New: Document Integrations with Aconex and Procore are live. See what your AI can read now

Platform Ground truth models Risk drivers

Schedule risk driver analysis

See exactly what’s driving your risk, and by how much.

Risk drivers ranks the activities and risk events behind any P-position, for any activity, straight from your QSRA results, so mitigation goes where it counts.

Book a demo
See the technical detail
DC Echo Risk drivers
Handover · P80
What drives the P80 of Handover of Data Hall 1
Activities
Risk events
Duration uncertainty, P.B.1010
44%
300-MATS Supply chain
27%
200-UTIL Utility relocation
25%
Duration uncertainty, E.G.2410
21%
100-PERM Permitting delays
12%
500-RAIN Weather events
9%
000-EARLY Early access
-5%
Top 20 shown. Remaining drivers grouped as one additional figure.

Why teams choose it

Three reasons it earns its place.

01

The real drivers, ranked.

The top 20 activities and risk events behind any P-position, split into duration uncertainty and risk events.

Top 20 drivers
02

In percentages or in days.

Normalized shares that add to 100%, converted to days using the P80 delay, or raw correlation when you need it.

Three views
03

Proof mitigation is working.

Compare drivers with the previous update or your mitigation scenario, and see which risks are actually coming down.

Before and after

Technical specification

How every driver is calculated.

Source
Monte Carlo results from your QSRA run
Method
Each iteration’s target finish delay is correlated with every activity’s duration change, with uncertainty and risk event contributions measured independently
Ranking
Normalized to 100%, top 20 shown, the rest aggregated
Units
Percentages, days (normalized share × P80 delay) or raw correlation coefficients
Target
Any activity, any P-position, or the whole project end
Views
Activities driving the risk, or risk events and the activities they impact
Comparisons
Previous update, or pre against post mitigation
Export
Tornado image, or Excel with the full activities and risk events breakdown

The math

Transparent from correlation to days.

Raw correlations are normalized so the drivers add to 100%, then converted to days using the target’s P80 delay, so the numbers are always explainable.

  • Normalized
  • Days
  • Raw correlation
From correlation to days
DRIVER A
DRIVER B
DRIVER C
Raw correlation
0.75
0.50
0.75
Normalized
37.5%
25%
37.5%
In days (P80 delay 200d)
75d
50d
75d

Mitigation

Show what each action really bought.

Turn on the mitigation comparison to see each driver before and after your mitigation scenario, and spot the risks that still need attention.

  • Compare with mitigation
  • Compare with previous update
Compare with mitigation
300-MATS Supply chain
27%
18%
200-UTIL Utility relocation
25%
22%
100-PERM Permitting delays
12%
6%

Two lenses

Activities or risk events, your choice.

Look at which activities drive the risk, or which risk events do and where they hit, then search for project-wide drivers in one click.

  • Activities view
  • Risk events view
  • Project End search
Activities view
Every activity that drives the P-position, with its bar split into duration uncertainty and risk events.
Risk events view
Every risk event that drives the P-position, broken down by the activities each one impacts.
Project-wide drivers
Search for Project End to see what drives the whole project, not just the last activity.

Put it to work

QSRA deep dive
Risk and forecasting use case
Industrial energy risk story

FAQ

Questions, answered.

Technical questions about risk drivers.

Talk to our team

From the Monte Carlo results. The target activity’s finish delay in each iteration is correlated with every other activity’s change in duration, with duration uncertainty and risk events measured separately.

The correlations are normalized so the drivers add up to 100%. The top 20 are shown, and the rest are grouped as one additional figure.

Each normalized percentage is multiplied by the target activity’s P80 delay. A 37.5% driver on a 200-day P80 delay is 75 days.

Yes. Switch off normalization to see each driver’s raw correlation coefficient as a percentage. Raw values don’t add up to 100%.