Segmenting for Perspective

In this:

* Obtaining insight through segmentation

* Following the segmentation process

* Transforming H.R. data into segments

Segmentation is a fundamental and essential part of people analytics — or any analytics, for that matter. Segments are crucial when it comes to helping you understand and derive insight from your data. A segment is a grouping of people who share common characteristics. A segment of people can be observed as one whole unit or as a portion of another unit. For example, the people who work together in an industry form a segment of the total job market. The people who work together in a company form a segment of an industry, and the people in that particular company form segments of that company. They also form segments of other categories in other dimensions.

Some simple examples of segments within a company are division, business unit, location, and job function (such as sales or engineering).

People can also be segmented by characteristics that have nothing to do with the company and everything to do with the person; for example, gender, ethnicity, socio-economic status, age, work experience, educational achievement, and personality type.

People can be described by things that stay the same or things that change. Some examples of things that change are years of work experience, company tenure, job tenure, pay, or attitude. There is nearly an infinite number of potential segments for any data set about people because people are describable in many different ways.

<Remember> Using data to observe a group of people with data is like looking at a diamond. The brilliance of a diamond is a function of two different things; the number of cuts (or facets) and the clarity of the stone. Like diamonds, companies can have many facets. The job of people analytics is to increase perspective and clarity by making the right cuts.

In this chapter, I introduce you to segmentation, show you some of the ways it’s used in people analytics, and point you toward the use of segmentation to improve the way you think about the people in your company.

Segmenting Based on Basic Employee Facts

As you explore your company’s various systems, such as enterprise resource planning (ERP), human resource information system (HRIS), and applicant tracking system (ATS), you find many hundreds of different facts about people stored in relational tables. In some cases, administrators input these facts. In other cases, individuals input the facts themselves — self-service, in other words. In other cases, the company actively seeks new facts by distributing surveys or forms. Finally, some facts collect with no deliberate planning — data is just coincidentally accumulated by way of other activities or processes – for example, email and meeting metadata.

<Technical Stuff> Metadata is "data [information] that describes other data [information]". Email or Meeting metadata refers to facts that describe the nature of conversation without getting into the specifics - this metadata is useful for analysis. For example, number of emails sent, number of words in emails, number of unique social connections, number of meetings, average meeting time, the concentration of social connections by job function or location, who are the people that various types of people communicates with the most, who shares more information, and so on.

<Remember> In their primary form, the facts collected as data can seem useless; however, in skilled hands, the facts can be transformed into useful insights.

 “Just the facts, ma’am."

What kind of people data are we talking about? Generally speaking, you can find facts from the following categories of information in one or more of your systems that contain employee data. (I list these facts first and then delve into how to make them more useful.)

Candidate facts

* Name, or the name the candidate answers to

* Candidate ID, which is a unique number representing the candidate

* Source — recruiting method/channel (for example, LinkedIn, job board, referral, recruiter source, university recruiting, recruitment partner outsourcing)

* Source — most recent employer

* Source — most recent university

* Education level, such as diploma/GED, some college, bachelor’s degree, master’s or MBA degree, or PhD

* Requisition ID, or the unique number representing the job opening

* Test score (For example, a technology company I worked for tested potential software engineer candidates for their level of acumen solving challenges with code. A pet retailer I worked for tested store customer service employees for their knowledge of pet topics following new hire training. A pharmaceutical company I worked for tested aspiring pharmaceutical sales representatives for their product knowledge following new hire training.)

* Application date, which is the date of the first contact

* Phone screen date, which is the date on which a recruiter first spoke to the candidate

* Date of candidate’s prescreen employment test, if applicable

* Onsite interview date

* Offer date

  * Offer acceptance or decline date

* Hire date

Basic employee facts

* Name

* Employee ID

* Company start date

* Company tenure (how long the employee has been working at the company based on the company start date)

Job facts

* Job title

* Job code

* Full-time or part-time

* Contractor or employee

* Salaried or hourly

* Temporary or permanent

* Job function, such as sales and marketing, manufacturing and operations, research and development, or general management and administration

* Job management levels, such as executive, director, manager, or individual contributor

* Job compensation grade

* Job start date

* Job tenure, or how long the employee has been working in the current job based on the job start date

* Annual pay

Managerial and financial structure facts

* Manager – this is the name of the manager of the employee

* Next-level manager – this is the name the manager’s manager (usually this is a director)

* Executive – if you look at an organization chart, this is the name of the highest-level management officer; this position reports up under the CEO on the organization chart tree. (Usually, this is a vice president)

* Financial unit, usually the lowest level unit is called cost center

* Next-level financial unit, usually called organization – there are multiple cost centers in each organization

* Division, which is the highest-level financial unit before the company – there are multiple organizations within each division

<Technical Stuff> Hierarchy. A hierarchy is a relational classification system in which people or groups are ranked one above the other according to status or authority. The most obvious example is an organization chart built by observing the combined result of the many reporting relationships. Every company has a different number of levels and different naming conventions for their management hierarchy.

