Purpose
Turn a set of engineering and operations cost inputs into a defensible, multi-year ROI model for a digital twin deployment — quantifying savings from reduced physical prototyping, compressed iteration cycles, predictive maintenance, and lower defect rates, then stress-testing the result so finance and engineering can align on the investment decision.
When to Use
Invoke this skill when the user asks to:
- Build the business case or ROI model for a digital twin or simulation platform
- Justify digital engineering spend to a finance business partner or CFO
- Compare the cost of physical prototyping against a virtual/simulation approach
- Quantify predictive maintenance savings for a specific asset or fleet
- Decide whether a digital twin investment clears a required payback or hurdle rate
- Model the payback period and NPV of a manufacturing or product-development simulation initiative
Inputs Required
Before starting, collect:
- Asset type: What is being twinned — e.g., a physical product, a production line, a fleet of machines, a facility
- Physical prototyping costs: Cost per prototype/test cycle and number of cycles per year (materials, labor, test rigs)
- Iteration cycle time: Current time per design iteration and estimated number of iterations per program per year
- Downtime costs: Cost of unplanned downtime per hour and current annual unplanned downtime hours for the asset
- Defect / rework data: Current defect or scrap rate (%) and cost per defect or rework event, plus annual production volume
- Digital twin investment: Estimated upfront cost (software, integration, sensors, modeling) and annual recurring cost
- Financial parameters: Analysis horizon (default 5 years), discount rate (default 10%), and any required payback threshold
If cost-per-hour or defect cost data is missing, use industry-reasonable placeholders drawn from the asset type, label them clearly as assumptions, and expose them in the sensitivity analysis.
Instructions
Step 1: Establish the Baseline (Current-State Cost Model)
Before modeling savings, quantify what the current physical-only approach costs annually.
- Prototyping baseline: cost per cycle × cycles per year
- Iteration baseline: value of time lost to slow iteration — (iteration time × iterations/year × loaded engineering cost/hour) plus any opportunity cost of delayed launch
- Downtime baseline: unplanned downtime hours/year × cost per hour
- Defect baseline: production volume × defect rate × cost per defect
Sum these into a single Annual Baseline Cost figure. Every savings claim later must trace back to a line in this baseline — no savings from a category that has no baseline.
Step 2: Model the Four Value Streams
For each stream, apply a stated improvement percentage (with a conservative / expected / aggressive band) and calculate annual savings.
Value Stream 1 — Reduced Physical Prototyping Digital twins replace a share of physical builds with virtual validation.
- Conservative 20% / Expected 40% / Aggressive 60% reduction in physical cycles
- Savings = prototyping baseline × reduction %
- Note: rarely 100% — physical validation of final designs usually remains
Value Stream 2 — Faster Iteration Cycles Virtual iteration compresses the design loop and pulls launches forward.
- Model both cost savings (fewer engineering hours per iteration) and revenue acceleration (earlier launch × monthly gross margin, if a launch is tied to the program)
- Conservative 15% / Expected 30% / Aggressive 50% cycle-time reduction
- State launch-acceleration savings separately — it is a different confidence class than cost savings
Value Stream 3 — Predictive Maintenance Savings Twins monitoring live assets predict failures before unplanned downtime occurs.
- Conservative 15% / Expected 30% / Aggressive 45% reduction in unplanned downtime hours
- Savings = downtime baseline × reduction %
- Add maintenance-labor savings if planned interventions replace emergency repairs
- Only apply this stream when the asset is operational/monitored — skip it for a pure design-stage product twin and say so
Value Stream 4 — Defect & Quality Reduction Simulation catches design and process faults before they reach production.
- Conservative 10% / Expected 25% / Aggressive 40% reduction in defect rate
- Savings = defect baseline × reduction %
Sum the four streams into Total Annual Gross Savings at each confidence band.
Step 3: Build the Multi-Year Cash Flow
Construct a year-by-year table across the analysis horizon.
