Skip to main content

8 Changes Machines Learning Impact Us as Employers

Form the employer’s perspective, machine learning is just something to be “employed” to meet their objective, more efficient in reaching out to potential consumers, less costly in manufacturing and employee management, and better serve existing customers. This article looks at how machine learning can be helpful in the corresponding aspects:

Identify and Engage with Matching Talents

Matching algorithms is among the most development area in Machine Learning. Today, its implementation an be found everywhere: from the type of content shown on our Facebook news feeds to the suggested TV shows that come up on Netflix, and even to the matches suggested on dating sites/apps like Match.com and Tinder. 

At the moment, most of the matching algorithms use strings and keywords in resume to filter candidates. It makes finding potential candidates faster and more accurate. Soon, it should be able to match candidates based on industry background, engagement style (if a candidate work mostly with international team or local team; remote or face-to-face); culture exposure (multi-national team or local team); ecosystem awareness (serving both external and internal team; hierarchy company comparing to flat structure etc).

INC survey suggested that over 40% of the employees left their work within the first 6 month; and other 15% left within the 1st year. In specific industries such as retail and leisure, the number can be as high as 66%. Culture and opportunity play a big role in this move.


Employee Turnover By Industry (%)

                            
Machine learning help employers understand a better match between the person and the team; match the employers’ expectation with the candidate’s skill and experience.

Write the Job Description Better

One of the most important thing about the company is its culture. A traditional Hong Kong bank is different from a IT startup; moreover, even all IT startups are different. It is important to apply accurate description on the hob post so candidates can get an accurate understanding of the culture and figure out, before going through the effort on rounds of interviews and only to find themselves feeling miserable as they can’t associate with the culture, if they would enjoy the time with a potential employer.

Furthermore, we all know working in the company is not just about carrying out a certain function. It involves working with the team, collaborating with other departments, participating in the various activities the company organize and initiating projects. Few of these activities made its way to the job post and even if it does, the descriptions are pale. 

Textio (https://textio.com/) is a pioneer in the field of “Hiring Smart”. It help recruiters to analyze job listings and hiring outcomes with data collected globally so as to find a pattern why some job post gets great results while others fail.

Improve Employee Evaluation Process

Stephen Bruce surveyed nearly 1,400 managers and found majority of the performance evaluation don’t go beyond supervisor rating and employee self-rating; in terms of frequency, 74% of the respondents go through review once in a year or less. Most of the employees do not hear about how they have done or what they can improve frequent enough. According to OfficeVibe, 65% of employees today say they want more feedback at work





Another problem of the performance evaluation lays in the supervision of the evaluator. 41% of the respondents credit not following up with employees after the review to check on progress as the top error supervisors make while another 39% complains that the supervisor place all employees in the middle of the scale.

Nevertheless, supervisor are rarely held accountable when they do a poor job of managing the performance appraisal process while 72% of the respondents rarely or never receive disciplinary action for their poor review


Performance Evaluation Structure (%)




Machine learning can help improve employee evaluation process in the following ways:

Build up a database with questionnaires personalized for the company and the role: instead of manually creating every questionnaire, the company can have a database of questions based on the role, experience level, project details and team structure, HR and supervisors define the templates together and the questionnaire can be automatically compiled and send out when it is time to do the evaluation; next time the employee may get a promotion, or work on different team, or did some specific task, and the machine will then generate another questionnaire based on the new context.

Makes it easy and quick to implement peer review and evaluator training: as machine replace HR to create, deliver and follow up on the evaluation, the cost of doing evaluation reduces. A machine executed process also ensure anonymity among the employees, making it a good alternative. 

Evaluators can engage with chat bots both to discuss techniques to conduce better evaluation and share tips from their own experience (which the machine can use later on to demonstrate to other evaluator that had similar problems). And they can do it at anytime they prefer.

Utilize alert and pop-up questions to create the habit of in-context feedbacks: end of a project, closing of a sale, release of an upgrade, these are all good checkpoints for review and feedback, but with the performance evaluation isolated from the day-to-day operation of a business, integrated machine learning collect the triggers and carry out work flows to collect concise feedbacks in the right context so that it is based on facts not memory.

Identify bad behaviors and provide feedback in real time

a constructive performance evaluation is the respect every employee deserves. Machine learning helps identify bad behavior (e.g place all employees in the middle of the scale; only base on recent experience) and provide real-time feedback to the evaluator and if no improvement is observed, escalate to HR and its supervisor.

Be Close To Employees

Replacing employees is a costly activity:
  • 16% of annual salary for high-turnover, low-paying job;
  • 20% of the annual salary for a mid-range position
  • Up to 213% of the annual salary for a highly educated executive position

And these numbers don’t even take into consideration loss of productivity and training cost spent on the employee.

on the other hand, there are little an HR could have done to retain the employee when s/he already decided to leave, while they often have no idea what happened before the resignation comes.

