Analytics
Quantavist analytics helps HSE and site leadership understand where risk is concentrated, how quickly teams respond, and what consequences follow from safety events. All analytics pages share common filters and date controls so you can explore the same question from different angles.
Access analytics from the sidebar under Analyse:
| Page | What it answers |
|---|---|
| Risk analytics | How many alerts occurred, where they cluster, how risk compares across locations and subcontractors, and PPE compliance (separate from alerts) |
| Response analytics | How fast alerts move through the workflow and where bottlenecks occur |
| Consequence analytics | What enforcement outcomes followed — fines, site removals, toolbox talks |
Common controls
Every analytics page uses the same core tools.
Date range
Select the period you want to analyse using presets (for example last 7 days, last 30 days) or a custom range.
Compare to previous period
Turn on comparison to see whether metrics improved or worsened versus the prior equivalent window. KPI cards and charts show the delta — for example whether alert volume rose or median response time fell.
Filters
Narrow the dataset using:
| Filter | Purpose |
|---|---|
| Location | Limit to a zone or area of site |
| Alert level | Critical, High, Medium, or Low |
| Detection type | The kind of safety issue (for example working at height, plant–person proximity) |
| Source | Automatically detected or manually created alerts |
| Assignee | Alerts assigned to a specific person |
Quick presets are available — for example Critical only, Critical & High, or Manual alerts.
Cross-filtering
Click a segment in many charts (a location bar, detection type, alert level slice) to filter the entire page to that selection. Active filters appear as badges at the bottom of the screen. Clear them to return to the full dataset.
View alerts
When a chart or KPI is selected, use View alerts to open the underlying alert list in a side panel. This lets you move from a trend to individual records without losing your filters.
Risk analytics
Risk analytics focuses on alert volume and distribution, plus a separate PPE compliance KPI.
KPI cards
At the top you will see totals for all alerts and for each alert level (Critical, High, Medium, Low), alongside PPE compliance rate. The PPE headline combines all checked equipment types; its deep dive separates Overall PPE, Hard hat, and High-visibility vest. Click a card to drill into that metric across the charts below.
PPE compliance measures how consistently required PPE is worn. It is not derived from alert counts — see PPE compliance.
Charts
Typical views include:
- Open vs resolved — balance of active and closed work
- Alerts over time — trend compared to the previous period
- By detection type — which hazard types dominate
- By location — zone-level hotspots
- By subcontractor — which contractors appear most often on alerts
- Day and hour heatmap — when alerts tend to occur during the week
Count vs Per 1k man-hours
Risk analytics can display metrics in two modes, toggled at the top of the page:
| Mode | What it shows |
|---|---|
| Count | Raw number of alerts |
| Per 1k man-hours | Alerts normalised by observed worker exposure |
See Alerts per 1,000 man-hours below for why this matters and how to interpret it.
Response analytics
Response analytics measures how efficiently your team works alerts.
Key metrics
| Metric | Meaning |
|---|---|
| Time to effective action | How long from detection until the corrective action that was later verified — the end-to-end measure of closing the safety loop |
| Time to assign | How long until someone takes ownership |
| Time to accept | How long until the assignee accepts |
| Time to action | How long from acceptance to first action or fix request |
| Time to fix | How long engineers take to submit corrective evidence |
| Expired rate | Share of alerts that timed out before resolution |
| Rejection rate | Share of closures or fixes sent back for rework |
| Escalation rate | Share of alerts escalated to higher roles |
Time to effective action
Time to effective action is the primary response KPI. It answers: once we spotted the issue, how long until we took the corrective action that stuck?
| Aspect | Detail |
|---|---|
| Start | Alert detection (creation) |
| End | The Actioned step that proceeds directly to Verified |
| Included | Assignment, acceptance, any rejected attempts, and the successful corrective action |
| Excluded | Time waiting for verification after that successful action; alerts that never reach Verified |
| Aggregation | Usually shown as a median across alerts in the selected period |
If an alert is actioned, rejected for rework, then actioned again and verified, the metric uses the second (successful) action — not the rejected attempt.
Do not confuse this with Time to action, which only covers one stage of the workflow (acceptance → first action or fix request). Time to effective action is the fuller measure of response performance.
Charts
- Trend charts — how a selected metric changes over time
- Jitter plots — spread of individual alert timings (not just averages)
- Bottleneck analysis — time spent in each workflow status
- Risk vs speed quadrant — whether high-risk alerts are being closed quickly (median time to effective action by risk level)
- Expired and rejected alert tables — specific records driving poor metrics
Use response analytics to find process delays, not just to count alerts.
