For context and to answer that question we need to look at the existing evaluation methods and results. First let's take a look at why this deeper scientific analysis of in-job evaluation is possible.
A convergence of technologies applied in unique ways enables performance predictions
There are many talent data, and business intelligence solutions available, intended to deliver valuable insights to help drive faster, more impactful business decisions. On the talent data side they do not measure the nuanced traits, and complexity of human behaviors that drive outcomes. They do not correlate to locally validated KPI outcomes to provide statistical and empirical evidence on what uniquely drives performance in a specific company role.
When combining local talent analytics and business intelligence, advanced multivariate regression machine learning algorithms identify models that maximize the prediction of actual on-job performance to guide human capital decisions.
An international team of data scientists, psychology professionals, IO consultants and leading experts on personnel performance underpin this unique approach. The Platform is designed as a decision support engine that ingests client business intelligence and employee talent data with the goal of hiring, developing, promoting and retaining people with the highest probability of being top performers.
An employee’s true on-job performance encompasses four domains with components that can be directly measured and incorporated in a Decision Support Platform. Scientifically quantifying potential, what someone is capable of, and will likely produce leads to greater confidence in human capital decisions.
Re-framing people assessments - a novel approach
We believe the current approach to talent selection and assessment of future on-job performance used by industry has great room for improvement:
1. Companies over-rely on resume screens and interviews when making hiring decisions. Neither of these methods has predictive validity and each of them introduce significant bias into the hiring process.
2. All hiring solutions using assessments utilize a one-size fits approach, none of which can predict actual on-job performance because they measure only a small handful of the characteristics that make us unique (e.g., traits, abilities, etc.).
3. Most assessments create generic hiring recommendations that lack context around the job role a candidate is being assessed for. Consequently, people are being denied jobs based on generalized benchmarks rather than client and role-specific validation data.
4. Too many talent evaluation tools oversimplify human behavior. These tools are built around assumptions such as more is always better, that good ‘traits’ are always good, and that all ‘traits’ are equal predictors - none of which hold up when validated with criterion data. People and their contextual performance are far more complex than that.
Our mission is to improve upon the status quo. To offer clients a contextually predictive, customized, bias-free solution that understands the complexities of human behavior and delivers talent intelligence to help enterprises optimize human capital.
The approach is overtly designed around predictive modeling. While the talent science adheres to well established principles of psychology, test-development and psychometrics when building the suite of talent evaluation measures, this solution is firmly planted in data science. The most common solutions in the hiring space are assessment-based. They are based on simple bivariate correlations which are used to create loose job-fit buckets(e.g., pass/fail; low fit vs. high fit), and simply are descriptive not predictive analytics.
From our perspective, the purpose of pre-employment talent evaluation and application to future on-job performance for current and prospective employees is not about the psychometric soundness of a simple assessment, but about the ability to accurately, precisely and consistently predict whether a candidate can and will perform in a job role for which they’re being considered. In short, while many talent evaluation tools exhibit excellent psychometric properties because of their specificity of measurement, they are inherently limited for the same reason. They are, quite frankly, too narrow, and therefore unable to measure the vast number of potential traits and attributes that capture human complexity and explain, more comprehensively, the variance in predicted job performance.
In contrast, our employee performance platform is designed from the ground up to generate predictability. While we start with a rigorous talent science foundation including the use of reliable construct valid psychological measures and criterion-validation, the end product is based on the use of advanced multivariate regression machine learning algorithms to identify models that maximize the prediction of actual on-job performance.
We offer a repeatable process that builds highly customized prediction models, not a repeatable assessment designed around explainable psychometrics. This is a different approach, a more predictive approach that is about helping clients hire and develop the best talent in an unbiased way.
The prediction engine can assess individual differences between people at a ‘forensic level'. That is because the system built is capable of assessing talent using over 400 broad attributes that cover personality, abilities, attitudes, soft-skills, interests, and more. We then configure each talent evaluation differently based on a single job role that’s being analyzed. This results in an ability to distinguish the unique combination and elaborate interaction between psychological measures including non-linear relationships that predict job performance with a high degree of accuracy.
