twice:

» The product frequently breaks: Okay, who designed it?

» You can’t produce profits: Okay, where do people fit into your business model, and where in the world are you sourcing them?

» New equipment causes production delays: Okay, were the workers involved in the equipment selection, and were these same workers fully trained on the new equipment?

» Nobody wants the product your company spent ten years developing:  Okay, who’s working in product strategy and marketing, and what went wrong?

» Sales are flagging: Okay, are the employees that your customers interact with miserable to be around? If you go on for hours in an investor conference call about the quarterly numbers and never spend a minute talking about what you’re going to do about the sad interactions between miserable employees and miserable customers, you have a problem.

Whenever you encounter a business problem, try asking “why” consecutively without eventually coming back to a people problem. If you find a situation where you don’t get to a people problem in fewer than five “whys,” mail me the context, the problem, a way to contact you, and a self-addressed stamped envelope. If we can’t figure out the people part of it together, I’ll send you $10.

The Actions of People Are Influenced by Forces (Seen or Unseen) That Operate as Vectors

A force can be expressed graphically by a vector. A vector’s length is its magnitude, and its direction is given in relation to the x, y, and z axes.

Every company has a natural attractive force that can be measured by the number of job applications it receives. The degree to which that force is insufficient to produce a specific quantity of candidates who meet the hiring criteria determines the additional energy that must be applied to acquire those candidates through active recruiting. At the same time, the attractive force of other nearby companies operates to make your employees into your competitor’s applicants.

Employees may be simultaneously attracted and repelled by different forces related to your efforts and the efforts of other companies — many of these forces can be measured mathematically and understood as competing influences of the choices people make.

Understanding more of these vectors can help you make the behavior of people more predictable and influenceable. For example, the abnormally high exit rate of unpromoted, high-performing managers in their third year of tenure may indicate that a nearby competitor has identified a vector that you have not — a vector that represents the employee's aspiration or need for career growth. If you can't satisfy the employee's needs, someone else may.

The more vectors you understand, the more predictable and controllable the behavior of the object.

A Company Receives Forces, Experiences Stresses, and Exhibits Strains

Force, stress, and strain are physics terms, applied more frequently in the context of physical engineering than a business; however, they apply equally well to the condition of a business.

A force is something outside of an object that acts on the object, causing it to change speed, direction, or shape. A physical example of force is the snow load on a rooftop: Given enough force, your roof could collapse. An example of a force on a company is a new competitor. An established restaurant experiences force on its business as a new restaurant opens right down the road. The nature of that force is determined by how exciting the neighboring restaurant concept is and how well it is executed.

Stress is a measurement of the impact of the force on the object. In the real example of snow on the roof, stress may be expressed as the estimated pounds per square inch of snow. In the case of the restaurant, stress may be measured in the increasing anxiety of wait staff, commensurate with the vector of the hype of the next-door restaurant opening. You may think the wait staff is unnecessarily nervous, but they live on tips. The level of stress could be picked up by survey or from the scrutiny of informal conversation.

A strain is the product of stress. In the real example of snow on the rooftop, a strain may be measured as the observable deformation or change in an object. In the case of the restaurant, a strain may be measured in the retention rate of servers. With all objects, there's a point at which they break under strain. Careful observation and response can help you address the changing forces that impact your workforce situation and influence the performance of your business.

Know Explicitly What Change You Want to Create and In What Direction You Want It to Move

When a force acts on a measurable attitude or behavior, a metric representing the attitude or behavior will either remain the same, increase, or decrease.

Different types of analysis can be applied. Your objective with people analytics could be to understand those things that create a change, clarify a change, or predict a change. One analysis may seek to illustrate whether something is changing and, if so, by how much. Another may try to show what is driving the change and what is not. Yet another may seek to understand whether a specific change influenced an outcome of interest. A final analysis could attempt to use all the historical information to predict future results.

Different methods are available to go about carrying out this analysis. Experimental Design seeks to control the context by isolating a change among one of two different randomly selected samples to determine with a certain threshold of confidence whether the change moved the measure you’re interested in compared to a segment that didn’t experience the change. A big data approach seeks to isolate the impact of the force by measuring every variable you can get your hands on. A big data approach may determine what matters using either multiple variable statistics (like multiple regression) or other methods (like machine learning).

Using whatever design is best, your first objective should be to determine whether there's a change in the outcome; from there, you need to determine what was responsible for that change before testing ideas about how you can influence outcomes. If you don't know what the result should be or what the probable influencers are, it doesn't matter what method of analysis you use, because it's more likely than not that nothing of real benefit will come from it. Again, this raises the most critical question, "What specifically do you want to change, and in what direction?" If you have no specific objective, you're unlikely to gain any insights that are in any way persuasive or compelling.

Segmentation Choices Can Both Illuminate and Darken Insight

Consider a company that has an overall employee attrition rate of 10 percent. Ten percent seems reasonable and may not suggest that any action is required. Now consider that if you took this same company and reported attrition by division, you may find one division with a 5 percent attrition rate, one division with a 10 percent attrition rate, and another division with a 20 percent attrition rate. Or, if you reported it by location, the range is wider, with some units even above 30 percent attrition. While the overall attrition rate seems okay, some smaller subsets of the whole are outside the range of what you determine to be good. The variability of smaller units can go unseen when aggregated into larger groups and left unreported. Applying a large number of relevant segments to a metric can help illuminate hidden variability that represents strengths or weaknesses that would otherwise go unseen. In this way, segmentation choices are fundamentally crucial for producing insight.

