04
We know that pretty much everything we do nowadays leaves a digital trace, and this means it is now possible to get data on almost any aspect of people management. After all, data are by-products of computing and, these days, everything is essentially a computer, from the smartphones we carry everywhere with us and the vehicles we drive to modern manufacturing equipment. Even basic office equipment can be made ‘smart’. For example, the humble office chair has been given a ‘big data’ facelift: BMA Ergonomics’ Axia Smart Chair has sensors fitted in the seat, and the data gathered from these sensors are used to give feedback on the employee’s posture, helping them to sit properly and avoid the health problems associated with bad posture.1 It is clear then that HR teams have exciting data options that go way beyond data that must be collected for statutory reasons or data from the normal annual performance review cycle. But even traditional HR data can be optimized and used more intelligently. Instead of using just a simple scoring mechanism on performance review data, for instance, you could use text analytics to reveal more detailed insights about your employees’ performance and level of satisfaction (there is more on text analytics in the next chapter).
To help you determine which data work best for you, this chapter sets out the different categories of data and explores specific types of HR-relevant data, namely activity data, conversation data, photo and video data, and sensor data. I will also look at options for sourcing and collecting the data you need, and address the question of whether any one type of datum is better or more effective than another.
Distinguishing between different types of data
Before we get into the various different sources of HR-relevant data, it is important to understand the key overarching ways in which data are categorized. As illustrated in Figure 4.1, data are either internal or external, and either structured, unstructured or semi-structured. Let us look at each category in turn.
FIGURE 4.1 Types of data
Internal data
Internal data comprise any proprietary data that are owned by your business, including all the data you currently have and all the data you have the potential to collect in the future. This includes employees’ personal data, performance review data, employee surveys, sales and financial data, customer feedback etc. Internal data can be structured, unstructured or semi-structured (more on this later). The obvious big advantage of internal data is that they are usually cheaper and easier to work with than buying in or paying to access external data. They are also uniquely tailored to your business and industry (as opposed to external data, which may not be), which makes internal data incredibly useful and valuable. This points to a disadvantage, however: because your internal data are so uniquely tied to your business, they may not provide a rich enough picture to meet all your strategic goals and may need to be supplemented with some external data (particularly when it comes to recruitment) to get all the information you need. Another important point to note about internal data is that you are responsible for looking after and properly securing those data, particularly sensitive personal data (more on this in Chapter 6).
There is a lot of hype and excitement concerning data but, often, internal data are not viewed as terribly exciting or cutting edge – after all, they might well be the same old data you have been collecting for years; however, the value of internal data should never be overlooked. US-based recruitment and training firm Kenexa was acquired by IBM in 2012 for US $1.3 billion,2 partly on the basis of its valuable internal data, gained from assessing millions of employees, managers and applicants each year. IBM was able to analyse these precious data to gather insights on the key characteristics of successful sales people. After comparing employee tests and surveys with assessments by managers, IBM found that, rather than the obvious ‘outgoing personality’ characteristic typically associated with sales positions, ‘emotional courage’ was actually the most important trait for a successful sales person to have. This insight allows IBM to test for emotional courage in future sales applicants and hire only those that score highly for this trait.
External data
External data comprise all the data that exist outside of your company, be it public data that are freely available or data that are privately held by another organization. These data include social-media profiles and posts, recruitment data from sites such as LinkedIn and Glassdoor, economic data, data on social trends and much more. Like internal data, external data also can be structured, unstructured or semi-structured. While some external data can be accessed for free, you may have to pay to use certain data, particularly data owned by private, for-profit companies. Therein lies an obvious downside: these data may not be cheap. Access rights also can be an issue in that there is always a risk that an external provider could cut off access or jack up prices. Therefore, if you are reliant on certain external data for absolutely critical insights, it is worth exploring whether you can possibly generate or gather those data yourselves in-house.
Despite these considerations, external data offer some huge advantages. They are often richer and more detailed than the average company’s internal data, giving HR teams access to vast and complex data sets that they could never hope to build internally. Complications concerning storing and managing data are reduced by working with external data providers, as they will look after and secure their own data. Silicon Valley-based networking solutions company Juniper Networks provides a great example of a company using external data intelligently. Juniper uses the vast amount of data available through LinkedIn to analyse where its most successful and best-performing employees come from – and where people move on to when they leave Juniper.3 This gives the HR team a useful picture of career paths in the networking solutions industry, information that they can use to help attract and keep the industry’s best talent.