Consequence analytics
Consequence analytics connects safety alerts to enforcement and coaching outcomes.
Outcome types
| Outcome | What it tracks |
|---|---|
| Toolbox talks | Coaching sessions issued after safety issues |
| Site removals | Workers removed from site for breaches |
| Fines | Financial penalties and their total value |
How to use it
- Compare periods to see whether consequences are rising or falling
- Use matrix charts to see which detection types and alert levels most often lead to fines, removals, or toolbox talks
- Review subcontractor and assignee impact cards to see who is most affected
Consequence analytics answers "what happened after the alert?" — complementing risk (how much) and response (how fast).
Alerts per 1,000 man-hours
The problem with raw counts
A location with twice as many workers will often produce twice as many alerts — even if safety performance per worker is the same. Comparing raw alert counts between zones, weeks, or subcontractors can therefore be misleading.
A quiet area with 2 alerts and a busy area with 20 alerts does not automatically mean the busy area is ten times less safe. It may simply have far more observed work activity.
What man-hours mean
Man-hours represent estimated worker exposure time on site — how much work activity Quantavist observed in a given location and period.
Quantavist estimates this from camera coverage: worker presence is observed over time and converted into exposure hours. For example, if an average of 20 workers are observed during one hour of coverage, that contributes roughly 20 man-hours to the denominator.
Man-hours are an exposure estimate, not a payroll or timesheet figure. They reflect observed activity in camera-covered areas during the selected period.
The rate
Alerts per 1,000 man-hours expresses alert volume relative to that exposure:
(alerts ÷ man-hours) × 1,000| Example | Calculation | Rate |
|---|---|---|
| 5 alerts, 250 man-hours | (5 ÷ 250) × 1,000 | 20 per 1k mh |
| 10 alerts, 2,500 man-hours | (10 ÷ 2,500) × 1,000 | 4 per 1k mh |
In this example the second location had more alerts in absolute terms, but a lower rate — suggesting less risk per unit of work activity.
Why it matters
Exposure-adjusted rates let you:
- Compare locations fairly — identify hotspots that are genuinely elevated, not just busy
- Track improvement over time — see whether risk per unit of work is falling, even as site activity grows
- Prioritise interventions — focus HSE effort where rate is high, not just where alert count is high
- Benchmark subcontractors — see relative exposure-adjusted performance (see caveats below)
This is standard practice in safety benchmarking: normalise observations against exposure so comparisons reflect risk intensity, not site size or headcount alone.
When to use each mode
| Use Count when… | Use Per 1k man-hours when… |
|---|---|
| You need total workload or backlog | You want to compare risk intensity across areas or periods |
| Reporting absolute alert volume to leadership | Identifying which locations are outliers relative to activity |
| Exposure data is unavailable for the range | Fair comparison between quiet and busy zones |
When rates are unavailable
You may see no rate, a dash (--), or a message that exposure is insufficient. This happens when:
- Camera coverage was too sparse in the selected period to estimate man-hours reliably
- The location or time window has too little observed activity for a stable rate
- Filters narrow the data to a slice with inadequate exposure
Quantavist deliberately avoids showing rates when the denominator is too small — a handful of alerts divided by minutes of observed activity can produce misleading spikes. No rate is more honest than a distorted one.
When rates are unavailable, use Count mode or widen the date range and location filter.
Subcontractor rates
Subcontractor rates use location man-hours as a proxy — observed activity in zones where that subcontractor's alerts occurred: (subcontractor alerts ÷ location man-hours) × 1,000. This is not payroll or timesheet hours. Use for relative ranking across peers, not as an exact workforce measure. Totals may not align precisely with site-wide figures.
Weekly reports
Weekly safety reports also use alerts per 1k man-hours for location rankings and alert-level breakdowns, using the same exposure logic as risk analytics.
Tips for effective use
- Start with a question — "Where is risk highest per unit of work?" (risk, rate mode) vs "Is our time to effective action slowing on critical alerts?" (response).
- Compare periods — a single week can be noisy; comparison highlights genuine trends.
- Drill down — cross-filter from a chart to location or detection type, then open View alerts to inspect individual records.
- Pair risk with response — a hotspot with slow time to effective action deserves different action than a hotspot that is already being closed quickly.
- Check exposure before acting on rates — if per 1k man-hours is unavailable, widen the range or confirm camera coverage in that area.
Related pages
- PPE compliance — PPE compliance rate (separate from alerts)
- Glossary — time to effective action, alert level, location, subcontractor, and other terms
- Alert status model — workflow statuses referenced in response analytics