Two minutes, two answers: does this problem even need machine learning — and if so, what would the ROI look like? Everything runs in your browser; nothing is stored or sent anywhere.
Step 1 — Should this be a model or a spreadsheet?
Answer four questions about the decision you want to automate. The verdict updates as you answer.
Step 2 — Estimate the ROI
Rough numbers are fine — the goal is order-of-magnitude honesty, not precision. Leave a field at 0 if it doesn't apply.
Hours of repetitive work the system would take over
Salary + overhead, typically 1.3–1.5× base pay
E.g. total nightly revenue if this is a pricing model
Be conservative: 1–5% is realistic for pricing/forecasting; double digits is a red flag
Scoped custom builds commonly run $20k–$50k; complex ones more
Cloud, monitoring, and maintenance (~15–30% of build cost per year)
Annual benefit
–
Labor savings + revenue lift, per year
Payback period
–
Time for net benefit to cover the build cost
3-year ROI
–
Net benefit over 3 years vs. total cost
These are planning estimates, not a quote or a guarantee. Real ROI depends on data quality, adoption, and measurement — which is why NexoLogic prices qualifying projects as a share of measured improvement rather than projections like these. See how our performance-based pricing works.
What to do with the result
A rough guide to reading the numbers:
Payback under 12 months: strong candidate — worth scoping seriously.
Payback 12–24 months: viable if the estimate is conservative and the system is core to your margin.
Payback over 24 months: usually a signal to buy an off-the-shelf tool or improve the process without ML first. Our build vs. buy guide covers when each answer is right.
Want a second opinion on your numbers?
Bring them to a free scoping call — if the honest answer is "don't build," we'll say so