Structured data
As we saw in Chapter 2, the term structured data essentially means any data that can be organized neatly in rows and columns, usually in a database or spreadsheet. This may include employee personal data, sales data, test scores, performance review scores, absence data, salary information, data points from sensors etc. Clearly, the average HR team has the potential to tap into vast amounts of structured data, particularly internal structured data. In fact, for most businesses, until very recently, the majority of data analysis was based on structured data because, by their very nature, they are much easier (not to mention cheaper) to organize, structure, store and interrogate. And they often can be interrogated by non-analysts (most employees can find their way around a database, for instance), which is another big advantage. Despite their fixed nature, structured data still can be incredibly powerful. One such example of the power of structured data was seen in Chapter 1, where a bank analysed the colleges their employees came from and determined their best-performing people came not from the top-tier schools, as expected, but from less-prestigious universities.
Another example can be found in the book Work Rules! written by Laszlo Bock, former Senior Vice President of People Operations at Google.4 Bock describes how structured data help Google to optimize the hiring process. Interview questions are all computer-generated and completely automated to make sure the company hires the best talent, without human bias getting in the way. But one big downside to structured data is that they account for only approximately 20 per cent of all the data in the world – the rest are unstructured or semi-structured data.5 This means, if you focus on structured data alone, you could be seriously limiting the number of insights available to you. And those insights are likely to present a less detailed and rich picture than if you combined structured data with unstructured data. Take employee turnover rate, for instance. Using structured data, you can determine that your employee churn rate is 20 per cent. But that is all the structured data can tell you. You will need to make use of unstructured data, such as the detailed answers given in an employee exit interview, to understand why employee churn is 20 per cent.
Unstructured and semi-structured data
Unstructured data are essentially all the data that do not fit neatly into a spreadsheet or database, whether they are internal or external data. Unstructured data are often text-heavy, but the term could also refer to audio or visual data. Examples include social-media posts, employee e-mails, employee and customer feedback, photos, videos (eg CCTV footage) and audio recordings (eg customer-service calls). These messy data types used to be too difficult or expensive for the average company to work with, but that has all changed in recent years. Now, thanks to massive advances in storage and computing power, increasingly more companies are benefitting from unstructured data.
As you can probably guess, semi-structured data sit somewhere between unstructured and structured data. They have some sort of structure (like descriptor tags, for instance), but lack the strict structure found in databases or spreadsheets. For example, a tweet can be categorized by author, date, time, length and even the sentiment behind the tweet, but the text itself in the tweet is generally unstructured and therefore would be a little more complex to analyse. This is the main disadvantage of unstructured or semi-structured data: they are more complex to work with. They tend to be bigger, which means they require more storage, and they are more difficult to organize and analyse, requiring specialist analytics tools. Obviously, all this has an impact on costs; however, that should not put anyone off using unstructured or semi-structured data as there are some serious benefits. Clearly one big benefit is that you are broadening your view much more than you would if you stuck only with structured data (which, as we know, excludes around 80 per cent of the data available). You should also be able to gain a much more detailed, rich picture by combining these messy data with structured data.
Facebook has long been able to make some scarily accurate predictions about us based on the unstructured and semi-structured data found in our profiles, our likes and the content of our posts. Recently, it has been working to use this predictive power to help save lives. In March 2017, the company announced AI-driven tools and algorithms that can spot users who may be at risk of self-harm or suicide.6 These algorithms mine data from users’ posts and comments added by concerned friends looking for words and phrases that are linked to suicide or self-harm. The idea is to flag posts that raise concern more quickly and easily, and connect those at risk with mental health services. Similarly, a few years ago, Microsoft announced that it had developed a method for identifying Twitter users at risk of developing depression.7 It is easy to see how developments like these can potentially benefit HR teams charged with looking after the wellbeing of their employees.
Identifying HR-relevant data
Now let us look at the main types of data that are relevant to HR. Essentially, HR-relevant data can be categorized as follows:
· activity data;
· conversation data;
· photo and video data;
· sensor data.
It is important to understand that all of the above are still either structured or unstructured/semi-structured data, and they can include both internal and external data. For example, sensor data are structured data and they can be internal or external, depending on which sensor data you are using; conversation data are likely to be unstructured or semi-structured and can be internal or external etc. Over the following sections I will look at each data type in turn.
