Resources›Guides›Choosing schedule risk analysis software: the questions to ask

Choosing schedule risk analysis software: the questions to ask

By the Epoch SRA team·Published 5 August 2026·Corrections: contact@epochsra.com

A buyer's guide from a vendor with an obvious interest, written to be useful anyway. August 2026 – corrections to contact@epochsra.com.

Every schedule risk tool runs a Monte Carlo simulation; none of them differ meaningfully at that part. The differences that decide whether the tool works for you live elsewhere – in where your data goes, what happens to your schedule on the way in, and whose numbers you are actually looking at. These are the questions worth asking any vendor, including us, and why each one matters. We answer each as a worked example of what an answer looks like; for any other tool, ask them.

1. Where does my schedule go?

A program schedule is one of the most sensitive documents a company holds: it encodes suppliers, capacity, and how late you really are. If any part of the analysis runs on a server – a cloud component, a licensing check that phones home with usage, an AI assistant that reads the plan – your schedule is leaving the machine, and for export-controlled or competitively sensitive programs that can end the evaluation before it starts. Ask specifically: what is transmitted, where, and can it be switched off?

Our worked example: Epoch SRA computes entirely locally; schedule data never leaves the machine. There is no server, no account, no telemetry, and license keys are verified offline. For any other vendor: ask them, and get it in writing.

2. What happens to constructs the import can't translate?

No importer translates every scheduling construct – summary links, date constraints, exotic relationship types, manually scheduled tasks. The question is what the tool does when it hits one. Silent approximation is the dangerous answer: the simulation runs, produces confident percentiles, and nobody knows the network it simulated is not the network you planned. Ask: show me the import report for a real schedule of ours, and show me where it lists what you could not translate.

Our worked example: Epoch SRA's import reports every untranslated construct by name, and the quality checks run before any percentile is shown. For any other tool: ask them what their importer silently changes.

3. Which percentiles are calibrated, and which are estimated?

A P80 is only a commitment level if something ties it to reality. Most tools compute percentiles from whatever distributions the user typed in; the output inherits the input's guesswork, with no way to tell evidence from estimate. Ask: what are the default distributions based on, has anyone checked them against completed projects, and can I see which numbers in the output carry that evidence?

Our worked example: Epoch SRA's impact priors are field-calibrated per category against replayed historical space-program actuals, and every percentile is labelled with its provenance – field-calibrated or model estimate. For any other vendor: ask what their defaults are calibrated against, and how you would know.

4. Who is going to operate it?

Tools in this category assume very different operators: a trained risk analyst with allocated hours, a scheduling department, or the program manager between two reviews. A tool built for an analyst is not wrong – but if the honest answer is "the PM, occasionally," a tool that needs a training course will be run once and abandoned. Ask: how long from a fresh install to a defensible P80 on our own schedule, operated by the person who will actually do it?

Our worked example: Epoch SRA assumes the PM in the chair – import, map risks from your own register, Compute, read the results sheet – in an afternoon. For enterprise suites: ask them what the onboarding plan and training budget look like.

5. What happens when the plan changes?

A risk analysis you run once is a report; the value is in re-running it every time the schedule moves. That makes the update loop the real workflow: does re-importing a revised plan preserve your risk mappings, or do you rebuild them? Does the tool show what moved since the last run, or do you diff PDFs? Ask to see a second import, not just a first one.

Our worked example: Epoch SRA re-imports keep task identities stable by the file's own UIDs so risk mappings survive, and every run reports what moved since the last one, attributed to its cause. For any other tool: ask them to demonstrate the second import.

6. What does a seat actually cost per year?

Quote-based pricing is not a red flag by itself – enterprise suites have enterprise costs – but it makes comparison impossible until you are deep in a sales cycle. Ask for the fully loaded number: license, mandatory maintenance or subscription, any per-module pricing, and what happens to your files if you stop paying. Then ask whether you can try it before any of those conversations.

Our worked example: Epoch SRA costs €119/month, €990/year, or €1,300 perpetual + €350/year maintenance, published on the pricing page, and the 7-day full-featured trial requires no signup, no card, and no sales call. For any quote-based vendor: ask for the three-year total in writing.

The one-afternoon test

Whatever shortlist you end with, the cheapest evaluation is the same: take your real schedule, put it through each tool's import, and read what comes back before you look at a single percentile. The import report and the schedule quality checks will tell you more about both the tool and your plan than any demo. A vendor who will not let you do that without a sales call has answered question six already.