Location structure facts

* Job location

* City

* Country

* Country region, such as northeast, northwest, southeast, southwest

* Global region, such as Asia Pacific, Europe, Middle East, Africa, North America, South America

Core demographic classifiers

* Gender

* Ethnicity

* Disability status

* Veteran status

* Age

* Generational cohort – Baby Boomers, Generation X, Millennials, ...

* Marital status

Demographics are essential for evaluating the changing composition of the workforce, completing the government-required Equal Employment Opportunity Commission (EEOC) reports*, and analyzing your process for unconscious bias.

<Remember> *If your company is headquartered in the United States and has more than 100 employees or a U.S. government contract, you are required by law to file EEO-1 reports with the EEOC.

<Remember> Though demographic information is useful for analyzing patterns, it is inappropriate to use non-job-related personal characteristics like gender, ethnicity, or age to directly make any employment decision, such as whom to hire, whom to promote, or how to pay. In the United States, some laws prohibit making employment-related decisions based on non-job related personal characteristics and specifically providing protection for gender, ethnicity, age, and religion.

 New insights about people are driven primarily by new and more productive types of data about those people. The brave new world of segmentation is psychographics and social. People are cognitively advanced social animals who have minds of their own. To understand and predict human behavior, you have to “see” patterns at the social level and sometimes you have to “see” inside minds — and to do this; you have to ask some questions.

Here are a few examples of the many characteristics you can measure by using survey instruments or tests that can open up a whole new world of insights to you:

* Personality types: Some standard personality instruments are the Big Five, the Myers-Briggs Type Indicator (MBTI), and StrengthsFinder.

* Attitudes: Some standard employee-survey measures are satisfaction, commitment, motivation, and engagement.

* Preferences: A range of topics can be determined using basic questionnaires or advanced survey analysis tools.

* Technology adoption profile: These factors include innovators, early adopters, early majority, late majority, and laggards.

* Opinions: Survey questions can measure the likelihood to recommend the company to friends or to exit the company for a better opportunity.

Look to survey instruments and tests to help you find differences between people that help you understand, predict, and influence behavior. These types of instruments can help you develop segmentation to unlock new insight.

Visualizing Headcount by Segment

In its most basic use, counting people by segment can help you see the company in new ways. For example, the numbers in the graphs shown in Figure 4-1 add up to a company’s total headcount of 3,100; however, each graph paints a different picture of the company based on the segmentation dimension. These are just six among hundreds of possibilities.

Figure 4-1: Different ways to segment the same company.

Analyzing Metrics by Segment

One of the main reasons to bother with segmentation is to provide a finer-grained analysis of the data you’re using to get to the root of a problem. Here's an example that illustrates the power of segmentation: Exit Rate % is a metric that measures the percentage of employees at your company who left to go to work elsewhere over some specified period.

<Remember> The exit rate is synonymous with the attrition rate, termination rate, and turnover rate.

The formula for Company Exit Rate % calculates this way:

(Total # Company Exits / Company Average Headcount) x 100

Company average headcount is calculated by counting the number of employees at the beginning and end of a period and averaging, or by counting the number of employees each day of a period and averaging, or any other consistent period sampling methodology. For example, you can average headcount over a year by weekly, monthly, or quarterly snapshots. The reason to use an average headcount is that if the company is changing (either increasing or decreasing headcount), you get a different answer for what the size of the group is depending on what day you count – average headcount standardizes.

If your company’s Exit Rate % is 10 percent, it means that 10 percent of your employees left to go to work elsewhere in the time frame of analysis.

When viewing Exit Rate % by segment, you calculate it this way:

 = ((Segment # Exits / Segment Average Headcount) x 100)

Segment average headcount calculates by averaging the number of employees in the segment over the period. You are not dividing segment exits by the total number of employees. You are dividing segment exits by segment average headcount.