Activity data
Activity data, which provide a record of human activities or actions (whether online or offline), can be incredibly valuable from an HR perspective. Think about all the things you do in the course of a normal day – they all generate activity data. If you wear a fitness band with a sleep tracker, like I do, even your sleep and the time you wake up generate activity data. Then you travel to work, perhaps paying for a ticket with a bank card or touching in with a travelcard. Assuming your phone is a smartphone, it will generate records of your location while you are on the move. If you make or receive a phone call on your way to work, or post a photo on Twitter, that generates data. Then you get to work and send countless e-mails, type hundreds or maybe thousands of words a day and look at numerous webpages. Maybe you buy something online or head to the supermarket in your lunchbreak. Even browsing online for ideas for your partner’s birthday next month generates data.
Understanding what your people really do
The sheer volume of activity data available to HR teams can be overwhelming, so it is important to always refer back to your strategic objectives and focus only on the data that help you to achieve your goals. But the real advantage of activity data is that they allow you to assess what your employees actually do, as opposed to what they are supposed to do or what you assume they do. Some companies take this to the extreme. Bloomberg, for example, reportedly gathers data on every single keystroke made by its employees.8 Others focus more on specific activity data. The Container Store, for example, uses wearable technology to track employee activity, how they communicate with customers and other staff, and where they spend most of their time.9 Performance-related data are particularly valuable for HR teams. By tracking such activity data, companies can accurately monitor individual performance and use this information to identify top performers and those who may need help.
Activity data in action: recruitment and retention
Of course, when you know who performs well and what characteristics top performers share, you can focus on hiring more people who match those characteristics. San Francisco software provider Evolv has created pioneering online tests to help refine the recruitment process. Evolv’s tools help businesses gather data on everyone who applies for a job at the company and everyone who gets hired. With a data set of over 300,000 candidates who have now taken their online assessments, Evolv has some incredibly valuable performance-related data. The company can pinpoint, with an extremely high level of accuracy, which characteristics make for a more successful retail sales person, for example.10
We briefly looked at Xerox in Chapter 1 and saw how the company had gathered valuable insights on what makes a successful call-centre worker. Xerox is just one of the big companies that has used Evolv’s tests to refine its hiring practices. And with roughly 45,000 employees working in Xerox’s 150 US customer care centres, finding those workers who will perform well and stay in the job is critical. Xerox switched to online assessment of candidates back in 2010.10 An algorithm analyses the applicant’s test scores, alongside factual information provided by them on their application, and allocates a traffic light rating to each application: green means they are good to hire, red means they should be avoided and orange means they are in the middle of the pile. It was these tests that showed Xerox that previous call-centre experience had no impact on retention or productivity – a major assumption on which hiring managers had been basing their decisions. The tests also showed that how close candidates lived to the office was a strong marker of retention and employee engagement. In the initial pilot period alone, employee attrition fell by 20 per cent.10 And the longer-term benefits of the improved quality of hires included an increase in the number of promotions.
Activity data in action: fostering innovation
Another example comes from video game start-up Knack’s collaboration with Royal Dutch Shell. Knack’s video games, which were designed by a team of data scientists, psychologists and neuroscientists, are not just about having fun – they are about measuring human potential. All sorts of factors are logged as a player participates in the games: every move they make, how they solve problems, how long they pause before taking action etc. This builds a thorough picture of the player’s level of persistence, creativity and even intelligence, as well as their ability to prioritize tasks and how quickly they learn from mistakes.
Royal Dutch Shell’s GameChanger unit, which is charged with identifying the best business ideas from inside and outside the company, was extremely interested in the potential of these games to improve and speed up the process of identifying the best ideas. So the unit devised an experiment: 1,400 Shell employees who had previously proposed ideas to the GameChanger team were asked to play a couple of Knack’s games. The GameChanger team then shared with Knack information on how well three-quarters of the players had fared as idea generators (whether their ideas made it all the way, for instance). Knack used this information to develop game-play profiles of the best idea generators in comparison to the weaker ones. Using information based on these top innovators’ game profiles, Knack was then asked to guess from the remaining quarter of the players who had had the best ideas. This was done with startling accuracy, clearly identifying those who had previously generated winning ideas based only on the way they played the games.10 Based on this experiment, Knack and Shell were able to identify the key characteristics of top idea generators, such as social intelligence and task-switching ability. This has allowed the GameChanger unit to devote more time to those employees whose ideas are likely to have more merit.
