# Predictive Hiring Analytics ExplainedPredictive Analytics in Smooth

Hiring uses data from your pre-employment assessments to forecast candidate performance, identify hiring trends, and optimize your assessment strategy over time. This page explains the predictive capabilities built into the platform.

---

## What Is Predictive Analytics?

![Analytics](/files/flow-analytics85386b.png)

Predictive Analytics goes beyond reporting what happened — it analyzes patterns in your assessment data to:

- **Forecast candidate success** — predict which candidates are most likely to succeed in the role
- **Identify hiring trends** — track patterns in candidate quality, assessment effectiveness, and hiring funnel performance
- **Optimize assessments** — determine which assessments are the best predictors of job success
- **Benchmark performance** — compare current candidates against historical cohorts

---

## Where Predictive Analytics Appear

### Assessment Insights Dashboard

The [Assessment Insights](/help/assessments-predictive/assessment-insights) dashboard is the primary home for predictive analytics, offering:

#### Activity Trend Analysis

A time-series chart showing:

- **Invited candidates** over time
- **Completed assessments** over time
- **Average score trends** — rising or falling candidate quality
- Period-over-period comparisons to identify improvements or declines

#### Conversion Funnel

The conversion funnel tracks candidates through four stages:

1. **Invited** — candidates who received an assessment invitation
2. **Accessed** — candidates who clicked the assessment link
3. **Started** — candidates who began answering questions
4. **Completed** — candidates who submitted the assessment

Drop-off rates between each stage reveal where candidates are leaving the process:

- **Invited → Accessed** — low conversion may indicate email deliverability issues or poor timing
- **Accessed → Started** — candidates may be deterred by the assessment landing page or technical requirements
- **Started → Completed** — high drop-off may indicate the assessment is too long or too difficult

#### Conversion Sankey Diagram

A visual flow diagram showing the same conversion data in a Sankey format, making it easy to see the relative volume at each stage and where candidates exit.

#### Score Distribution Analysis

A histogram of candidate scores showing:

- How many candidates fall in each score range
- The distribution of flagged candidates across score bands
- Whether your assessments are producing a useful distribution or clustering too many candidates in one range

#### Score Stream (Time-Based Bands)

A stacked area chart showing how the distribution of High, Medium, and Low score bands changes over time, revealing:

- Whether candidate quality is improving or declining
- Seasonal patterns in assessment performance
- The impact of changes to your assessment strategy

---

## Predictive Survey Analytics

[Predictive Surveys](/help/assessments-predictive/personality-job-fit) provide specialized predictive capabilities:

### Role Fit Prediction

Based on personality trait analysis, the predictive model classifies candidates into fit levels:

| Fit Level | Prediction |
|---|---|
| **Strong** | Candidate is highly likely to succeed in the role |
| **Good** | Candidate has solid potential with minor development areas |
| **Fair** | Mixed signals — additional evaluation recommended |
| **Weak** | Significant gaps between the candidate's profile and role requirements |

### Trait Cohort Comparison

Each predictive trait is compared against cohort averages, showing:

- Traits where the candidate is above average
- Traits where the candidate is below average
- The magnitude of deviation from the norm

### Job Fingerprint Matching

A visual "fingerprint" chart maps:

- The job's requirements across multiple dimensions
- The candidate's assessed capabilities on those same dimensions
- The degree of overlap, which drives the fit prediction

---

## Time-Based Analytics

### Completion Timing Patterns

- **Completion Time Distribution** — histogram showing how long candidates take, with average and median statistics
- **Time to Start** — how quickly candidates begin after receiving an invitation
- **Completion by Day of Week** — which days see the most completions
- **Completion Heatmap** — a day-of-week × hour-of-day heatmap showing when candidates are most active

These patterns help you:

- Set appropriate assessment time limits
- Time assessment sends for maximum completion rates
- Understand your candidate pool's behavior patterns

---

## Assessment Effectiveness Analytics

### By Assessment Performance

The Assessment Performance table in [Assessment Insights](/help/assessments-predictive/assessment-insights) shows per-assessment metrics that help you evaluate which assessments are most effective:

| Metric | What It Tells You |
|---|---|
| **Completion Rate** | Whether the assessment is too long, too difficult, or has technical issues |
| **Average Score** | Whether the difficulty is calibrated correctly |
| **Integrity Flagged** | Whether security settings need adjustment |
| **Invited vs. Completed** | The overall effectiveness of the assessment in your pipeline |

### By Type, Category, and Difficulty

Breakdowns by these dimensions help you identify:

- Which assessment types produce the best signal
- Which categories are most predictive of success
- Whether difficulty levels are calibrated appropriately

---

## Integrity Predictive Signals

The Integrity tab provides predictive signals about assessment reliability:

- **Integrity Breakdown** — proportion of clean vs. flagged assessments
- **Integrity Event Types** — specific types of suspicious activity
- **Flagged Candidates** — candidates with the most integrity concerns
- **Multiple Flags** — candidates with multiple types of integrity issues (highest risk)

High integrity flag rates may indicate:

- Security settings need to be strengthened
- The assessment environment instructions need improvement
- The candidate pool may benefit from proctored assessment options

---

## Using Predictive Data

### Optimizing Your Assessment Strategy

1. **Monitor completion rates** — if rates drop below 70%, consider shortening assessments or adjusting difficulty
2. **Track score distributions** — a bell curve indicates good calibration; heavy skew suggests the assessment may be too easy or too hard
3. **Review conversion funnels** — identify and address the biggest drop-off points
4. **Compare assessments** — use the per-assessment table to determine which ones provide the best signal for your roles

### Making Better Hiring Decisions

1. **Prioritize candidates** by assessment signal and cohort standing
2. **Weight predictive survey fit** alongside traditional assessment scores
3. **Consider integrity alongside performance** — a high score with integrity flags is less reliable
4. **Use AI recommendations** as a starting point, then review details

---

## Related Pages

- [Assessment Insights](/help/assessments-predictive/assessment-insights)
- [Scoring Profiles](/help/assessments-predictive/scoring-profiles)
- [AI Candidate Recommendations](/help/assessments-predictive/ai-recommendations)
- [Personality Traits & Job Fit Report](/help/assessments-predictive/personality-job-fit)
- [Job Fit & Assessments](/help/assessments-predictive/job-fit-assessments)