For example, if a segment called Segment A had an average head count of 100 people over a year and 20 people left in that year, then the Segment A Exit Rate % = 20%, or ((20 / 100) x 100 = 20). In this example, Segment A has double the exit rate of the average, which is, as I mentioned, 10 percent. The differences between segment A and average tells you that something may be going on in Segment A.

Figure 4-2 shows quite clearly the explanatory power of moving beyond mere Company Exit Rate % and looking at specific segments within a company — Region, for example, or Business Function or Last Performance Rating. It lets you see the percentage of people within that particular segment who left the company during that period, not the percentage of the total population of exits.

Figure 4-2: 2017 Exit Rate % by segment.

<Remember> The reason for calculating segment exits as a percentage of segment headcount is that it allows a fair and consistent comparison between segments, regardless of the segment's size. If the calculation is not done as a percentage of segment average headcount, then the larger groups show a higher percentage of overall exits — which would tell you only that these were larger groups, not that there were more people exiting (relative to the groups size), from which you may infer a comparison about exits.

When you report Exit Rate % per segment, you can see how much each segment’s exit rate varies from the company average. You want to know where each segment is in the range of values. You can use segmentation to identify the segments that require more attention, which helps you move the overall company average the most with the least amount of effort.

Understanding Segmentation Hierarchies

In people analytics, you can use many hierarchical dimensions to describe people, such as manager hierarchy, financial unit hierarchy (cost center hierarchy), location unit hierarchy, or job unit hierarchy. The details vary by company.

<Tip> Start with the method of segmenting business units used by your executives for finance and accounting. I refer to this as the financial unit hierarchy, but you may call it something different at your company. Most often, it is called a cost center or business unit structure.

Here’s an example of how a single individual sits in a location hierarchy:

Location hierarchy example

Region = “North America”

     Country = “United States”

         City = “Mountain View”

             Location = “401 Castro Street”

                  Floor = “3”

                     Desk = “401-3-5901” Employee = “John Smith” ID = “11158”

This outline illustrates that there are six possible levels to describe the geographic location of John Smith. The example reflects a hierarchical tree for one person. Of course, there are many more people in a company. You can count the number of people by any of the levels in this hierarchical structure – each level contains multiple segments, and each segment contains multiple people (except for the very bottom). Using this example, you can count the number of people by region, country, city, location, or location by floor if you want to. For example, at your company, you may find only two countries you can count by but twenty different cities. If your company is in two countries, then you have only two segments to compare. If you count by city, then you have twenty segments to compare. From this example, you can say that you can choose many different ways to count, even if we are only talking about location. 

In Figure 4-3, you can see that the company has 2,000 employees in North America, 1,800 employees in the United States, 980 employees in Mountain View, 500 employees at the building at 401 Castro Street, 200 employees on the third floor, and one employee, at desk 401-3-5901, named John Smith.

Figure 4-3: You can keep slicing the pie until you get no more pieces.

As Figure 4-3 illustrates, each person exists only once; however, the same person is in many different hierarchical segment structures. Here are three other ways to describe where John Smith is in the company:

Financial unit hierarchy

Division = “Sales Division”

     Organization = “Enterprise Go-Team”

         Department = “Widgets”

             Cost Center Name = “Widgets – Southwest Territory”

                  Employee = “John Smith”, ID = “11158”

Manager hierarchy

CEO = “Sally Rodgers”

     VP = “Bob Woodward”

         Director = “Chris Henderson”

             Manager = “George Harris”

                  Employee = “John Smith”, ID = “11158”

Job hierarchy

Job Function = “Sales”

     Job Level = “Individual Contributor”

         Job Family = “Inside Sales”

             Job = “Inside Sales Rep 3”

                  Employee = “John Smith”, ID = “11158”

When conducting each analysis, you decide the right level of summarization that’s useful for your analysis. For example, you can count by Division, Location or Job or any combination. All math and science begin with counting. What you count is determined by context and need.

<Remember> When you first begin reporting data, you find lots of inconsistencies between data recorded in systems and what people think in their minds. Without a place to store data, an agreed segmentation structure, and regular reporting, what you and anyone else sees in their mind’s eye is likely to be very different. Counting provides perspective.

In a particular light, the reconciliation between our mind, others’ minds, and the data in systems is the point of analytics.

Creating Calculated Segments

Most of the segments I describe earlier in this chapter exist as a single entry in a database structure: for example, Division = “Sales” or Department = “Inside Sales." In the earlier examples, no calculation is involved in creating a segment — the segment is in the system in the exact how it is.

