Adapting the title from the famous book, I pose this question to HR leaders: “What differentiates Good from Great HR departments?”
In my view, understanding the key factors that contribute to greatness is something that every HR leader that wants to excel should be doing continually. However, it turns out that because HR is a copycat industry, it’s hard to identify any obvious differentiators between the HR functions at different companies. Almost all HR departments of a similar size share the same programs and features.
Through my research, I have found that a commitment to “hypothesis testing” is the single “good to great differentiator” that stands out, such as at the elite HR departments of Amazon and Google. I also found that the impact of experimentation and hypothesis testing is so powerful in disrupting the status quo and driving quantum improvements in HR and business results that it should be considered an indispensable step on the road to HR greatness.
My research journey started with identifying the elite top five most valuable US firms by market cap (Apple, Microsoft, Alphabet, Amazon, and Facebook). And, because I had advised managers at each of these firms, I was able to identify any prominent differentiating features. Only one differentiating HR feature stood out because it was only present at two firms, Amazon and Alphabet: A formal emphasis on experimentation and hypothesis testing. The goals of that feature include accelerated HR innovation, more rapid continuous improvement, quality control and faster learning within HR.
Incidentally, I also discovered that it’s possible to be a great performing firm without having a great HR department. For example, in my opinion, Apple outperforms as a firm despite the fact its HR function doesn’t have a single exceptional feature. Instead, Apple succeeds because it has a great culture of innovation and a great product brand.
The two greatest HR functions emphasize experimentation and hypothesis testing
Amazon and Alphabet (i.e. Google) clearly stand out because of their data-driven experimentation and hypothesis testing. Here is how they internally codify and communicate this unique HR cultural element.
Amazon’s HR tenets include hypothesis testing – “We seek to be the most scientific HR organization in the world. We form hypotheses about the best talent acquisition, talent retention, and talent development techniques and then set out to prove or disprove them with experiments and careful data collection.” “We rigorously audit ourselves to disrupt and reinvent HR industry standards.”
Alphabet’s HR tenets include experimentation – “We apply the same level of rigor, analysis, and experimentation on people as we do the tech side.” “Our mission is to have all people decisions to be informed by data.” Google has formalized its commitment to experimentation by creating their PiLab. It runs experiments on employees in an effort to answer questions about the best way to manage at Google.
Hypothesis testing challenges best practices in HR
Almost every HR department relies on established “best practices” that have been adopted throughout the industry to guide their actions. Unfortunately, utilizing the same best practices as everyone else guarantees you won’t have a competitive advantage in HR. Also, this approach is based on the premise that the utilized best practices will continue to be effective over time. We live in a VUCA world where there are constant dramatic changes and new technologies continually being developed. What is needed in order to achieve and maintain greatness in HR is an ongoing assumption that all existing HR practices will soon become obsolete. And, under this approach, you form hypotheses about what still works (e.g. diverse interviewers hire more diverse candidates). Then you run statistical tests or conduct an experiment to find out if your original premise or hypothesis is still true (or it isn’t).
Examples of hypotheses that when tested, often produce startling results
Because most HR functions are not data-driven, the actual testing of hypotheses often produces surprising results. Here are some examples of hypotheses that major firms have tested, along with their yes/no results. Perhaps seeing these normally unanticipated results will stimulate you to begin challenging your common HR assumptions using a hypothesis testing program.
- Effective managers – Managers have the highest positive impact on team productivity of any factor (yes).
- Customer complaints – Inexperienced staff are the source of most customer complaints (no).
- Performance management – The performance and the behavior of employees that complete a performance management program improves significantly (no). And, standard performance appraisal processes have a measurable positive impact on performance (no).
- The causes of turnover – Delaying exit interviews for at least a month produces 40% more accurate causes of the turnover (yes).
- Increasing retention – You can predict with a 95% accuracy which employees are likely to quit (yes). And, a promotion has the highest positive impact on a target employee’s retention (yes).
- What factors predict new hire success – Past experience in the job is the best predictor of new hire performance (no). And, boomerang rehires routinely become top performers (yes). Also, the size of a new hire’s external network helps to predict their on-the-job performance (yes).
- Lengthy new-hire deliberations – Taking longer to deliberate and make hiring decisions increases the probability you will hire top performers (no).
- Onboarding results – Extended onboarding produces longer new-hire retention and higher initial productivity than standard onboarding when an A/B test is used to compare (yes).
