A Data-Driven Approach to Advancing Meritocracy
Instead of simply relying on best practices, employers should adopt a talent management strategy that addresses bias and inequity while ensuring efficient, fair, and merit-based decisions
Topics
News
- When Cables Snap: Decoding the Red Sea’s Internet Disruption
- Apple’s New Products Reinforce its Long-Term Bet on Intelligence and Sustainability
- Global Spending on Generative AI Smartphones to Hit Nearly $300 Billion This Year
- Hero MotoCorp names Chitale as CEO
- Mistral AI Raises €1.7B in Series C Funding Led by ASML
- Databricks Forecasts $4B in Annual Revenue, Fueled by AI Boom
Building meritocratic and equitable organizations is a complex yet critical endeavor, as many corporate leaders have come to realize. It requires effective talent management systems to attract, develop, and retain qualified and motivated individuals — key drivers of organizational success. In my book, The Meritocracy Paradox, I caution that certain organizational efforts to foster meritocracy and excellence in organizations may paradoxically deepen inequities and unfairness. And I have presented evidence of three key findings — along with their related warnings — that highlight what I call the meritocracy paradox.
The first warning is that simply having organizational processes in place to hire, evaluate, and promote the best does not automatically guarantee fairness. In fact, any talent management process can be subject to bias and inefficiencies, and there is a risk that, rather than fostering excellence and opportunity for all, people-based management systems may actually reinforce or create advantages for certain groups over others.
The second warning is that emphasizing meritocracy — whether implicitly or explicitly — as the foundation of hiring, promotion, and reward practices may backfire on women, racial minorities, immigrants, and other historically disadvantaged groups. When individuals believe their organization is meritocratic, they may be less likely to recognize and correct for biases in their decision-making. This can lead to unfair treatment of certain individuals or groups and, inadvertently, exclude candidates whose skills and talents merit inclusion.
The third warning is that there is no universal agreement on what merit actually is. Even managers and executives with similar training and experience within the same organization often hold differing views. This lack of consensus on what constitutes merit or talent can ultimately derail efforts to build a truly meritocratic organization.
The encouraging news is that fostering true meritocracy in the workplace does not entail an extravagant amount of time or resources but rather a strategic and intentional focus on debiasing and improving talent management processes.
Taking action is essential. But pressure to act often leads companies to implement generic solutions — such as diversity and implicit bias training, blind selection and hiring, tweaks to job posting language, and the use of AI recruitment tools, all of which have been shown to have limited effectiveness — without first diagnosing the organization’s specific challenges or needs. The “best practice” approach falls short because it fails to account for the specific context of an organization. In this article, drawn from my book, I propose a more effective approach — a data-driven talent management strategy that actively addresses bias and inequity while ensuring efficient, fair, and meritocratic decision-making.
An Approach Grounded in Talent Analytics
My strategic approach, based on talent analytics, aims to foster meritocracy — that is, organizational systems that reward and advance individuals solely on the basis of their demonstrated intelligence, efforts, skills, abilities, or performance, without regard to their demographics or personal characteristics. By collecting, coding, and analyzing employment-relevant data on people-related processes and outcomes, meritocratic organizations strive to improve fairness and equal opportunity in hiring, advancing, and rewarding individuals — three key career outcomes that not only benefit individuals but also contribute to the success of organizations.
Two conditions are essential for achieving meritocracy in practice. First, there should be equal opportunity for all individuals at crucial decision points concerning selection, advancement, and rewards. A talent analytics approach can support this first goal by identifying any people-related processes that may be restricting equal access to opportunities. Second, once equal opportunity has been established — ensuring individuals can attain positions and rewards based solely on their merit — inequality in outcomes may be acceptable in a meritocratic system. The main principle here is that disparities in pay, rewards, and promotions should be driven by employment-relevant factors rather than demographics or personal characteristics. A talent analytics approach can play a crucial role in this regard, helping organizations assess whether their people-related processes are operating fairly and pinpointing any areas where bias and other inefficiencies may be present.
A note of clarification, even caution, is warranted. Implicit in the first condition is that opportunity for all is a win-win type of situation, where everyone stands to benefit equally from organizational efforts aimed at creating opportunity, such as employer initiatives that provide training and employment benefits. But the second condition acknowledges that the distribution of positions and rewards, particularly in the workplace, could be a zero-sum game, in that organizations have a finite number of jobs and promotions to offer, as well as fixed budgets for raises and bonuses, and consequently not everyone will “win” those jobs, promotions, or rewards. In such zero-sum situations, and to foster meritocracy, it is therefore fundamental that organizations guarantee that everyone has an equal chance to compete and succeed.
