For CTO · CFO

Cloud Replatform ROI.

"It'll pay for itself" is an opinion. This is the arithmetic. Six inputs — current run-rate, target run-rate, migration project cost, productivity uplift, risk premium, hurdle rate — produce payback period and 3-year NPV. Adjust assumptions; verdict moves live.

Your inputs

All dollar values in $K. Use loaded costs (cloud bill + ops people + tooling + on-call burden).

Current annual run-rate ($K/yr)

today's loaded cost of the platform you're considering replacing

Target annual run-rate after migration ($K/yr)

honest steady-state, after 6-month settling

Migration project cost ($K)

one-time: vendor + internal labor + parallel-run + risk-buffer

Annual productivity uplift ($K/yr)

eng hours saved, faster releases, reduced incidents — money equivalent. Be honest.

Project risk premium (0–100%)

% of project cost you'll mentally add to cover slippage / scope creep / refactor

Discount rate / cost of capital (%)

your hurdle rate for NPV. 10% is a common SaaS default.

Verdict

Net positive

The math, openly

Annual savings = (currentRunRate − targetRunRate) + productivityUplift

Risk-adjusted project cost = projectCost × (1 + riskPremium)

Payback months = (riskAdjustedProjectCost / annualSavings) × 12 — capped to "never" if savings ≤ 0.

3-year NPV = Σ(annualSavings / (1 + discountRate)^t) for t=1..3 − riskAdjustedProjectCost

Verdict: NPV positive AND payback ≤ 24 months → go. NPV negative OR payback > 36 months → wait. In between → genuine toss-up.

This does not model salvage value of legacy capex, vendor-lock change-of-mind costs, or the cost of NOT migrating (technical-debt interest). If those are material, run the calc twice — once with optimistic inputs, once with pessimistic — and look at the spread.