Conversation data
Conversation does not just mean two employees having a conversation around the coffee machine. It covers any conversation people may have in any format, whether it is a call with a customer, an instant message sent via phone or computer, company e-mails or written survey responses, social-media posts etc. These are all examples of conversation data. These types of data are incredibly valuable to HR teams because they can give in-depth insights into how happy and engaged your employees really are, as well as how positive your employer brand is (eg by analysing data on Glassdoor). Thanks to advances in analytics, conversations now can be mined for the content itself (what is said) as well as context (how it is said). In other words, you can understand what is going on from the words used and the mood of the person engaged in the conversation. This means companies can tell how happy, irritated or stressed an employee is, or even if they are telling the truth, just by analysing the tone of their voice. For example, the US Department of Homeland Security is using voice analytics to detect when those entering the country are lying. Using its Avatar system, a computerized ‘agent’ with a virtual, human face and voice asks several questions.11 The person’s responses are monitored to detect fluctuations in tone of voice, as well as the content of what exactly was said. These data are compared against a database, and matched against factors that indicate someone may be lying, based on previous experience. While this may seem a little hard core for most HR teams, it shows how far the technology has progressed.
Conversation data in action
Let us look at a more viable example for the average HR team. Very few people enjoy filling in employee surveys. In fact, staff surveys are notorious for either not being completed at all or not being completed truthfully (because employees feel they should say what the company wants to hear or worry that their responses could be individually tied to them). Conversation data allow you to assess how people are actually feeling, as opposed to them saying what they think you want to hear. These data are particularly useful in understanding what makes an employee want to leave (or, for that matter, stay with) the company. By analysing text from open-ended questions in surveys and exit interviews, as well as social-media posts, e-mails and team assessments, HR teams now can accurately predict what makes an employee more likely to leave or stay with the company.
Hiring is another area that can benefit from conversation data, and it is not uncommon for employers to scour social-media profiles for glimpses into what potential hires are really like. This potentially could be done on a larger scale to identify the types of content and the sentiment behind things that successful employees post on Twitter and Facebook, and then use that knowledge to assess potential candidates in the future.
You could also assess conversations employees are having with each other and customers to assess their satisfaction (both the employee and the customer) and pinpoint behaviours of successful team members. Wearable technology potentially has a big role to play in this area in the future. Wearable technology company Sociometric Solutions (now part of Humanyze) has created electronic employee badges that capture information from conversations as employees go about their day, including the length of the conversation, the tone of voice involved, how often people interrupt, how well they show empathy etc.12 Sandy Pentland, the brains behind the badges and head of the Human Dynamics Lab at MIT, says data from the badges can be used to predict which teams are likely to be more successful, which employees are more productive and creative, and which show signs of being great leaders.13
Mining e-mails for insights
For many companies, e-mail is an especially rich source of conversation data, giving insights into employees’ productivity, treatment of colleagues etc. Text analytics software is getting better and cheaper all the time, making it possible for companies to search through employees’ e-mail traffic, hunting for words, phrases or patterns of communication that are linked to certain success (or failure or attrition) metrics.
It goes without saying that there are implications to gathering and analysing conversation data. When it comes to phone calls, for instance, generally speaking, you cannot record customers or employees just because you feel like it; what you are recording must be relevant to the business. You may also need to inform the parties that they are being recorded. Use of e-mail data also can be restricted depending on where you are in the world. In the United States, any e-mail sent in the course of work can be used for analysis. In Europe, employers should be more cautious about reading communications, especially if the conversation is of a private nature. There is more on such pitfalls in Chapter 6.
Photo and video data
Photo and video data refer to any kind of photo or video image (such as CCTV footage). The amount of photo and video data has exploded in recent years, largely thanks to the advent of smartphones and the increasing use of CCTV (especially in the United Kingdom). Photo and video data can be big, which can make them trickier (and potentially more expensive) to store and manage; however, your company may be collecting these data as a matter of routine already (perhaps through security footage), so it may not be very difficult or expensive to find new ways to use these data more intelligently. If your company is not collecting photo or video data already and you are interested in doing so, make sure you have a clear business case for working with these data (purely because it can be expensive) and that the benefits clearly link to your objectives.
