Descriptive & Diagnostic Insights: What Happened and Why

Andras Rusznyak

artificial intelligence expert

Ha magyarul szeretnéd olvasni a cikket, kattints ide

In our last article we explained why laying the foundations of clean data, well defined metrics and objectives is critical. Building on these foundations, most HR teams begin with descriptive analytics—reports that tell you what happened, such as headcount changes, turnover spikes, or training completion rates. But reporting alone leaves gaps: it shows symptoms without revealing causes.

Enter diagnostic analytics, which asks why these patterns emerged. Why did attrition spike? What triggered engagement dips? Why has absenteeism risen past the norm? Diagnostic analytics connects the dots and empowers HR to act with clarity and confidence, not guesswork — bridging the gap between awareness and action.

IMPORTANT NOTE:

We utilized generative AI in the making of this article.

🔍 Descriptive Analytics – “What happened?”

• Answers basic questions, like: “Turnover was 12% last quarter.”

• Uses summary statistics (e.g., headcount trends, average training hours) to identify anomalies and establish benchmarks.

🧠 Diagnostic Analytics – “Why did it happen?”

• Seeks explanations: “Attrition rose in Team A following a manager change and two negative engagement surveys.”

• Uses correlation, root-cause analysis, and hypothesis testing to uncover systemic factors.

These stages are part of a layered analytics stack, where “what” leads to “why,” then to prediction and prescription.

What is Descriptive vs Diagnostic analytics?

Why HR Needs Both — The Strategic Advantage

Relying solely on descriptive analytics leaves HR reactive—always reporting, never solving. You know something happened (e.g. turnover rose), but you don't know why, which makes intervention ineffective.

By performing diagnostic analytics, you uncover whether root causes like poor leadership, low engagement, role stress, or career stagnation are driving the issue. You then can target solutions, such as leadership coaching or role redesign, where they matter most.

Common diagnostic techniques in HR

Practical Example: Diagnosing high turnover in Customer Support

Here are core diagnostic methods HR teams can practically apply:

1. Root-cause analysis

   • Use exit surveys or pulse surveys to map common dissatisfaction drivers (e.g., work-life balance, role clarity).

2. Drill-down analysis

   • Segment by department, tenure, location, job level to pinpoint where issues concentrate.

3. Correlation analysis

   • Analyze links: Do low engagement and high turnover cluster together?

   • E.g., employees rating manager support below 3/5 are twice as likely to leave.

4. Hypothesis testing

   • Formalize a cause—like “managerie turnover increases team flight risk”—then test it (e.g., T-tests).

   • Helps distinguish meaningful signals from random variance.

5. Cohort and time series analysis

   • Track groups (e.g., hires by quarter) to see how performance or retention trends evolve relative to policies or events.

These techniques are accessible using spreadsheets, SQL, or BI tools—no advanced software required. What matters is disciplined thinking and dataset structure.

Scenario

In Q1, turnover in Customer Support rises sharply from a 11% average to 16%, prompting concern.

Step 1: Descriptive — Confirm the trend

What you do:

• Calculate overall turnover for Customer Support last quarter vs. usual.

Data required:

• HRIS data: termination dates, department codes, headcount by period.

Tools & skills:

• Use Excel pivot tables or BI filters (e.g. Looker, Power BI).

• Understand simple aggregation (turnover rate = exits ÷ avg headcount).

This confirms the issue and focuses the analysis on one team and timeframe.

Step 2: Drill-down — Find where it’s most severe

What you do:

• Break down turnover by tenure, location, or hire date cohort.

Data required:

• HRIS: hire date, termination date, location, tenure calculation.

Tools & skills:

• Pivot tables with filters, or BI tools for cohort charts.

• Basic knowledge of slicing and dicing data by segments.

You discover ~80% of exits are new hires (<6 months) in Office X.

Step 3: Correlation & Root-Cause — Link causes

What you do:

• Compare turnover with engagement survey scores or manager ratings from onboarding surveys.

Data required:

• Engagement data: survey responses linked to employee IDs.

• HRIS: manager assignment, hire cohorts.

Tools & skills:

• Create scatter plots or correlation matrices in Excel or BI.

• Basic stats: calculating correlation coefficient (e.g., Pearson’s r).

You see that new hires with manager support scores ≤2.4/5 leave at much higher rates than those scoring ≥3.8.

Step 4: Hypothesis Testing — Confirm significance

What you do:

• Test: "New hires with low manager support have higher turnover than others."

Data required:

• Two groups: low manager support vs. higher support — with retention outcomes.

Tools & skills:

• Conduct t-test in Excel (Data Analysis Toolpak), Analyse-it add-in, or open-source tools like Rattle GUI for non-coders.

• Interpret p-values (<0.05 means statistically significant difference).

The test confirms the difference is unlikely due to random variation.

Step 5: Theory — Shape the cause-and-effect narrative

From the evidence, you form a hypothesis: a recent leadership change left new hires without sufficient support in Office X during early months, driving voluntary exits.

Required competencies & tools:

• Data storytelling skills to summarize findings clearly.

• Presentation tools (PowerPoint or slides within BI dashboards) to communicate causality to stakeholders.

Turning insight into action

Once you understand why, you can craft targeted interventions:

• Short-term: Mandate managers to hold weekly 1:1s with new hires during the first 90 days.

• Medium-term: Pilot a buddy/mentorship program paired with manager training on early-stage support.

• Assessment: After 3–6 months, re-check new-hire turnover, engagement, and manager satisfaction metrics.

This structured loop—from data to hypothesis to action to evaluation—transforms HR into a strategic architect, not a reactive unit.

Pitfalls to avoid

Scaling your diagnostic capability

Once you’ve built trust and demonstrated results, you can automate and scale:

• Alerts & thresholds: Dashboard triggers when metrics deviate (e.g., turnover +3% over baseline).

• Linked surveys & outcomes: Combine engagement, exit, and onboarding survey responses with turnover data to identify patterns automatically.

• Anomaly detection: BI tools like Looker or Power BI can flag unusual deviations for investigation.

This evolving maturity allows HR to shift from periodic reviews to proactive monitoring and intervention.

Summary & key takeaways

• Descriptive analytics answers what happened. Diagnostic analytics explains why it happened. Together, they empower meaningful interventions.

• Techniques like drill-down, correlation, hypothesis testing, and root-cause analysis are accessible, even without sophisticated tools.

• Real-world example from Customer Support demonstrates how a few diagnostic steps led to targeted actions and potential retention improvement.

• Diagnostic analytics builds credibility and opens the way to predictive and prescriptive practices.

The heart of insight: moving beyond numbers to understanding

Why This Matters

• Transparency: You’re not guessing—every insight is backed by data and statistics.

• Clarity: Stakeholders can see what and why, and then follow clear logic to how to act.

• Scalability: The same framework (descriptive → drill-down → correlation → test) can be reused for any HR question, from engagement dips to training completion.

Up next: In our next article, we’ll explore Predictive Analytics—where HR goes from understanding past events to forecasting future trends and staffing needs.

Have you read our other articles? Go to Motioo Insights

Do you have any questions or comments? Contact us