A risk model without a traceable, quantifiable foundation is not an analysis. It is an opinion with a histogram attached.
The Risk Data Engine is IQRM's proprietary methodology for replacing workshop-derived three-point estimates with calibrated probability distributions extracted from real project data. It is the foundation underneath every QSRA, QCRA, and Risk Radar that IQRM delivers.
Monte Carlo simulation is a mathematical engine. It does not know whether its inputs are evidence or opinion.
The histogram looks the same either way. The S-curve renders. The P80 number appears in the report. And for the last twenty years, the industry has accepted this as quantitative risk analysis.
It is not. It is qualitative inputs dressed in quantitative clothing. The moment the executive committee asks "how did you arrive at the optimistic, most likely, and pessimistic values?", the analysis collapses to its true foundation: a room full of people agreeing on numbers.
The Risk Data Engine exists to solve this at the only layer that matters: the input layer. Not better software. Not more iterations. Not smarter visualisation. Better inputs. From real data. Traceable. Defensible.
Five project data domains in. Defensible Monte Carlo forecast out.
The Risk Data Engine extracts, analyses, and calibrates data from your project records into probability distributions that feed the Monte Carlo model.
This is what it looks like when real project data becomes a calibrated Monte Carlo input.
Three worked examples. Raw data from project records on the left. Fitted probability distribution on the right. This is what the Risk Data Engine produces.
PO lead times to fitted Triangular
Measured rates to fitted PERT
SPI variance to fitted Lognormal
The distribution is not assumed. It is derived. That is the difference between an RDE-calibrated model and a workshop-based model.
Where defensible risk inputs actually come from.
Every Monte Carlo input in an IQRM model is traceable to one or more of these project data domains. Not to a workshop. Not to expert judgement. To records.
Productivity
Labour output rates from completed and in-progress activities. Measured vs planned. Activity-level data, not programme averages.
Procurement
Purchase order lead times: planned vs actual. Vendor delivery performance. Cost variances on procured items.
Progress
Earned value curves. Physical completion percentages. Progress against baseline by discipline, area, and system.
Performance
Schedule Performance Index (SPI) and Cost Performance Index (CPI) at activity and WBS levels. Trend analysis.
Resources
Equipment utilisation. Personnel availability and mobilisation lead times. Material supply constraints.
Five domains. Real records. Traceable distributions. That is the difference between an opinion and an analysis.
Standard practice vs. an RDE-calibrated model.
- Three-point estimates from a workshop
- "Expert judgement" with no traceable source
- Same distributions recycled across projects
- Inputs derived in 2 days, model runs for 6 months
- P80 that collapses under the first question
- Risk register as input to Monte Carlo
- Looks quantitative. Foundation is qualitative.
- Distributions calibrated from 5 project data domains
- Every input traceable to a specific data source
- Distributions fitted to this project's actual performance
- Inputs earn the same rigour as the simulation
- P80 that survives board-level challenge
- Project data as input to Monte Carlo
- Quantitative from input to output.
Five phases. Data to defensible forecast. 6-8 weeks.
Scope and data audit
Map the programme's data landscape. Identify usable data across the 5 domains. Define gaps. Agree deliverables.
Data extraction
Extract raw data from project systems: P6, cost reports, PO registers, EV reports, resource logs. Structure for analysis.
Statistical fitting
Analyse distributions. Fit probability distributions to each input based on data shape, not assumptions.
Model rebuild
Rebuild the QSRA/QCRA in Safran Risk using calibrated distributions. Run Monte Carlo. Produce P50/P80/P90.
Report and transfer
Board-ready report. Executive presentation. Model handover. Methodology documentation for your team.
RDE deployment on a $1.2bn brownfield expansion.
Two years of workshop-derived three-point estimates. Monthly risk updates with the same top 10 risks. P80 schedule contingency the executive team could not defend at the annual sanction review.
8-week engagement. Productivity rates extracted from daily reports across 12 disciplines. Procurement lead times from 340+ purchase orders. Progress curves from EV reports. Distributions recalibrated in Safran Risk.
Tighter, evidence-based forecast. Every P80 input traceable to source data. Re-baseline approved at the next executive review. RDE adopted as standard for two follow-on programmes.
The methodology underneath every IQRM service.
QSRA Consulting
Schedule risk analysis built on RDE-calibrated inputs. P50/P80/P90 completion forecasts with traceable distributions.
See QSRA consulting →QCRA Consulting
Cost risk analysis using compound distributions calibrated from real procurement and cost data via the RDE.
See QCRA consulting →Risk Radar Report
Monthly risk intelligence built on a live RDE-calibrated Monte Carlo model. Decision-ready, not register-ready.
See Risk Radar →Download the RDE methodology paper. Or discuss a deployment on your programme.
The methodology paper is a detailed technical overview: the 5 data domains, the distribution fitting process, and a worked example from a real programme. Written for risk practitioners and programme directors.
- The 5 data domains explained with source examples
- Distribution fitting methodology and selection criteria
- Worked example: procurement lead time calibration
- How the RDE connects to QSRA, QCRA, and Risk Radar
- When to commission an RDE deployment (4 triggers)
Download the methodology paper
Check your inbox.
The RDE methodology paper is on its way. If you want to discuss a deployment, reply to the email or email us directly.
Questions about the Risk Data Engine.
No. The RDE is a methodology, not software. It is a structured process for extracting, analysing, and calibrating data from project systems into probability distributions that feed Monte Carlo models in Safran Risk. The methodology is proprietary to IQRM.
Most projects have more usable data than they realise. Phase 1 includes a data audit that maps what is available across the 5 domains. Where gaps exist, IQRM uses analogous data from similar projects and documents the basis explicitly. Maximum traceability with whatever data exists.
Typically 6-8 weeks. Phase 1 (scope and data audit) is 1 week. Phases 2-3 (extraction and calibration) take 2-4 weeks. Phases 4-5 (model rebuild and reporting) take 2-3 weeks.
Yes. The QRM Professional Programme teaches the RDE methodology as Module 5. For consulting clients, IQRM offers methodology transfer as part of the deployment so your risk team can maintain and update the model independently.
Yes. Early-stage projects use analogous data from comparable completed projects. The RDE documents the analogous basis explicitly, so the executive team understands which inputs are from direct data and which are from comparable programmes. More defensible than unstated expert judgement.
Maximum 3 consulting engagements per month. Every engagement is led by Rami Salem directly. Current availability is communicated during the scoping call.
Your Monte Carlo model is only as defensible as its inputs. The Risk Data Engine makes the inputs defensible.
IQRM accepts 3 consulting engagements per month. If your programme's risk analysis needs an evidence base, not just a simulation, start the conversation.
Policy Pages
Get in touch
Copyright © 2026