On the other hand, if you have reported attrition in a thousand different ways and you share this information with others, they may not know what to do with it, and, in any case, they may never get to the end of it. Analysts must do their best to get to the right level of segmentation. If you haven't included the most important segments in your analysis, they will be missed. The nature of this problem represents a conundrum for analysts, which is why I propose starting with theoretical models. Learn more about Models in Chapter 10.

Accuracy and Precision Are Different Things

Accuracy is the absence of error; precision is the level of detail. Effective analysis always requires accuracy, but precision should be guided by what is useful to communicate a defensible finding. Early in the problem-solving process, accurate but imprecise methods, rather than precise methods, will help to minimize the cost of pursuing unnecessary detail that would not change your decision or aid in communication. This understanding can help you head off at the pass any wasted activity that prevents you from doing more useful things.

For example, if you know that 20 people left the company, it isn't much difference if you calculated the attrition rate from an average headcount assumption of 99, 100, or 101. Your attrition rates would be 20.20 percent, 20.00 percent, and 19.80 percent, respectively. Each can be rounded two decimal places to 20 percent and provide the same conclusion. Accuracy represents the nearness to the actual number, which you know to be somewhere between 99 and 101. You cannot determine the answer more precisely without careful investigation. The +.20 to –.20 difference in the attrition rate represents the error of precision. Improving the precision wouldn’t change your interpretation of the 20 exits, so increasing the precision isn’t worth the effort.

REMEMBER: This example is simple. It can get more complicated the deeper you get into sampling techniques. A set of data can be said to be precise if the values are close to each other, whereas the set can be said to be accurate if their average is close to the real value of the quantity intended to be measured. A particular set of data can be said to be either accurate, or precise, or both, or neither because the concepts are independent of each other. The good news is that science and statistics are intended to address the problems of both types. 

Quantification Is Approximation

People analytics is contained within the boundaries of current math and science, but reality exceeds these boundaries. As a system for understanding reality created by humans with imperfect instruments of observation, science both observes and is contained within the same reality. Reality is the reality; science is an ambitious but incomplete attempt to explain reality. These should not be confused. Quantification is an imprecise representation of reality for analytical purposes in the service of problem-solving. You don't have to debate reality if you can still solve problems.

REMEMBER: You cannot see, smell, touch, or hear concepts such as happiness, engagement, motivation, and commitment. However, you can define these concepts as a series of statements that employees can react to.

When you measure happiness, you don't know ultimately whether you have captured a perfect representation of joy by the statements you provided, or even whether happiness exists; you can measure your idea of happiness based on an agreed-on definition. The importance of the measurement is less about determining the ultimate truth as it is about the usability of the information found. If you discover that, by responding in a way you define as unhappy, a person is statistically more likely to disengage from their work or leave the company, then it doesn't matter whether your measure of happiness is imperfect. What matters is the ability of your analysis to explain, predict, and help you control the objective outcomes you care about.

You Don’t Understand Something Until You Quantify It, But You Understand Nothing At All If All You Do Is Quantify

It is now known that at a level of observation below what can be seen in a microscope, the material of human beings is coiled in twisted strands called DNA. The coiling of DNA strands allows 3 billion base pairs to fit in a cell just 6 microns across. If you stretched the DNA in one cell out, it would be about 2 meters long. If you stretched out all the DNA in all your cells, your DNA sequence is something close to twice the diameter of the solar system. DNA represents a current perspective of the biological basis of the human body characterized by the composition and sequence of the same 46 chromosomes.

No one has yet developed a DNA type guide for understanding a company entity at a submicroscopic scale the same way they have with DNA, but given the variety of concepts you can measure and the options for granularity, it approaches infinity. Yet this doesn’t mean you cannot infer a useful observation without a complete measurement of everything.

There comes the point when you should stop analyzing and use your analysis to make some changes in the real world with the imperfect information you have and then see whether it worked. Your objective is not to know everything that can be known; it is to identify the information that is useful to help someone solve a problem and get on their way. The earlier you can identify the problem your company needs to address, the more deliberate you can be about what data is necessary to help illuminate a new path forward. You don't need all the data — you only need specific data. To determine this, you have to get away from your desk and talk to other people. You have to take your first step along the path and then go wherever it takes you.

This is an excerpt from the book People Analytics for Dummies, published by Wiley, written by me. Please buy it.

More on People Analytics For Dummies here

More samples here: Introduction to People Analytics for Dummies, Introducing People Analytics, Making the Business Case for People Analytics, Contrasting People Analytics Approaches, Segmenting for Perspective, Finding Useful Insight in Differences, Making Sense of HR Metrics, Estimating Employee Lifetime Value, Mapping the Employee Journey, Survey Questions To Collect Analyzable Data For Your Employee Journey Map, Attraction: Quantifying the Talent Acquisition Phase, Activating Employee Value, Analyzing Employee Attrition

The beloved tens: Ten Counterintuitive but Unifying People Analytics Design Principles, Ten Myths of People Analytics, Ten Pitfalls of People Analytics, Ten Things to Look for Yourself When You Apply People Analytics to Talent Acquisition

You will find many differences between these samples and the physical copy in the book - notably my posts lack the excellent editing, finish, and binding applied by the print publisher. If you find these samples interesting, you think the book sounds useful; please buy a copy, or two, or twenty-four.

Three Easy Steps

• Follow and connect with me on LinkedIn here: https://www.linkedin.com/in/michaelcwest
• Join the People Analytics Community here: https://www.linkedin.com/groups/6663060
• Check out an index of my writing and other resources here: Index of my writing on people analytics on PeopleAnalyst.com
•