Why is getting the data right the only way to get people decisions right?

When companies say “data-driven HR” they often mean dashboards with turnover, hiring speed and costs. Those numbers matter, but they’re not the real goal.

The board should be asking: How do we make people decisions that increase productivity, revenue and customer outcomes, and make them faster?

The answer is simple: Define the insight, measure what creates impact and act on evidence.

1. Start with the decision you need to make

Too often HR starts with the systems they already have in place. Data in your HR system tells you what happened: hires, attrition, payroll, time-to-fill. That’s useful, but not enough. First, name the business outcome you want to improve (for example: sales per FTE, NPS, margin per rep). Then ask: what would we need to know to improve that outcome? Only after that, we collect the right data to measure those insights. If boards treated OPEX like CAPEX, asking for expected returns and evidence before spending, HR investments would change (1,2).

2. Move from gut to evidence

Managers' intuition and experience matter. Yet, they don’t replace evidence.

Many leadership frameworks are built from opinions: “what leaders think matters.” Instead, build frameworks from data: test which competencies actually link to business results. The academic and applied literature on selection and prediction shows that structured, validated measures outperform unstructured opinion when predicting job performance (3,4,5,6).

We run statistical checks that measure how behaviors and skills relate to outcomes like revenue, client satisfaction, or time-to-ramp. That needs structured inputs: comparable competency ratings, clear outcome metrics, and enough linked data to test the ideas. Do this right, and you stop training skills that don’t move the needle. Over the last ten years, we have transformed this into a scalable process that requires minimal extra effort from the organization.

3. Collect better data: quality over quantity

Collecting lots of data looks effective, but volume alone doesn’t solve the problem. From a decade of work and hundreds of thousands of datapoints, we learned two clear things:

• Common inputs often don’t predict outcomes. Things like years of experience, resume keywords, and single manager ratings frequently have low predictive value to performance. In particular, managers’ ratings often don’t match real business results and, sometimes, promotions are more based on personality (e.g., extroversion) than performance (7).
• Multi-rater behavioral observations do predict outcomes. When multiple people describe someone’s behavior (360° observations) and you combine those ratings, the link to business KPIs becomes strong and useful. This makes development and selection much more effective (8).

Example: one client spent a lot on teamwork training. However, our analysis showed that teamwork had nearly zero impact on sales for their roles. That training was unlikely to pay off. When the client shifted to competencies that did matter, the training produced measurable gains (9).

4. Turn measurements into predictive models and ROI

Once you measure the right things, you can build AI models that predict who will deliver on your strategic metrics. That’s when HR becomes a competitive advantage: hiring, promoting, and developing people who match the data-driven profile for success.

We’ve seen this in practice: in some client cases, hires selected using models delivered much higher revenue, in one group nearly 94% more than in the first two years. That turns talent decisions from a cost into a lever for growth (10).

This only works when analytics, governance, and HR processes are aligned.

Ethics, fairness and governance

When you move from intuition to prediction, governance matters. A.I. is often perceived as being only as good, or as biased, as the data and opinions it is trained on. Yet, while that perception holds some truth, it is far from the full story.

You can avoid bias in different stages of the modelling process: 

1. When collecting data,

2. When modelling,

3. During implementation.

Our work has shown that when designed intentionally, algorithms can be trained in ways that reduce human bias rather than reinforce it. We’ll cover fairness and bias controls in a separate blog.

Final thought - The new ROI of people decisions

Boardrooms that demand evidence will see HR shift from a cost-center to a strategic value driver. The work starts with defining the insight, then aligning data collection and analysis to business outcomes. The companies that will win the next decade will be the ones where HR decisions are based on evidence. When done right, the results are not only better hires and development: it’s measurable business advantage.

Sources:

1. Deloitte (2018). Global Human Capital Trends — People data and analytics overview.
2. McKinsey (2023). Performance through people: transforming human capital into competitive advantage.
3. Garcia-Arroyo, J., & Osca, A. (2021). Big data contributions to human resource management: a systematic review. The International Journal of Human Resource Management, 32(20), 4337–4362.
4. Garg, S., Sinha, S., Kar, A. K., & Mani, M. (2022). A review of machine learning applications in human resource management. International Journal of Productivity and Performance Management, 71(5), 1590-1610.
5. Polyakova, A., Kolmakov, V., & Pokamestov, I. (2020). Data-driven HR Analytics in a Quality Management System. Quality-Access to Success, 21(176). https://doi.org/10.1080/09585192.2019.1674357
6. Schmidt, F. L. & Hunter, J. E. (1998). The validity and utility of selection methods in personnel psychology: Practical and theoretical implications of 85 years of research findings. Psychological Bulletin.
7. Murphy, K. R. & Cleveland, J. N. (1995). Performance appraisal: An organizational perspective.
8. Smither, J. W., London, M., & Reilly, R. R. (2005). Does performance improve following multisource feedback? A theoretical model, meta-analysis, and review. Personnel Psychology.
9. Burke, L. A. & Hutchins, H. M. (2007). Training transfer: An integrative literature review. Human Resource Development Review.
10. Pera — The Science behind AI recruitment (whitepaper). (GetPera) — the whitepaper and product science page describing benchmarks and methodology.