Photo and video data in action
So, how could you use photo and video data? One example comes from the brightly lit world of Las Vegas. The analytics team at Harrah’s worked out that card dealers and waiters smiling had an impact on customer satisfaction. The hotel and casino now reportedly tracks the smiles of card dealers and waiting staff, presumably to monitor who could be performing better in the smile department.10 Facial recognition software means that individuals now can be easily identified in photos and videos. Facebook has been at the forefront of this movement and its own facial recognition software can now accurately identify an individual 98 per cent of the time, which makes it more accurate than the FBI’s facial recognition technology.14
Connecting photo and video data to AI
Microsoft also gave some interesting examples of how photo and video data could be used in the workplace at its 2017 Build developer conference.15 Its vision of the workplace of the future includes, among other things, cameras connected to artificial intelligence (AI) programmes. One example given for the use of video data involved a camera detecting that an employee was not wearing the appropriate safety gear, prompting a notification to be sent to the employee’s supervisor. In another example, an employee was captured on video taking a selfie while in charge of a dangerous piece of machinery. The cloud-based computer support system recognized the activity and the potentially hazardous setting, concluded that the individual was acting recklessly and notified a supervisor.
Sensor data
Sensors are being built into an increasing number of products, ranging from factory machinery to office chairs and yoga mats. And these sensors generate a wealth of data that can help HR departments improve their functions, including employee performance, employee safety etc.
Combining sensor data with other data sources
Because sensor data tend to lack context (they are just telling you what the sensor recorded at any given time, not what might have caused the event), keep in mind that they may need to be combined with another data set to get the best results, depending on what you are trying to achieve. But sensor data are self-generating, meaning they are very easy to capture, once the data capture tools have been put in place. Some devices, such as smartphones, contain ready-to-use sensors that can be used to the company’s advantage (eg a delivery company using their drivers’ phones to track driver behaviour).
Sensor data in action
Clearly, wearable technology has a huge role to play in utilizing sensor data, and the workplace wearables market is booming. One survey by Forrester Research of over 2,000 technology decision makers found that one-third of respondents said workplace wearables were a ‘critical’ or ‘high’ priority.16 Honeywell’s ‘Connected Worker’ solution is one example of this.17 Using a series of connected wearables sensors, the solution measures an employee’s heart rate, breathing, motion and posture to assess whether they are under physical stress or in potential danger (it can detect toxic gas, for instance). This kind of technology will become increasingly more common in the future, especially for workers in physically demanding jobs, or those who work in dangerous or isolated locations.
Some of the most innovative uses of sensor data come from the world of sport, and it gives us a glimpse of how companies might be using these sorts of data in the future. In American football, for example, injury levels have been reduced due to wearable sensors that monitor the intensity of activity and impact of collisions and compare this information with historical data to determine when a player might be in danger of overexerting or injuring themselves. One Olympic sports team I worked with used wearable devices to track how well athletes slept at night, and then correlated those data with track performance. This enabled coaches to assemble their teams based not only on past performance, but also on how well individual team members had slept the night before.
Sourcing and collecting the data you need
Having identified what you want to achieve with data (Chapter 3), it is a good idea to start by seeing whether the data you need already exist internally. If they do not, you will need to look at whether you can generate those data in-house.
Sourcing data internally
As we have already seen in this chapter, nowadays, you can gather activity data from almost any activity undertaken by the company’s employees. From test scores, to interview answers, to performance reviews, there is the potential to gather valuable performance-related data from any sort of activity. And wherever the HR team and your company’s employees are currently having conversations, there is an opportunity to collect conversation data. For example, if you operate a telephone sales department or customer-service department, you can record those conversations and analyse the content and sentiment for useful insights into how staff are performing. Surveys, e-mails, customer feedback comments, social-media platforms etc all provide useful sources of conversation data.
Video and photo data can be obtained by simply starting to collect them using digital cameras. For example, retailers can use their network of CCTV cameras to analyse how the presence of staff members in certain sections of the shop floor impacts on how likely a customer is to buy something. And with modern sensors being smaller and cheaper than ever (small enough to fit into an employee’s badge, for instance), they can be incorporated into almost anything, from manufacturing equipment to office equipment.
It is clear that internal data comprise a vital part of any data-driven HR strategy. But you may also need to combine those data with some external data to get a fuller picture that truly answers your strategic questions.
Sourcing external data
There is a wealth of external data already out there. As increasingly more companies view data as a business commodity, a market is emerging where practically any organization can buy, sell and trade data. (Indeed, many companies exist purely to supply other companies with data.) LinkedIn and Glassdoor are perhaps two of the biggest sources of HR-related data. In addition, there are lots of smaller, more industry-focused, data providers. So, even if you are looking for quite specialized data, there is a good chance someone out there has them.