Getting to a Great Talent Analytics Strategy
Many organizations and businesses, even large and successful ones, lack a strategic approach to talent management that identifies and solves meritocratic and fairness challenges. I present five crucial steps to help you create and develop this powerful strategic approach toward fostering meritocracy in your own organization.
Step 1: Identify, Develop, and Define Key Criteria. You should follow this process for each primary employment decision in your organization. When recruiters and managers screen application materials, they should be clear about what criteria — such as required qualifications, experiences, and skills — the position requires. Some use the word competencies to refer to these criteria and the term competency modeling to refer to the process of defining such criteria. Get specific (and realistic) here. Without such clarity, biases and social processes may lead screeners to look for applicants in a limited number of ways or apply different criteria depending on the candidate, potentially reinforcing biases and social barriers.
As part of this first step, it is necessary to assess whether you observe significant demographic differences in the attainment of the criteria used to screen and select hires. If that is the case, check whether those criteria should be used, especially if only some groups have had an opportunity to meet such criteria in the past. If such criteria are essential (as they often are), consider providing the necessary resources for all hires to learn about those criteria (or become better at them) so that, ultimately, all hires have an equal opportunity to succeed. In this respect, post-hire resources and training can help level the playing field. Indeed, many organizations have implemented specific onboarding, training, and development programs precisely for this reason.
But this first step is not always straightforward. In many jobs and occupations, you need people who possess specific skills or credentials, which can lead to demographic imbalances in the pool of qualified applicants. Additionally, simply knowing the criteria for specific jobs doesn’t ensure an equal opportunity to obtain them; many existing societal barriers are difficult for individual organizations to overcome alone. For example, suppose someone enters a medical organization as a certified nursing assistant and is already an adult with children and family responsibilities and no college education. In that case, there is very little chance they will become a doctor, even if they have access to information on how to pursue a career in medicine. In these situations, the challenges are clearly societal and difficult for any one organization alone to address — meritocratic societies and institutions, for instance, could provide the opportunities and resources needed for talented individuals to attend college and medical school.
That said, sometimes not requiring specific qualifications or credentials can be an effective way to tap into a wider pool of talent. For example, in 2023, the state of Pennsylvania ceased requiring college degrees for certain jobs, and other states, such as Utah, Maryland, and Alaska, followed suit. Researchers and labor advocates have urged employers to eliminate college and university degrees for jobs that do not actually need them.1 Companies like AT&T, Mastercard, Microsoft, and Southwest Airlines have also created alternative pathways to good, stable jobs for those without a college degree.2
In implementing this first step, it is advisable to follow a holistic approach when selecting and hiring individuals, focusing on the employee as a whole rather than on specific skills and abilities, if feasible. Many large companies and undergraduate and graduate programs use this approach to ensure a talented cohort that is professionally and demographically diverse. Candidates are screened not only on a few abilities required to do a good job currently but also on their potential to do a great job in the future.
Simply knowing the criteria for specific jobs doesn’t ensure an equal opportunity to obtain them.
This holistic approach, therefore, takes a long-term and often strategic perspective for the management of talent in organizations. It allows leaders to experiment and figure out whether specific abilities required in the past are still necessary. It can offer an opportunity for organizations to discover untapped pools of candidates that competitors are not yet aware of, as they often rely on narrow, rigid, and outdated hiring criteria. For instance, when recruiters and hiring managers establish selection thresholds, such as a certain number of years of work experience or a particular credential or degree, they may have to revise those thresholds after examining the pool of applicants and the subsequent performance of the hires.
The same approach can be applied when making post-hire decisions related to bonuses, promotions, or development opportunities. Many employers, in this regard, could benefit from blinding irrelevant information when making promotions or awarding bonuses. Others could follow a more holistic approach that considers the employee’s current ability to do the job and how that employee could develop in the future.
Once again, however, this holistic approach may work only in certain situations — for instance, organizations and businesses that hire many individuals for relatively entry-level or midlevel professional positions, or those that periodically consider a large number of promotions or bonuses. This approach may not work when you are hiring for high-level or senior positions or when specific skills and qualifications are an essential part of the job.