As you might have suspected, other types of segmentation are out there — more specifically, ones that require calculation. In the next few sections, I walk you through a few examples.

Company tenure

The graph showing company headcount by tenure, as shown in Figure 4-4, looks like data that just come out of a system the way it is. However, tenure is a calculated field that results in continuous data, which wouldn’t look good on a graph the way it is. If tenure is calculated as the number of days since an employee started, the graph would have as many bars as there are people because everyone would have a different tenure. It would, therefore, show nothing useful. However, if you take the tenure calculation and create a new variable that describes tenure as a category characterized by ranges of days (0 to 365 days, 366 to 730 days, etc.), then you can count people that fit within a range of days to produce a graph that has a more useful number of segments to produce insight.

Figure 4-4: Company headcount, with all employees categorized into tenure group categories.

For example, you may want to count the number of people in their first year of employment. What you have in the HRIS is the start date. Tenure can be calculated by counting the number of days between the start date and current date and expressing it as the number of days or years or months. You might find people with 1.1 years, 1.5 years, 1.7 years, .89 years, .5 years, 20.7 years, and so on. Counting by data in this way wouldn’t produce a useful graph, because everyone’s number would be unique. To graph by tenure, you need to count the number of people who fall within a range. Figure 4-4 uses the following five ranges:

Tenure =

• < 1 Year
• 1 Year to 2.9 Years
• 3 Years to 4.9 Years
• 5 Years to 9.9 Years
• 10+ Years

Depending on whether you’re working in Excel or SQL (a database querying language) or another data environment, the formula is different. You also always have more than one way to do anything with data. For purposes of example, to count all people in their first year of tenure, one way to do it is to establish a formula that first calculates tenure and then another formula that counts those who have a tenure less than 1 year. The logic works like this: If tenure < 1.0 years, then assign a 1 or else assign a 0. Then sum. Stepping through the calculation of the number of employees with tenure < 1.0 years, see figures 4-5, 4-6, 4-7 and 4-8 below.

In figure 4-5 below, you see a simplified employee roster exported into Excel, where each row represents a unique currently active employee, and each column represents a data fact about that employee. Column C is the employee hire date.  I have added the formula in column D, =today()-C2, to calculate the number of days tenure between the employee hire date and today. After inputting this formula and hitting enter excel calculates the number of days the employee has worked for the company. You can then drag the formula down into the other rows or use other standard excel functions to apply it to all rows.

Figure 4-5: Calculating employee tenure from Hire Date in Excel

With figure 4-5, you can see employee tenure; however, this is not a perfect end state. In figure 4-6 below, I add a formula to column E, which divides the number of days found in column D by 365 so you can see employee tenure converted into years.

Figure 4-6: Converting employee tenure from days into years in Excel

In figure 4-6, you can see that I have extended column D to all rows and added a formula in column E to convert employee tenure from days into years in Excel. Next, drag the formula down into the other rows or use other standard excel functions to apply it to all rows. Having employee tenure in years is useful; however, you can see that almost every employee has a different tenure. If you were to count by now column E, you would not get a useful table or graph. In this example, what you want to do is count those employees with tenure less than 1 year. What I do in figure 4-7 below is use the Formula Builder to add an if-then statement to column F. What the if-then statement does is to add a 1 to column F if tenure (calculated in column E) is less than 1 or a 0 if tenure is equal to or greater than 1. 

Figure 4-7: adding an if-then statement

Continuing with this illustration, you can see in Figure 4-8 that I have extended the if-then statement to all rows in column F and highlighted some example rows for individuals that have less than 1-year tenure.

Figure 4-8: in column F employees with tenure less than 1 year are indicated with a 1 and all others a 0

Figure 4-8 concludes the example of how you would use start date to calculate a continuous variable, tenure, which you then convert to a categorical variable for purposes of counting. 

<Tip> Assigning a 1 or 0 to a data point changes continuous data into categorical data. A formula like this one may be embedded in the HRIS or in your analytics and reporting environments that assign all employees into a classification segment and then changes this segmentation dynamically as tenure increases.