Social-media platforms are obviously key sources of useful data, and they provide a wealth of information on current, past and potential employees. After LinkedIn, Facebook and Twitter are likely to be your first stop for social-network data. You can, for example, use sentiment analysis (more on this in Chapter 5) to find out what past employees are saying about your company culture online or how happy current employees are with their working environment. Or, say if there was a change in company policy or working conditions, you could assess employees’ tweets to find out how people are reacting to the changes. Sentiment analysis can tell us a lot about users’ feelings, opinions and experiences, without having to trawl through individual tweets one at a time – just like the example given earlier in the chapter where it was possible to identify people at risk of developing depression based on the content of their tweets.
Other sources of HR-relevant external data may include census data, which provide a very useful source of population data, geographic data and education data. This could be useful, for example, if you were looking to set up a new office in a new location and wanted to assess the potential workforce in that area using local demographics. In addition, weather data are often used by companies to help plan staffing levels according to the number of visitors expected on a sunny weekend, for example.
The importance of automating data collection
I keep coming back to the automation point, but with good reason. Wherever possible, you should put in place systems to collect or generate the data you need automatically. Whether you want to collect activity data related to employee productivity, or sensor data in a hazardous working environment, or whatever, the data collection should ideally take care of itself. The whole point of data-driven, intelligent HR is to free up HR time and resources to focus on adding greater value to the company. If HR professionals are engaged in lengthy data collection exercises, this completely defeats that purpose. Of course, with any new data project, time is needed to set up, fine-tune, maintain and assess those processes, but once that time has passed, you should be looking to collect data with the minimum effort possible, which leaves the HR team to focus on turning those data into insights, and then acting on those insights.
Identifying the most effective data type
Ultimately, no one data type is better than another. It all comes down to knowing what it is you want to achieve and finding the data that best help you to do that. A lot of the more exciting big data case studies focus on innovative uses of unstructured data, and it is easy to see why. But if it is possible to achieve your goals by working with structured data only, then why would you not do that? Unstructured data are not inherently more valuable than structured data. External data are not ‘better’ than internal data, just because there are more of them. And remember, what is best for one business may not be best for yours. With so many data available these days, the trick is to focus on finding the exact, specific pieces of data that will best benefit your organization, which is why it is vital you have a robust data strategy as your starting point (see Chapter 3). By working out what it is you need to know in order to achieve your objectives, you can go from there and identify the best data to give you those answers. Some data obviously will be more viable than others, and you can assess data options based on the ease and cost of sourcing those data versus how close they get you to your goals. But, ultimately, it does not matter whether the data are structured, unstructured, internal or external, or a combination of all these, the main thing is that they do the job you need them to do.
It is most likely that you will actually need a combination of data sets across different types of data, which is a good thing. If you rely on just one data set (employee responses to a survey, for example) to make critical decisions, you may be getting a very limited picture indeed. By combining those survey data with other data (such as activity data or conversation data), you can create a much richer picture of what is really going on. You can also verify insights rather than continuing down a road based on false assumptions from a limited data set. Say, for example, you are working to improve the company culture. To do this, you might well need a combination of internal structured data (like yes/no answers or 1–10 scores on an employee survey) with some external structured data (like your Glassdoor score), as well as internal unstructured data (like conversation data from employee interactions or open-ended survey questions) and external unstructured data (like social-media posts). The most intelligent HR teams will combine data to get at the most useful insights in relation to their goals. In my experience, it is often the combination of internal and external data and structured and unstructured data that provides the most valuable insights.
Key takeaways
The wealth of data options available can be overwhelming, but generally speaking, data boil down to a handful of key categories, as follows:
· Data are either internal or external, and either structured, unstructured or semi-structured. HR-relevant data generally can be categorized as activity data, conversation data, photo and video data, and sensor data.
· Activity data, which provide a record of human activities or actions (whether online or offline), can be incredibly valuable for understanding what your people really do.
· Conversation data cover any conversation people may have in any format. They are particularly useful for understanding what people really think and feel.
· Photo and video data cover photos, CCTV footage and any other kind of photo or video image. They can be particularly helpful in improving employee safety and security.
· Sensors generate a wealth of data that can help HR departments to improve their functions, including employee performance, employee safety etc.
