ROI calculations for supply chain projects have a credibility problem, and it’s largely self-inflicted. Numbers get built from vendor benchmarks, benefits get counted twice, costs get understated, and the resulting figure is high enough that experienced finance people stop reading. Here’s a structure that survives scrutiny.

Establish The Baseline First

Measure, over twelve months where possible: average inventory value by category, expedited freight spend, premium and spot purchase spend, overtime attributable to scheduling problems, obsolete stock written off, documented stockouts and their cost, and planner hours spent on manual data work.
Twelve months matters because it captures seasonality. A baseline taken during a quiet quarter will make the benefits look larger than they are, and that gets discovered later at some cost to your credibility.

Model Benefits Conservatively and Separately

Inventory reduction. Apply a defensible percentage to your current holding, then value the release at your cost of capital — not at the full inventory value, which is the most common overstatement in these models. Also net off any inventory increase you’d deliberately make to improve service.
Expedite reduction. Take your current expedited freight and premium purchase spend and estimate what share was caused by poor visibility rather than genuine external events. Being honest here typically lands well below half.
Obsolescence reduction. Better forecasting and stock policy reduces write-offs. Your historical write-off figure is the ceiling; a portion of it is design-driven and won’t move.
Productivity. Hours recovered, valued at loaded cost, discounted heavily unless headcount actually changes. State it as growth capacity if that’s the truth.
Service improvement. Leave it unquantified. Note it as upside and let the reviewer decide what it’s worth.

Count Every Cost

Licensing over the full evaluation period, implementation and integration, data cleansing, internal time at loaded cost, training, ongoing support, and a transition productivity dip. That last one is real and omitting it is the fastest way to look naive.
Internal time is the line most often left out and frequently the largest single cost. Your best planners will spend months partly on this project. That has a value, and it appears in your case whether or not you write it down.

Present It as a Range

Single-point ROI figures invite disbelief. Give conservative, expected, and optimistic scenarios with the assumptions that separate them stated explicitly. A case showing a defensible return even in the conservative case is far more persuasive than one showing a spectacular return in a scenario nobody believes.
Show payback period alongside return. Finance thinks in payback for operational investments, and eighteen months of payback is a more useful sentence than a three-year percentage.

Commit to Measuring Afterwards

Set review points at six and twelve months against the agreed baseline, and publish the results whatever they say. This is uncomfortable and it’s the single best thing you can do for the credibility of your next business case.
Expect the benefits to arrive unevenly. Expedite reduction usually shows within a quarter because it’s a behavioural change. Inventory reduction lags by two or three quarters, because working stock down takes time and nobody sensibly does it in one step. A six-month review that only looks at inventory will understate what’s happening.
ticktick.ai tracks the baseline metrics from implementation onward, so the post-project review compares like with like.