QHSE Software ROI Calculator

Record where every input comes from

Note the source, date and owner of each value. Use zero only when a cost or benefit is deliberately excluded; a blank or unknown amount is not a measured zero, and the calculator's optional zero values must be reviewed.

• Measured task volume, time per task and loaded hourly rate.
• A time-saving assumption supported by a representative pilot.
• The share that actually reduces spending, such as overtime or external services.
• Recurring software and support costs, one-time implementation and rollout delay.
• Other savings only where separately supported and not already counted in task savings.
• If the current process runs on spreadsheets, the hidden work as well: updating and consolidating files, chasing missing entries, re-formatting the same data for management reports and recovering conflicting or lost versions. Measure it; do not use a generic estimate.

How the model calculates the result

Annual labor cash benefit is released hours multiplied by the loaded hourly rate and the cash-realization share. Capacity released for other work remains useful but is not automatically a cash benefit. First-year benefits begin after the selected rollout delay; recurring costs begin immediately and implementation is counted once at the start.

Compare like periods and currencies. The model assumes constant costs and benefits after rollout and excludes discounting, tax, inflation, financing and changing adoption. It does not model incident reduction or avoided fines.

What the code computes, step by step

Every field must be filled in and every number must be zero or more; the two percentages cannot exceed 100 and the rollout delay cannot exceed 12 months, otherwise no result is shown. The steps are: hours released = annual hours × time reduction; annual cash benefit = hours released × hourly cost × cash realization + other savings; annual cost = subscription + other operating costs. Year 1 benefit is the annual benefit multiplied by (12 − rollout months) ÷ 12, and other savings are delayed with it. Year 1 cost is the annual cost plus implementation. The three-year net benefit is the Year 1 benefit plus two full years of benefit, minus three years of annual cost and one implementation cost.

ROI is the net benefit divided by the cost, as a percentage, and is left undefined when the cost is zero. Payback is calculated only when the annual benefit exceeds the annual cost. It then equals the rollout months plus (implementation + annual cost × rollout months ÷ 12) ÷ the monthly net benefit, measured from project start. Otherwise the page says payback is not reached.

A worked example that does not pay back

These inputs are invented for teaching and are separate from the three examples further down. A team spends 1,800 hours a year on a workflow at a loaded rate of 35 per hour. The assumption is a 20% time reduction, with 40% of the released time turning into a real spending reduction. Other evidenced savings are 1,500 a year. The subscription is 6,000, other operating costs 1,200, implementation 3,000, and benefits begin after 4 months.

• Hours released: 1,800 × 20% = 360
• Annual cash benefit: 360 × 35 × 40% + 1,500 = 5,040 + 1,500 = 6,540
• Annual cost: 6,000 + 1,200 = 7,200
• Year 1: benefit 6,540 × 8 ÷ 12 = 4,360; cost 7,200 + 3,000 = 10,200; net -5,840; ROI -57.3%
• Three years: benefit 4,360 + 2 × 6,540 = 17,440; cost 3 × 7,200 + 3,000 = 24,600; net -7,160; ROI -29.1%
• Payback: the annual benefit is below the annual cost, so it is not reached

The calculation function returns these values for these inputs. The result is sensitive to the cash-realization share alone. At 50% the annual benefit is 7,800 and payback arrives after 112 months, and at 60% it is 9,060 and payback arrives after about 38.8 months with a three-year net of -440. A model that flips from not reached to over three years on a ten-point change in one unmeasured assumption is telling you which assumption to measure first.

Why the result can mislead

Benefits switch on in full on the day the delay ends, whereas real adoption usually ramps. Costs stay flat for three years, so a renewal increase is not seen. Nothing is discounted, so money received in year three counts the same as money today. Year 1 ROI is almost always poor because implementation sits entirely in that year, and it should not be read as the long-run return. The hourly rate matters too: a loaded rate includes benefits and overheads, but the cash that actually falls is often an overtime premium or a contractor fee, which can be higher or lower than the loaded rate. Other savings are not scaled by cash realization, so entering a saving there that is already inside the labour line counts it twice.

Two public sources explain the inputs the model leans on. The US Bureau of Labor Statistics' Employer Costs for Employee Compensation release measures wages and salaries plus benefit costs per hour worked, which is the sense of a loaded hourly cost. HM Treasury's supplementary Green Book guidance on optimism bias says appraisers tend to be over-optimistic and recommends explicit adjustment of cost, benefit and duration estimates; it is written for UK public-sector appraisal and gives no figure for software. Both pages were opened on 3 October 2026.

If finance asks about incident costs and insurance

The calculator leaves incident reduction, avoided fines and premium changes out. We know of no published benchmark or controlled study that separates the effect of EHS software from the safety programme around it. If your business case discusses these costs anyway, keep them outside the model, source each figure and label anything that is not your own measured data as a working assumption, with a pessimistic case beside it.

• Cost per injury: OSHA's $afety Pays tool uses injury cost data compiled by the National Council on Compensation Insurance (NCCI) to estimate the direct and indirect costs of 40 injury and illness types, and the extra sales needed to cover them. Run your own injury mix through it, or use your claims data, instead of applying one cost or one indirect-cost multiplier to every incident.
• Scale: the US Bureau of Labor Statistics reports a 2024 private-industry rate of 2.3 total recordable cases per 100 full-time equivalent workers (BLS news release). At that rate, 250 full-time workers would record about six cases a year. A change of one or two cases in a year is a small count, not evidence of a software effect.
• Insurance timing: where NCCI's plan applies, the experience modification is generally based on three years of payroll and loss data, and the policy period just before the rating date is left out because its losses are not yet valued and reported (NCCI, ABCs of Experience Rating). Do not place a premium reduction in the first year.

The examples use the same undiscounted model as the calculator: costs start immediately, implementation occurs once, and full constant benefits start after the delay. They exclude tax, inflation, financing, changing adoption, incident reduction and avoided fines. Internal effort is a resource cost; record its cash effect separately when preparing a cash budget. A positive model result does not establish operational readiness.

In your own business case, identify which overtime, external service or other budget line will actually fall. If the evidence supports capacity alone, retain that benefit without relabelling it as cash. Test a longer delay, lower time reduction and higher support cost as further scenarios.

Measure the pilot before claiming an ROI

Use these proposed measures to collect your own evidence. Keep workflow scope, site, team, sample size and observation period alongside the result. Differences in case complexity or reporting access can change the measure without establishing a software effect.

Keep a conservative scenario with no realized cash saving, your pilot-informed scenario and a delayed-rollout scenario. These are sensitivity checks, not probabilities or promised returns. More hazard reports may reflect easier reporting; do not convert the change into assumed incident reduction or avoided fines.