Automation ROI calculator

Every formula is visible

Monthly hours released = weekly hours affected × share released ÷ 100 × working weeks ÷ 12. Monthly theoretical capacity value = released hours × loaded hourly cost. The share released should describe the work the system can reliably remove, not the entire process just because one step is automated.

Monthly realised cash benefit = capacity value × cash realisation percentage ÷ 100. Monthly recurring cost = care + platform, AI and hosting + ongoing review. Monthly cash net = realised cash benefit − recurring cost. Positive cash payback in months = build cost ÷ positive monthly cash net.

The chart starts at negative build cost in month zero. Each later cash point adds the same monthly cash net. The separate capacity point adds capacity value less recurring costs. These are alternative views of the same released work. Adding them together would double count the benefit.

Values are compared using the decimals you enter, so equal benefits and costs produce zero net. Currency amounts are shown to cents. Smaller non-zero amounts use signed scientific notation so a small gain or loss does not disappear as zero. The table displays rounded amounts, while the graph and payback use the underlying calculated values. No positive cash payback is shown when monthly cash net is zero or negative. With a zero build cost there is no initial outlay to recover, but a negative monthly cash net still means an ongoing loss.

Set the cash percentage from a real change in spending

A team member spending less time retyping records may be able to serve more customers. Their salary may remain exactly the same. Set the cash percentage to zero for that released time unless an identifiable expense is actually avoided.

If part of the work is covered by paid overtime or a contractor, use the expense you can realistically remove. For example, $500 avoided out of $2,000 theoretical capacity value corresponds to 25%. The percentage is a convenient assumption, not evidence that the saving will occur.

The model does not include speculative sales revenue. If extra capacity could support new work, build a separate case using contribution margin, demand evidence and the additional delivery costs. Do not put total potential sales into an hourly labour saving field.

Include the cost of keeping the result useful

Use the actual proposed build fee and include other one-off setup or training spend in that input if relevant. Our published builds start at $3,500 AUD, typical projects are $3,500 to $12,000 AUD and complex AI agent systems are $12,000 to $25,000 AUD. The $8,000 default is only an illustrative assumption within that range.

Our care range is $750 to $2,500 AUD per month. Add incremental platform subscriptions, AI use and hosting separately. The default $150 software allowance is an illustrative assumption, not a vendor quote. Enter your expected costs from the relevant plans.

Ongoing review includes exception handling, checking drafted outputs and routine operational decisions. Cost that time consistently with your labour assumption. If you already reduced the share released to account for the same review time, do not subtract it a second time in the review field.

Try the case that makes you hesitate

Start with zero cash realisation and look at the monthly running cost. Then enter the specific expense you believe can be avoided. Reduce the automatable share or increase the review cost to see how fragile that assumption is.

This model assumes steady operation from the first month. It excludes a ramp-up period, financing, discounting, tax effects, price inflation, project delays and changes in demand. Align the inputs to the same tax basis. The displayed payback is simple recovery, not discounted cash flow or personalised financial advice.

Collect a baseline before the build and measure actual review effort during acceptance. Bring those records to a scope conversation. A useful investment decision can be to make the process smaller, improve the input or wait until there is enough recurring work.

Read the positive, break-even and negative scenarios together

Keep the illustrative time assumptions unchanged at twenty hours a week, half released, forty-eight working weeks and a loaded hourly cost of $50 AUD. They create forty monthly hours with a theoretical capacity value of $2,000. Keep the build at $8,000 and the combined monthly care, software and review cost at $1,000. Now change only the cash realisation assumption so you can see exactly what drives the decision.

At 75% cash realisation, the assumed avoided expense is $1,500 a month. Subtracting $1,000 of recurring costs leaves $500 monthly cash net. Simple recovery takes sixteen operating months. The chart only shows twelve months, so its final cash balance is still negative $2,000. A positive monthly net does not mean the build has already paid back within the displayed period.

At 50% realisation, the assumed cash benefit is $1,000 and the recurring cost is also $1,000. Monthly cash net is exactly zero. The business continues carrying the $8,000 initial build cost, and the cash line stays flat at negative $8,000. There is no positive cash payback to calculate, even though the system could still release useful time.

At the default 25% realisation, the cash benefit is $500 against $1,000 of recurring costs. Monthly cash net is negative $500. After twelve steady months, the cumulative cash balance including the build is negative $14,000. The separate capacity-value balance can be positive in the same scenario because it values all released time. That difference is the reason the two lines must remain separately named.

A zero build cost removes the initial outlay, but it does not change monthly economics. Positive cash net then represents an ongoing gain with no build amount to recover. Zero cash net produces neither an initial outlay nor a monthly gain. Negative cash net remains a recurring loss. Treat those as three different decisions rather than giving every zero-build scenario a successful payback badge.

Work out the review cost before choosing the automation share

There are two reasonable ways to account for ongoing human checks, but using both for the same work will understate the benefit. You can enter the gross work released by the system and cost its new review effort separately. Or you can reduce the hours released to a net figure after that review. Choose one convention and keep it consistent through the baseline and the proposal.

For a simple fictional example, suppose automation removes forty hours of repetitive work a month but adds five hours of reviewing held records. At an entered loaded cost of $50 an hour, those new checks have a capacity cost of $250. Under the gross approach, retain the forty released hours and put $250 into monthly review cost. Under a net-hours approach, count thirty-five released hours and leave that same review work out of the separate cost line.

The cash interpretation needs another decision. If an existing employee performs those checks during ordinary paid hours, that does not necessarily create a new cash payment. This calculator subtracts the entered review cost from both the cash and capacity comparisons, so it is a conservative cost allowance when you include salaried review there. Use an actual incremental review expense for a strictly cash-only estimate and assess the remaining employee capacity separately.

Do not mix a net-hours estimate with a full gross review charge simply because both look cautious. Write down which tasks remain with your team, how often they occur and what each one costs. If you cannot yet estimate the exception rate, use two scenarios and collect evidence during a pilot. The spread between them tells you how much that uncertainty matters.

Change one uncertain assumption at a time

Start from the positive scenario above with $500 monthly cash net and sixteen-month recovery. Increase the combined operating cost by $250 a month while leaving the benefit unchanged. Cash net falls to $250 and simple recovery becomes thirty-two months. A modest recurring change can matter more than a small discount on the build fee because it affects every operating month.

Next return costs to $1,000 and reduce the share of work released from 50% to 40%. Monthly capacity becomes $1,600. At 75% cash realisation, the cash benefit becomes $1,200 and cash net becomes $200. The same $8,000 build now needs forty steady months to recover. You have changed one assumption about the work, not the provider's fee, and the investment case has changed materially.

Use these comparisons to decide what to measure next. If a small change in review effort doubles payback, inspect the difficult records before commissioning a broad build. If the decision depends on eliminating a contractor cost, establish whether that expense can actually stop. A sensitivity exercise is useful when it directs discovery toward the assumption with the greatest effect on your decision.

Very small positive balances can produce very long payback estimates. The model preserves fractional gains and losses rather than labelling them zero just because the screen shows currency. If an estimate exceeds the numeric range the calculator can display, it says so instead of presenting infinity as a result. If the assumptions produce an amount too small to distinguish from zero at browser display precision, the results pause and explain the limit. These limits are a signal to revisit the assumptions and economic scale, not a useful recovery period for a business decision.