Work out the ROI of quality inspection with numbers you already have

Executives always ask the same question before approving budget for AI quality inspection: "Is it worth it, and when?" The answer does not have to come from a vendor's numbers at all — the figures already sitting in your accounting and QC systems are enough to work it out yourself.

The basic formula, which works for almost any plant

Payback period comes from the scrap cost you avoid per month, against the cost of installing the system:

Payback period (months) = installation cost ÷ monthly value of scrap avoided

Monthly value of scrap avoided =
  (old defect rate − defect rate after installation) × units per month × cost per unit

A worked example (illustrative numbers, to demonstrate the formula only)

Say a plant makes 50,000 units a month at ฿40 per unit, with a defect rate of 3% falling to 1% after the inspection system goes in:

Monthly value of scrap avoided = (3% − 1%) × 50,000 × ฿40
                               = 0.02 × 50,000 × 40
                               = ฿40,000 per month

If installation costs ฿480,000, payback is 480,000 ÷ 40,000 = 12 months.

What to gather before you calculate for real

1. Your current defect rate — pull 3–6 months from the QC system, not a single month, so monthly swings do not distort it.
2. The real cost per unit — materials plus labour plus the cost of the line time lost, not just the material price.
3. The real monthly volume — use the average, not peak season, or the number will look better than it is.

Costs that get forgotten

Beyond installation (hardware plus labour), there are costs that regularly get left out of an ROI calculation:

• The time the team spends learning the new system in the first weeks
• Long-term maintenance of the lights and cameras — not a one-off cost
• The hidden cost of reshaping the workflow on the floor around the camera position

For more on why lighting matters in an inspection system, see why camera inspection systems fail — it is the hidden cost that most often makes an ROI estimate wrong.