Without first locating where biases and barriers may exist in your organization, it is difficult to know where to focus your efforts.
First, it is important to collect individual data (often including demographic or personal information) for both employees and candidates applying for your jobs. Getting this data is frequently challenging when it has not been collected previously; often, in certain countries, lawyers recommend against getting such data (the thinking being that if a company does not have the data, it cannot be found legally liable in discrimination cases). That said, many institutions, experts, and executives have encouraged collecting and reporting demographic data, especially for midsize and large organizations, to ensure that there is no discrimination against particular groups of employees who are protected by national, regional, and local employment laws and regulations.3 (Please be sure to carefully review the legal constraints regarding what data you are allowed to collect, as regulations and rules vary depending on the jurisdiction in which your organization is located.) Many of these recommendations even promote the exercise of transparency — that is, making such data and analyses publicly available in easy-to-understand formats to show how much progress organizations are making toward equity and fairness.4
Individuals may be rightly reluctant to provide demographic data if they have experienced bias and unfair treatment in the workplace in the past. However, as an organization proves over time that it is serious about correcting imbalances, addressing barriers, and giving opportunities for all, individuals may feel more comfortable about sharing such information. Although it is difficult to find a comprehensive list of companies where employees feel comfortable sharing their demographic information, some companies have publicly shared their efforts to promote diversity and equal opportunity for all. They may even have initiatives in place to encourage employees to disclose information.5
Such data can also help address concerns about reverse discrimination, where members of a dominant group might feel they are being unfairly treated in favor of historically disadvantaged groups. This is one common reason for backlashes against DEI practices: the argument that many such practices unfairly favor disadvantaged groups. The approach I propose here can indeed help identify situations where this could be the case and, consequently, help organizations intervene to resolve such tensions. A truly meritocratic approach should help everyone have equal opportunity by eliminating biases and improving talent management processes. That approach requires ensuring that no groups of individuals get unfairly penalized or favored in the pursuit of meritocracy.
- Recruiting outcomes. These should include the number of applicants for any given position, their professional and personal background, and their relevant skills, abilities, and experiences for the position.
- Screening and selection outcomes. These should include who advances in the selection process at the interview, offer, and offer acceptance stages, for instance.
- Post-hiring outcomes. These should include performance ratings, promotions, transfers, terminations, base pay, pay increases, bonuses, and benefits.
These are the most common employment-relevant outcomes to track. However, it is up to you and your team of professionals to identify your primary employment outcomes/stages that can affect the careers of your applicants and employees. This is a critical part of this second step because it will later allow you to critically examine the extent to which your organization is meritocratic and fair in making employment decisions.
Measuring employment-related outcomes is as important as measuring and collecting information about the inputs and the processes behind the attainment of such outcomes. For instance, it is critical to measure inputs about job seekers and employees at the time of hiring and beyond to learn about their relevant skills and abilities. This task can typically be accomplished by capturing and coding information from resumes and other employment-relevant application materials. In the case of employees, such data can be complemented with training and experience within the organization that enhances their contributions and performance. All of this information is relevant for Step 3, which involves analyzing the data to identify any disparities — and, if they exist, the reasons behind them — so that you can find the best solutions to any challenges uncovered.
Furthermore, while collecting this information can assist companies in identifying areas for improvement and monitoring progress over time, it is imperative to do so in a way that respects employees’ privacy and ensures that the information is used responsibly and ethically. (Once again, review the current legal constraints regarding what data you are allowed to collect given your organization’s location.)
Such collection efforts also need to be preceded by informing employees and being transparent about why the data is being collected, how the data will be used, and how the analyses of the data will be reported. The data collection processes, analyses, and results also need to reinforce that privacy will always be protected.6
Step 3: Analyze Collected Data Not Only on Outcomes but Also on Processes. Once you have collected and stored the information over time in a database, you can start analyzing that data. First, you can explore aggregate patterns in each of the measured employment outcomes by variables of interest. For example, I have seen companies calculate the percentage of applicants who receive an interview by gender and race and detect important disparities. In the case of base salary, a simple approach consists of computing the average base salary by gender and then testing whether any observed base salary differences are statistically significant.
Every single employment-related decision point could potentially introduce unintended and undesirable biases and social processes.