Machine learning can be very helpful in implementing preventive actions before the situation becomes un-irreparable

Identify irregular behaviors: long hours of work; fail or just meet the deadline; not getting enough customer support tickets comparing to before etc, these are all indications that an employee is going through a hard time. While in a busy working environment, these indications can be neglected, a machine can easily identify the exception case and notify the manager in line or the HR department for such situation. A chatbot can even go through some preliminary problem solving with the employee to collect data and feedback to the managers.

Remove frustration, improve efficiency: and adopting tools to improve efficiency will always be favored by the employees; Wise.io, for example, utilize machine learning to helps identify and prioritize ideal customers  and match them with the right sales people. 

Collect and Analyze Survey Results in an Agile Way: Josh Bersin said to Forbes in 2015 that employee feedback solutions will redefine how we manage our organisations. Machine learning is being utilized in 3 areas: using text analytics to make sense of vast amounts of open text answers; using patternspotting techniques to make probabilistic assessments of which populations are most likely to raise certain topics; and to use survey data to answer business questions.

Develop New Business Model Not Available Before: mid of 2016, Foxxon replaced 60K employees with robotics; Fanuc’r robots can learn a task in 8 hours with 90% accuracy.  In certain industry, robots with learning abilities has gradually replaced human labors, creating a whole new business and operational model.

Of course, the future is not splendid for all employers, there will also be many employer that suffer from the wave of machine learning, pushed out of the market; forced to reduce the margin or pay off their staff. It is up to the employers to decide what path to take, how to innovate and transform. Nevertheless, as the economist said, it is easier to say that people need to keep learning throughout their career, the practicalities are daunting. And it is more complex for employers as they have more to lose. 

Why Employees Quit Jobs Right After They've Started

Fifth of Staff on Probation Fail Trial Period or Have it Extended

Textio

Performance Management Survey Results in How do You Compare

Employee Retention the Real Cost of Losing an Employee

Wise.io

Betterwork Powers Employee Feedback and Recognition with Machine Learning

50 Smartest Companies 2016

Japan Robotics Giant Gives its Arms Some Brains

Comments

Popular posts from this blog

Will AI Believe in God?

I came across an interesting article recently on Gizmono ( Link ) titled “when superintelligent AI arrives, will religions try to convert it”,  The key question seems to be simple at the first glance. The commonality among all different interviewees suggest that there is some kind of criteria the Strong AI has to pass to be qualify as human to be relevant to the religion.  The complication comes in that different party seems to have different criteria, to summarize: If Strong AI has a soul? If Strong AI has a soul an spirit? And for the very fact that it is about AI and religion, let’s refer to Bible for the definition of spirit.  What is the definition of Soul:   the   spiritual   part   of   humans   regarded   in   its   moral  aspect ,   or   as   believed   to   survive   death   and   be subject  to   happiness   or   misery   in   a   life   to   come.  “In Noah’s day . . . a few people, that is, eight souls, were carried safely through the water.” (1 Pe

8 Takeaways on Productive Usability Tests

Please, test it.  The fact you believe in your idea, or that your investor do as well, it means a lot about creating a wonderful team and culture. It does not mean your customer will buy into that.  There is a long way between what the market opportunity (as the market analysis and the founder’s vision is) and what the actual product is. Of the thousands of products in different countries that seems to solve the same problem (you may call them competitors if you want), which one will win the customers’ favor now and in the future? Usability test helps to provides insight on that. Or in other words. usability test tells you, at the current state of mind, your customer will most likely find your product to be value if you do … in the following way … and … in the following way ... As you can see, these findings can be translated to what you can include in your MVP, why a certain design or feature did not meet the expectation (or will ever meet) and what you may decide

2 Myth & 2 Truth About Bio-authentication

Bio-metrics using our physical characteristics to verify our identity. What has been advertised on the market is that it improves security level and make user experience much more simple.  While at the same time, we also hear the opposite voice, from time to time, about the failure of the technology and the risk of using it. In this article, I aim to summarize the research findings to understand the actual benefit and risk of bio-metrics and what what are the key hurdles for implementing biometrics in end-consumer products. TechTarget defined Biometric payment as a point of sale (POS) technology that uses biometric authentication to identify the users and authorize the deduction of funds from a bank account. Fingerprint payment, based on finger-scanning, is the most common biometric payment method.  As the authentication involved physical characteristics (e.g. fingerprint; voice; pupil etc), there certainly need a “point” on which the “Sale” is confirmed. Nevertheless, the POS concept