- Year 0: upfront investment (negative)
- Years 1–N: gross savings − annual recurring cost = net annual benefit
- Apply a ramp curve: benefits rarely hit 100% in Year 1 — use 50% in Year 1, 85% in Year 2, 100% thereafter unless the user specifies otherwise
- Discount each year's net benefit to present value using the discount rate
Step 4: Calculate the Headline Metrics
- Payback period: the point (in months) where cumulative net cash flow turns positive
- NPV: sum of discounted net benefits minus upfront investment
- ROI %: (cumulative net benefit over horizon ÷ total investment) × 100
- IRR: report if computable; otherwise note the discount rate at which NPV = 0 is above/below the hurdle Report all four at the Expected band, and NPV/payback at Conservative and Aggressive bands.
Step 5: Sensitivity Analysis
Identify the 3–4 assumptions the outcome is most sensitive to (typically downtime cost/hour, prototyping cost, and adoption ramp).
- Show how NPV and payback move under ±25% swings on each
- Flag any single assumption that alone flips the decision from "clears hurdle" to "fails hurdle"
- State the break-even value for the riskiest input (e.g., "downtime must cost at least $X/hour for this to pay back within the threshold")
Step 6: Verdict and Recommendation
State plainly whether the investment clears the required payback/hurdle at the Expected band, what has to be true for it to work, and the single biggest risk to the case. Be decisive.
Output Format
- Structure: Baseline table → four value-stream calculations → multi-year cash flow table → headline metrics → sensitivity table → verdict
- Length: 900–1,300 words including tables
- Format: Structured document with numeric tables (baseline, cash flow, sensitivity) and short interpretive prose between them
- Sections: Baseline Cost Model → Value Stream Savings (4) → Multi-Year Cash Flow → Headline Metrics (Payback / NPV / ROI / IRR) → Sensitivity Analysis → Verdict & Recommendation
- Tone: Finance-literate engineering director — quantitative, conservative by default, every number traceable to an input
Quality Checklist
Before delivering output, verify:
- [ ] Every savings figure traces to a line in the Step 1 baseline — no orphan savings
- [ ] All four value streams are shown at conservative / expected / aggressive bands (or a stream is explicitly skipped with a reason)
- [ ] The cash flow applies an adoption ramp — Year 1 is not modeled at 100% benefit
- [ ] Payback period, NPV, ROI %, and IRR (or its status) are all reported with the numbers behind them
- [ ] Every assumed input (any value not provided by the user) is labeled as an assumption and appears in the sensitivity analysis
- [ ] The verdict states clearly whether the case clears the hurdle and names the single biggest risk
Notes
- The most common overstatement is assuming 100% elimination of physical prototyping — final-design validation almost always remains, so cap the reduction and say why.
- Launch-acceleration revenue is the highest-upside, lowest-confidence stream; keep it separated from hard cost savings so finance can discount it independently.
- For pure design-stage product twins with no live asset, predictive maintenance savings do not apply — do not manufacture a downtime baseline to make the number look better.
- When inputs are thin, a defensible model with clearly flagged assumptions beats a precise-looking model built on hidden guesses.
How to use this skill
This file is a Claude Code skill in Anthropic's standard format — markdown with YAML frontmatter at the top.
Installing in Claude Code:
~/.claude/skills/<folder-name>/SKILL.md— auto-loads in every session anywhere on your machine (recommended for skills you'll use across multiple projects)<your-project>/.claude/skills/<folder-name>/SKILL.md— auto-loads only when working in that project (recommended for project-specific skills)
Restart Claude Code or open a new session to register.
Using in Claude.ai, ChatGPT, Gemini, or other AI tools:
- Drag this file into a new chat, or copy and paste its contents
- Send your task or request as your next message — the AI will apply the skill where it fits
For Claude Code skill details: https://docs.anthropic.com/en/docs/claude-code/skills