· Options available for collecting data are:
1. – making use of existing internal data;
2. – generating new internal data (eg through sensors or surveys);
3. – sourcing external data (eg through sites like Glassdoor).
· Ultimately, no one data type is better than another. What matters is knowing what you want to achieve and finding the data that best help you to do that.
Once you have identified the data you need, the next step in intelligent HR is turning those data into insights through data analysis. In the next chapter I look at the various types of data analytics and how they can provide critical insights for HR teams.
Endnotes
1 BMA Ergonomics [accessed 23 October 2017] Axia Smart Chair [Online] https://www.bma-ergonomics.com/en/product/axia-smart-chair/#ad-image-0
2 Bersin, J (2012) [accessed 23 October 2017] Why IBM Acquired Kenexa [Online] https://www.forbes.com/sites/joshbersin/2012/08/27/why-ibm-acquired-kenexa/#d6aa4f71372f
3 Roberts, B (2013) [accessed 23 October 2017] The Benefits of Big Data [Online] https://www.shrm.org/hr-today/news/hr-magazine/pages/1013-big-data.aspx
4 Bock, L (2015) [accessed 23 October 2017] Work Rules! [Online] https://www.workrules.net
5 Schneider, C (2016) [accessed 23 October 2017] The Biggest Data Challenges That You Might Not Even Know You Have [Online] https://www.ibm.com/blogs/watson/2016/05/biggest-data-challenges-might-not-even-know
6 Callison-Burch, V, Guadagno, J and Davis, A (2017) [accessed 23 October 2017] Building a Safer Community with New Suicide Prevention Tools [Online] https://newsroom.fb.com/news/2017/03/building-a-safer-community-with-new-suicide-prevention-tools
7 De Choudhury, M et al (2013) [accessed 23 October 2017] Predicting Depression via Social Media [Online] https://www.microsoft.com/en-us/research/publication/predicting-depression-via-social-media
8 Seward, Z M (2013) [accessed 23 October 2017] Bloomberg’s Culture Is All About Omniscience, down to the Last Keystroke [Online] https://qz.com/83862/bloomberg-culture-is-all-about-omniscience-down-to-the-last-keystroke
9 Sacco, A (2014) [accessed 23 October 2017] How the Container Store Uses Wearable Tech to Think outside the Box [Online] http://www.cio.com/article/2378126/infrastructure/how-the-container-store-uses-wearable-tech-to-think-outside-the-box.html
10 Peck, D (2013) [accessed 23 October 2017] They’re Watching You at Work [Online] https://www.theatlantic.com/magazine/archive/2013/12/theyre-watching-you-at-work/354681
11 US Department of Homeland Security (2014) [accessed 23 October 2017] Rapid Screening Tool: the Avatar [Online] https://www.dhs.gov/sites/default/files/publications/Rapid%20Screening%20Tool-NCBSI-AVATAR-Jan2014.pdf
12 Humanyze [accessed 23 October 2017] People Analytics. Better Performance [Online] https://www.humanyze.com
13 Berman, A E (2016) [accessed 23 October 2017] MIT’s Sandy Pentland: Big Data Can Be a Profoundly Humanizing Force in Industry [Online] https://singularityhub.com/2016/05/16/mits-sandy-pentland-big-data-can-be-a-profoundly-humanizing-force-in-industry
14 Lachance, N (2016) [accessed 23 October 2017] Facebook’s Facial Recognition Software Is Different from the FBI’s. Here’s Why [Online] http://www.npr.org/sections/alltechconsidered/2016/05/18/477819617/facebooks-facial-recognition-software-is-different-from-the-fbis-heres-why
15 Sullivan, M (2017) [accessed 23 October 2017] At Build, Microsoft’s Vision of the Future Workplace Looks Both Helpful and Intrusive [Online] https://www.fastcompany.com/40419938/at-build-microsofts-vision-of-the-future-workplace-looks-both-helpful-and-intrusive
16 Gownder, J P (2014) [accessed 23 October 2017] How’s Your Enterprise Wearables Strategy? [Online] http://www.informationweek.com/mobile/mobile-business/hows-your-enterprise-wearables-strategy/a/d-id/1316342
17 Galman, D (2017) [accessed 23 October 2017] Honeywell Launches New Connected Worker Software Aimed at Boosting Safety, Productivity, press release [Online] https://www.honeywell.com/newsroom/pressreleases/2017/05/honeywell-launches-new-connected-worker-software-aimed-at-boosting-safety-productivity