This multivariate modeling strategy ultimately allows you to compare individuals with the same control variables. For instance, to analyze who gets a merit-based bonus or a promotion, you may want to compare employees with the same jobs, performing at equivalent levels, and account for all other factors that could influence the reward or promotion outcome. This is why steps 1 and 2 are essential, as they enable you to consider, evaluate, and collect data to estimate such models.
As an example, the gender pay gap is a commonly cited statistic. According to a 2024 Payscale analysis of salaries of over 627,000 individuals in the United States, women earned 83 cents for every $1 earned by men when comparing median salaries.7 That number was reported by Payscale as a measure of the “uncontrolled gender pay gap” because such a statistic does not control for different types of jobs or qualifications. Its reported “controlled gender pay gap” statistic reflects that such a gap is estimated to be much lower, with women earning 99 cents for every dollar a man earns when controlling for such factors. The key here is to determine which variables you need to incorporate in your analyses (i.e., to control for) in order to accurately compute the demographic gap in your organization. Those variables can be job title, education, professional requirements, skills, and geography, among many others. Then you can decide whether the factors you consider essential indeed determine salary and whether, on top of such factors, demographics still play a role.
Another important part of this analysis is to investigate and check every single people-related process and practice in your organization. At first, you may not have enough information on relevant variables, but the more you conduct this investigation strategically, the more you will revisit steps 1 and 2 to refine your analytical data-driven approach. Critically, these checks must take place with some regularity because processes that were once effective but have not been properly updated can often become distorted or less effective over time. Like your car, your talent management system is constantly being affected by external forces, and parts of it periodically break down under those pressures. In this regard, every single employment-related decision point could potentially introduce unintended and undesirable biases and social processes that hinder your organization’s progress toward becoming meritocratic, fair, and excellent. It is important to regularly focus on key people-related processes:
- Processes used to find and recruit diverse, talented applicants. This might include which recruitment sources you use to attract job seekers, what kind of messages and information about your company and your jobs could influence who applies and who does not apply, and what activities and processes are being used by your recruiters and hiring managers.
- Processes and criteria used to select candidates for the next steps in your selection process. For example, data on who gets their application reviewed, who gets interviewed, who receives an offer, and who ultimately accepts the offer and becomes a prospective hire can provide valuable insights. Here, pay close attention to each step a successful candidate follows and the extent to which all candidates alike have equal opportunity to advance to the next stage. If certain groups of candidates are not progressing, investigate which factors drive the selection process and the extent to which such factors are employment-relevant, valid, reliable, and useful.
- Processes aiming to maximize the number of offers accepted by applicants, typically by improving recruitment and onboarding activities. Many businesses ignore this part of the selection process, even though their challenge may be convincing potential employees to join the organization. When particular groups of candidates decide not to join an organization, even when given a good offer and hiring package, for example, the organization should consider the extent to which it is an appealing employer for everyone.
- Processes behind onboarding efforts and other training opportunities. These processes should be designed to ensure that every employee can succeed from the outset. Here, also guarantee that all hires alike have equal opportunity to be trained.
- Processes behind measuring and evaluating performance. These processes should clearly define and establish expected performance standards and targets that are achievable and relevant to the position.
- Performance measurement processes used for training and developmental purposes. Pay special attention to those procedures aimed at helping develop those who underperform. They should also be clear, consistent, and relevant for the success of the job and the organization.
- Processes used to reward those who meet or exceed standards, and also those used when deciding advancement and other employee career outcomes. This includes promotions, transfers, and terminations.
When you are analyzing such key talent management processes, I recommend examining not only how such employment decisions are being made but also who is ultimately responsible for them and how others (including managers and employees) may respond to them. You should also carefully examine the effectiveness of these organizational processes with the data you collect to see if they are working as intended.
Step 4: Decide Which Intervention to Employ. By monitoring people-related decisions and outcomes, organizations and businesses can learn how much they may be deviating from meritocracy and act on their findings. For instance, in one large global company I worked with, we analyzed all promotion decisions and merit-based pay increases based on employee performance tracked over a decade, assisted by a company team that put in place a system to collect, clean, and prepare data for analysis. We soon identified demographic patterns in both the promotion and the merit-based pay decisions. We first calculated promotion rates and average salary increases every year and then computed those numbers based on a few demographic characteristics to check for significant differences. We found a large difference between male and female employees in that women were not getting the same merit-based bonuses as men.
