Understanding Our Scoring System
Our scoring system provides a fair and data-driven assessment of candidates by leveraging data from over a million reference checks. For each scorable question, we’ve established percentile-based benchmarks to categorize responses into three tiers:
Neutral (lowest 25%)
Good (middle 50%)
Great (top 25%)
This ensures that each candidate’s performance is measured against a comprehensive dataset, making the evaluation meaningful and comparative.
Weighted Scoring for Greater Accuracy
To emphasize the significance of certain references, we apply specific weights to their responses:
Manager References: Valued for their insight into a candidate’s performance and weighted at 150%.
Old Work Relationships: If a reference pertains to a work relationship that is more than three years old, a 50% penalty is applied to account for its decreased relevance. However, 90% of references fall within the recent three-year window, ensuring that most data remains current.
How Scores Are Calculated
We calculate a weighted average of responses for each candidate on every scorable question. Any ratings marked as “Not Applicable” by references are excluded to ensure accuracy. These averages are then aggregated to produce the candidate’s overall score. Thresholds for these overall scores are derived from extensive data, allowing us to classify candidates accurately into their respective categories.
The HiPeople Score is only calculated after at least two references have been completed. The score takes into account responses to key reference check questions, including:
Overall Performance
Leadership Skills
Hiring Recommendation
Custom Skill Questions
If none of these questions are included in the reference check, the score is not available.
Insights from Our Data
Our analysis of 63,000 candidates shows:
Quality Matters More Than Quantity: There is no correlation between the number of references provided and the overall score, indicating that the quality of references is more important than their quantity.
Number of Questions Answered: There is a small negative correlation (-0.1) between the number of questions answered and the overall score. While this trend is minor, we continue to monitor our scoring system to ensure it remains fair and effective.