More calculated segment examples

Here are some fast-and-quick calculated segments that are useful for most employment-related analysis. If you do any work with people analytics, you find that you use these calculated segments over and over again:

Tenure

• < 1 Year
• 1 Year to 2.9 Years
• 3 Years to 4.9 Years
• 5 Years to 9.9 Years
• 10+ Years

Total job experience

• < 5 Years
• 5 Years to 9.9 Years
• 10 Years to 14.9 Years
• 15 Years to 20 Years
• 20+ Years

Generation cohort

• WWII and Silent Generation
• Baby Boomer
• Gen X
• Gen Y — millennial

U.S. minority status (simple classification of ethnicity into two segments)

• Minority
• Non-Minority
• Base Pay
•  >$100K
• $76K to $100K
• $51K to $75K
• $25K to $50K
• <$25K

Market pay group – variation A

•  >60th Percentile
•  40th to 60th Percentile
• <40th Percentile

Market pay group – variation B

•  >75th Percentile
•  25th to 75th Percentile
• <25th Percentile

Market pay group – variation C

•  >90th Percentile
•  10th to 90th Percentile
• <10th Percentile

Cross-Tabbing for Insight

This section outlines how to get started working with data in segments. Cross Tabbing (also called cross-tabulation or cross-tab) is putting data together in a table in a particular way that allows you to see if there is a relationship among variables. A cross tab puts one method of segmentation on the columns and one method of segmentation on the rows. You use a cross tab if you want to compare the similarities or differences of segments as divided by some other segment. For example, what proportion of employees in each division are men or women? Or do sales employees exit the company disproportionately to the size of their division when compared to the exit rate of other divisions? These and many other questions can be answered without further complication - merely by constructing a cross-tab table and reviewing the numbers.

Setting up a dataset for cross-tabs

The key to working successfully with cross-tabs is to do the prep work correctly. The prep usually involves these two major steps:

1. Extract the data from wherever it is.

2. Organize the data in a way that is useful for reporting and analysis, using either a statistical application or a spreadsheet.

<Tip> The majority of analysts extract data from a company's HRIS and work with data in Excel. Eventually, you should replace Excel with a more robust and permanent reporting solution; however, it’s a great place to get started.

You may have descriptive information that you need to change into numbers — location, for example. For each employee, you find 1 of 20 different locations. If you want to analyze the U.S. versus non-U.S., you can create another field, named U.S.–Reference, and make the value a 1 for any employee at a location in the U.S. and a 0 for any other location. Then you can count the U.S. or use that variable in any other more advanced statistical procedure.

Table 4-1 shows you what I mean, by representing a few records from a simple dataset.

Table 4-1: A Simple Dataset

In the table, all U.S. locations are 1s, and those outside the U.S. are 0's. Also, I took the 0–10 Likelihood to Recommend the Company variable and added another variable (Likelihood to Recommend: High Reference (0,1), which indicates 1 when Likelihood to Recommend is greater than 7 and 0 if not.

Getting started with cross-tabs

When you organize data in a table, the next step is to count by “crossing” two or more variables against each other — known as creating a cross-tab. Cross-tabbing helps you see the interactions between two variables by explicitly revealing that some measured characteristics usually appear (or do not appear) in conjunction with other characteristics.

In the example in Table 4-2, you cross two columns from table 4-1 with each other into a two-dimensional table: Division by Likelihood to Recommend: High Reference.

Table 4-2: Working with Two Variables

Each cell in Table 4-2 contains the number of employees who fit the category listed as the row heading and the column heading. The 300 employees in Sales, for example, rated the company higher than 7 on Likelihood to Recommend, which was thus coded as a 1 in Likelihood to Recommend: High Reference.

The next step is to convert the numbers in Table 4-2 into percentages. You can calculate the percentage of rows or columns or the overall total, depending on what you want to know.

In Table 4-3, I calculated the percentage of each row, which tells you the percentage of each division segment that is likely to recommend the company. I used this formula:

Percentage of employees in cell = ((number of employees in cell) / (number of employees in row total)) * 100

Table 4-3: Percentage of Row Total

You can draw a lot of information from Table 4-3. For example, there appears to be some association between Division and Likelihood to Recommend the Company. Twice as many sales associates would recommend the company over those who would not, whereas the Engineering and Operations groups don’t fare as well among their employees.

Using segment definitions to defuse data quality concerns

If you put the wrong data into the system, then there’s nothing that the systems or the systems people can do about that. You have probably heard the term garbage-in, garbage-out get thrown around among systems people. It merely means that whatever you put in is what you get out. If you add bad data to a database, what you get out is the same. It is that simple.