Upon further investigation of this finding, we concluded that the problem probably stemmed from differing asking rates. Men were more likely to ask for and ensure that they received their merit-based bonus, while women tended to trust that the bonus was already included in their paychecks. To solve this problem, one HR professional was asked to monitor and confirm that all employees would get the bonus automatically once the level of performance required for such a bonus was met, thus eliminating the need to ask for it. After this simple intervention, gender differences in merit-based bonuses disappeared.8
When assisting organizations in evaluating their meritocracies, I often encounter leaders and managers who wish they had gathered better employment-relevant data to analyze workplace trends and patterns, thereby gaining a deeper understanding and diagnosis of the observed outcomes. This step-by-step analytical framework becomes dynamic and interactive because discovery allows you to revisit steps 1 and 2 to reassess and gather additional information to better address your challenges.
Step 5. Stay Continually Alert and Monitor Results Regularly. Finally, implement processes that frequently alert you to potential future challenges that could affect the successful functioning of your talent management strategies and procedures in your organization. Accordingly, regularly reassess and reevaluate each of the prior steps. Because success criteria may change over time, and because technology, organizational practices, and labor markets are highly dynamic, Step 1 is very relevant in that it allows you to reassess and validate additional skills, abilities, merits, or talents that are necessary for hiring and, later, for promotions, as well as for rewarding top performers fairly.
References
1. For example, see J.B. Fuller and M. Raman, “Dismissed by Degrees: How Degree Inflation Is Undermining U.S. Competitiveness and Hurting America’s Middle Class,” PDF file (Accenture, Grads of Life, and Harvard Business School, 2017), www.hbs.edu; and R.C. Booth, “Stop Requiring College Degrees for Jobs That Don’t Need Them,” Vox, March 19, 2023, www.vox.com.
2. For more information about employers that are proactively advancing and developing workers, visit the American Opportunity Index website, www.americanopportunityindex.org.
3. For example, see N. Lewis, “Technology Can Be Used to Achieve Pay Equity,” SHRM, June 19, 2023, www.shrm.org.
4. Y. Slan, “Viewpoint: A Reflection on Juneteenth, Transparency in Diversity Reporting,” SHRM, June 16, 2023, www.shrm.org.
5. Some employees seem to feel comfortable sharing their demographic information in certain companies and organizations worldwide. As of December 2024, for instance, Salesforce was committed to DEI and regularly releasing data on its workforce demographics, encouraging employees to disclose their race, ethnicity, gender, and other demographic details. Similarly, Airbnb has implemented Project Lighthouse to collect data on discrimination and bias experienced by its users while also collecting demographic information from employees to inform its DEI initiatives. Microsoft and Target follow similar practices, encouraging employees to share demographic data. Target encourages employees to share their demographic information to help inform its diversity and inclusion initiatives and to track progress over time. However, as workplace policies, privacy concerns, and legal regulations continue to evolve worldwide, companies may continue to adapt their approaches to demographic data collection and DEI transparency.
6. See D. Habtemariam, “3 Must-Dos for Collecting Employee Demographic Data Beyond Race and Gender,” Senior Executive, April 21, 2022, https://seniorexecutive.com; B. Melloy, “What Demographic Question Should You Ask in Surveys?” Culture Amp, updated July 7, 2020, www.cultureamp.com; and K. Magoon, M.-J. Robinson, A. Kissling, et al., “Best Practice for Demographic Data Collection & Reporting: Evaluator’s Guide,” PDF file (Boston: Public Consulting Group, August 2022), www.publicconsultinggroup.com.
7. “2024 Gender Pay Gap Report,” PDF file (Boston: Payscale, March 2024), 5, www.payscale.com.
8. Google offers a similar example. As reported on its re:Work site, the company identified a gender disparity in one of its promotion cycles: Junior female software engineers were being promoted at a lower rate than their male peers. Google’s People Analytics team discovered that the issue was rooted in differing self-nomination rates. Since engineers at Google can self-nominate for promotion when they feel ready, the data revealed that men were doing so more frequently than equally qualified women. To address this problem, a senior leader shared the findings with employees and encouraged all engineers to self-nominate when they felt prepared. After this simple intervention, Google reported that the gender gap in promotion rates was eliminated.