Action steps for making hypothesis testing part of your HR culture
If you want to make the transition to greatness in HR, your function will need to implement a formal data-driven process for continuous experimentation and hypothesis testing. Rather than operating as a separate program, hypothesis testing must become an integrated part of your HR culture. If you decide to adopt this strategy, here are some action steps that can get you started.
Fully understand the approaches taken by the best in HR – In order to avoid learning by trial and error, you should closely study the experimentation approach taken by Alphabet and its PiLab, as well as the HR hypothesis testing practices that guide Amazon.
Adopt a continuous obsolescence assumption – Start with a “glass is half-empty and leaking” assumption about all of your major processes and programs. Assume everything you do will soon become obsolete. Also, adopt the Amazon approach of rigorous audits in order to disrupt and reinvent HR. And, finally, only hire HR professionals that have the capabilities and the willingness to test hypotheses that challenge the status quo.
Assemble a hypothesis testing team – Don’t make the mistake of assuming that those with traditional HR backgrounds will be able to do effective experimentation. Instead, follow the lead of Amazon and Google and get the help of engineers, data scientists, and psychologists to ensure accurate hypothesis testing. Closely study the work and the methods of Google’s Pilab (short for People & Innovation Lab) and their analytics team leader, Prasad Setty.
Partner with other business functions that regularly test – Hypothesis testing is only rare within HR. However, it’s quite common in marketing, product development and CRM. So, work within and learn from the business side. Look for those functions that practice key elements of hypothesis testing, including beta testing, A/B testing, split-sample testing, pilot implementations, failure analysis, root cause analysis, and correlations to identify critical success factors.
Start by identifying the underlining assumptions of your major HR processes – Begin by identifying the primary assumptions that underlie each major HR process. Prioritize these assumptions based on the premise that if any single one was wrong, which faulty assumptions could have the greatest negative impact on business and HR results? Begin your work by testing which of these impact hypotheses are actually true. Then try data-driven experiments in order to determine which modifications would produce the largest improvement in results.
Next, prioritize the HR functions with the highest business impacts – Research has shown that recruiting, retention, onboarding and internal movement usually have the highest business impacts of all HR processes. So, prioritize your hypothesis development and experimentation in these areas because they will likely have the highest impact on future productivity and innovation.
Work with the right vendors – A few data-focused vendors utilize statistics to test hypotheses (Visier and Entelo come to mind). Speed up your transition by identifying and working exclusively with vendors that can help you use data and machine learning to discover what works and what doesn’t.
Be prepared for vigorous HR resistance – Even more than operating managers, I have found HR professionals to be extremely conservative. Don’t be surprised when HR professionals vigorously challenge even the strongest evidence about the validity of an assumption or the effectiveness of a current best practice. Be aware that experimentation also requires HR leaders to accept a high level of tolerance for risk and to relentlessly find a way around every obstacle and resistance factor that will block a new idea.
Move towards becoming a scientific HR function – Hypothesis testing and experimentation all require a significant amount of data. Adopting a hypothesis testing approach should be only part of HR’s overall efforts to follow Amazon’s lead in becoming more scientific. The overall goal should be becoming a data-driven function that measures its business impacts in dollars. It’s also important to remember the increasing importance of data. You can’t utilize AI, machine learning or most new technologies unless you first gather copious amounts of data.
A sample hypothesis testing template
If you’re wondering about the process, here is a sample hypothesis testing template.
- Business problem: High turnover rate and low productivity among new hires.
- Hypothesis to test: The amount and length of onboarding impacts new hire turnover and performance.
- Required decision: Whether to make a significant investment in extending onboarding.
- Data required: Data from an A/B test comparing simultaneous short and long-form onboarding results in areas of retention and productivity.
- Data supported insight gained: Stretching out onboarding has a measurable impact on new-hire productivity and retention.
- Suggested future action: Stretch out onboarding and determine how many days are optimal.
The goal of every HR leader should be to help their firm dominate their industry as a result of excellent people management. It should also be obvious that you won’t be able to make a significant contribution to that dominance. Unless you are providing a competitive advantage by continuously replacing your HR approaches with innovative ones that produce a significantly higher business impact. Since almost no one in your industry is likely to be currently using this hypothesis testing approach. Adopting it will provide you with an immediate competitive advantage and increased executive support because you will have hard data showing your impact on business results.
Incidentally, if you’re looking to adopt other tenets that distinguish great HR departments, they include transitioning to a data-driven model, focusing on increasing and quantifying your business impacts, adopting AI and other new HR technologies and prioritizing your HR actions that directly increase innovation and corporate revenue.