At other times, the data in the system is technically accurate, but when it comes to reporting that data, the choices you make may make the data appear wrong to others. When people make different choices, they get different answers — it's as simple as that. When data don't match, sometimes people assume that there’s a data quality problem when the two reports may merely be using two different segmentation definitions.

Usually, disagreements about the data you’re using come to the fore when you report to a group of executives. They may say, “We don’t have 30 people in that group” or “Finance showed us a report with headcount yesterday, and on their report, it said 15 and yours says 10 — yours must be wrong.” Often, the answer lies simply in the definition of headcount: Finance may be counting contractors plus employees, whereas you may be counting only employees, for example.

There are many ways in which you have to make inclusion/exclusions decisions that result in different numbers. Your data set for headcount can be affected by any of these factors:

   *  Employees vs. contractors

   *  Full-time vs. part-time

   *  U.S. vs. other countries

   *  Exempt vs. non-exempt

   *  Regular vs. temporary

As if that weren't enough, here are even more ways to affect headcount:

   *  Date: Often, you get a different count, depending on the date you pull the data. People are always coming and going, which means that one day may be different from the next. Do you count the number of people at the beginning of the period or the end of the period, or do you use an average? Do you pull headcount on the day you present? If you run all reports at the same time, they have a better chance of matching — but they usually aren’t run at the same time.

   *  Fractions: You can count part-time employees as partials based on the hours worked (.3) or count each one as (.5) or count them as a whole, just like a full-time employee (1), or you might not count them at all. As they add up on your report, you end up with a different number.

   *  Hierarchy choices: If you’re counting employees by a unit, you might count them by the financial structure definition of that unit (Example Sales), or you might count them as being anyone under the head of that financial structure unit (everyone that reports to managers that report to directors that report to the Vice President of Sales John Smith). You may think that these two methods should give you the same answer, but it may not — and it usually doesn’t.

Work with others to create a standard definition or to make your definition clear. You’ll have valid reasons to count things in different ways.

Good Advice for Segmenting

Here are some additional tips you can use to create accurate and useful employee segmentation:

* Tackle one task at a time. Don’t try to try to do everything at once. You’ll become overwhelmed, as will everyone else.

* Focus your efforts. Start with some research objectives, hypotheses, and questions to answer before you spend much time on the data.

* Be open-minded. You may begin with a particular segmentation scheme, but it may evoke a new question, and you may need to add additional segments to your report to answer the new question.

* Remember that segmentation can help you see things you might not have otherwise seen or thought to ask. After collecting any data, most good researchers run reports by a bunch of fundamental segments, just to get a sense of what is going on in the data.

* Expand your outlook. By that, I mean that you shouldn’t confine your segmentation design to only the data from a single data source. Many valuable people-related insights come from data found in transient sources like employee surveys. Assuming that you have collected and stored data the right way, there’s no reason you shouldn’t be able to combine data from multiple sources. If you want to increase the possibility of producing insight - increase the number of sources, you can add to your data for segmentation and analysis. The implications of new segmentation for analytical productivity are vast because the new segmentation options are very large. The implications of new segmentation for analytical waste are vast for the same reasons! While I recognize the two last sentences are a confounding puzzle, a paradox, master this puzzle, and you open the key to insight. Control of the destiny of your productivity of analysis lies in control of your segmentation parameters.

* If you’re sharing data with others, pick out the essential segments to share based on what you have to communicate. Leave out data, that while entertaining, is nonessential. Put segment data that you don’t plan to present live in an appendix or another location that you can locate quickly in case you need it to address questions.

There are a variety of ways you can segment employee data for reporting and analysis. The options for segmentation are constrained only by the facts you collect (location, start date, and pay, for example) and your imagination. Imagination is vital to help you figure out how you can use data to answer questions and how to get new data when necessary. If you don’t have the data that you need to create a segment you want to create, you can use your imagination to find a way to get it. The options for segmentation are, therefore, infinite. What you do with segmentation should be determined by your purpose — the questions you want to answer — not what was is recorded coincidentally. 

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

Don't judge a book by its cover. More on People Analytics For Dummies here

I have moved the growing list of pre-publication writing samples here: Index of People Analytics for Dummies sample chapters on PeopleAnalyst.com

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.dhirubhai.net/in/michaelcwest
• Join the People Analytics Community here: https://www.dhirubhai.net/groups/6663060
• Check out my blog and an index of other people analytics related writing and resources here: Index of my writing on people analytics on PeopleAnalyst