Accordingly, its unnecessary to create exhaustive psychometric evaluations for all psychological measures because the same items are not used repeatedly in the prediction models. Not only would it be unfeasible to analyze and document the psychometrics of 400+ measures, it would also essentially be meaningless for a use case. Ultimately, the methodology is about what is predictive and forward looking.
Capturing the complexity of human behavior.
Performance Fingerprints use a ‘Scientist-in-the-loop’ talent and data science approach.
An example of 2 measured traits and correlation to a company performance KPI of revenue. A performance fingerprint typically comprises a combination and weighting of 11 to 25 traits and facets achieving a minimum correlation coefficient of 0.70. companies report predictions of KPIs track to within 85% of observed results. In recalibrated fingerprints, on a 3rd iteration companies are experiencing predictions within 90% of what actually ends up happening.
First, the Science Team conducts a job analysis to understand the role, the traits that help or hinder successful performance in the role, and the intricacies of a client’s performance data for that role.
Next, incumbent employees complete a custom-configured psychographic questionnaire in parallel with the collection of performance data for those same employees.
Finally, the talent data and the performance data are analyzed using a multivariate regression machine-learning algorithm to distinguish the unique combination of psychographic characteristics that best predict actual performance in the role.
Because a Performance Fingerprint is locally validated through this analytics process, it is generally Equal Employment Opportunity Commission (EEOC) compliant from the outset. Unlike traditional assessments that focus on simple linear relationships between traits and performance, a Performance Fingerprint truly represents the intricate interplay between specific psychographics and performance outcomes, capturing the complexities of human behavior ‘at work’ in a defined context.
The prediction engine is not a one-size-fits-all assessment or a one-time event. Performance Fingerprints are customized for each job role based on a process of local validation and continue to get more accurate and deliver more intelligence as they are re-calibrated at regular intervals. The Platform can quantify actual performance (e.g., predicted sales revenue) before a hire or promotion is made, establishing a repeatable, bias-free hiring formula that is improved in accuracy and integrity over time.
The ideal user for the prediction platform recognizes the importance of people analytics for making evidence-based decisions and the value that on-going validation has on the quality of talent decision-making. While the system was designed for hiring and promotion, it unlocks unseen value “potential” and three specific use cases
Improvement quality of job fit impacts performance throughout the entire employee lifecycle lifting almost all productivity metrics from ramp to retention.
Perhaps the strongest ROI impact a business can experience is elevating workforce performance. C-Suite, HR, Talent acquisition, Sales and Business Leaders can develop scientific and empirical evidence for insight for faster, better more reliable decisions to predictably inflect higher team performance.
Better Incoming Talent:
Quite simply, the system is designed to create a bias-free method for hiring the right talent with the greatest probability of being a top performer in a specific job-role.
Improve Existing Team:
Secondly, the talent intelligence from the platform provides data about who to promote, potential career paths, and how to invest enablement resources (e.g., training and coaching) most effectively, for promotion and career paths. As new Performance Fingerprints are added and new performance data ingested, the system identifies how effective employees will be in other job roles. As for enablement, the platform provides clear data about where candidates may struggle, providing clear direction for training and coaching. And because the predictive models make it possible to identify who is reaching their potential versus who is not, the platform provides intelligence about where to invest valuable resources.
The third use case is for more formal strategic workforce planning and people analytics. As the Platform continues to get smarter with more data, it can then begin to predict more complex longitudinal outcomes such as tenure, turnover, absenteeism, or low-base rate criteria like safety incidents. From employee fixed costs (salary, benefits, etc.) versus performance, leaders can rethink their organizational structures and human capital needs. For example, it may be possible to realize the same performance with a smaller but more capable team or for the same cost, bring on more junior but equally effective talent.