Theme

support@eliteworks.ai

A collaborative robot arm packed in an open wooden crate with a blank clipboard and a pencil on the lid.

Automation Payback Calculator

Will this automation pay for itself?

Enter the quote, your volumes and your costs. See the payback, what you stand to lose if it doesn't work, and how much a test before you buy is really worth. Your figures stay in your browser.

Private. Built on your numbers, not industry averages.

Enter the quote, your volumes and your costs. See the payback, what you stand to lose if it doesn't work, and how much a test before you buy is really worth. Your figures stay in your browser.

Private. Built on your numbers, not industry averages.

Check before you sign

Every quote shows what you gain if it works. Check what you lose if it doesn't.

Put in your own figures. Not sure of one? Give a low and a high. You'll see the payback, what the project is worth if it works and if it fails, and how much it's worth spending on a test before you commit.

This calculator needs JavaScript to run.

The formulas are set out below, so you can work them through by hand. Or send your figures to Arjun and he'll reply personally within one business day, Monday to Friday, India time.

arjun@eliteworks.ai

Enter your own figures. Not sure of one? Give a low and a high instead of a single hopeful number. Nothing is filled in for you, because a default would be our guess about your line, not yours.

Currency for your figures

Labels only. Nothing is converted. Enter every amount in the currency you pick.

You'll see a cautious case, a central case and a hopeful case.

01What it costs

Everything you pay before it runs: equipment, integration, installation, acceptance testing and your own team's project time.

Maintenance, spares, software licences, power, support contracts and the time of the people who look after it.

per year
02What it saves or earns each year if it works

Only hours that really stop, after taking off any re-checking or supervision the new system adds. Hours that just move to another task don't count.

hours a year

Wages plus benefits, overheads and shift allowances, for the people whose hours are saved.

per hour

Defects the system would catch inside your plant that today get further, to a later station or to the customer. Leave at zero if this isn't about quality.

defects a year

Scrap, rework and handling when the defect is caught before the part leaves.

per defect

Returns, replacements, warranty, penalties and the time spent handling the complaint. It should cost more than catching the defect inside the plant.

per defect

What you spend each year on safety stock, overtime or spare capacity that this system would make unnecessary. Not sure? Leave it at zero.

per year
03How your finance team judges it

The return your company expects on money it invests. Your finance team will have a figure. Use theirs, not ours.

per cent a year

How long the system will run before a major replacement. Don't use a longer life than your finance team would accept for this kind of equipment.

years
04If it doesn't work

What you could get back by selling, moving or returning the equipment if the project doesn't work. It can't be more than the capital cost.

Lost output, rush deliveries, rework and management time while a failed system is taken out or worked around.

Your honest estimate that the system hits its targets on your line. With no evidence yet, a figure near the middle is more honest than a high one.

per cent

Example, not a client's numbers

Example: a robot cell where a test before buying is worth up to 30 per cent of the price.

A plant is weighing up a collaborative robot cell that costs INR 60 lakh. (One lakh is 100,000 rupees, so that is 6 million rupees.) On the plant's own figures, the cell is worth plus INR 90 lakh over its life, in today's money, if it works as planned. If it doesn't work, the plant expects to get INR 20 lakh back by moving or returning the equipment, and to lose INR 5 lakh in disruption while it comes out. So a failure is worth minus INR 45 lakh. The plant head thinks there is a 60 per cent chance it works.

-50 0 +50 +100 INR lakh If it works +90 If it fails -45 Commit now 36 Know first 54 up to 18 Probability break-even 33 estimate 60 0 100
Worked example, not client data. Outcomes, expected values and the break-even point for a cell costing INR 60 lakh.
  1. Step 1. Two outcomes

    If the cell works it is worth plus INR 90 lakh. If it fails, minus INR 45 lakh.

    The failure figure is the capital cost of INR 60 lakh, less INR 20 lakh recovered by redeploying or returning equipment, plus INR 5 lakh of disruption while the cell comes out.

  2. Step 2. Commit now

    Committing now is worth INR 36 lakh, on average.

    At the plant head's 60 per cent: 0.6 x 90, plus 0.4 x minus 45, which is 54 minus 18.

  3. Step 3. Know first

    Knowing first makes it INR 54 lakh, so finding out is worth up to INR 18 lakh.

    If the plant could learn whether the cell works before committing, it would commit only when it will, and the failure branch drops to zero. The gap between 54 and 36 is the ceiling on what any test before committing can be worth.

  4. Step 4. Break-even

    The decision turns at 33 per cent. The estimate of 60 is 27 points clear of it.

    Below 45 divided by (90 plus 45), committing loses money on average. At 60 per cent the answer does not turn on small changes in confidence. At 40 per cent it would.

The sums, in INR lakh (1 lakh = 100,000 rupees)
StepCalculationResult
Expected value if you buy now0.6 x 90 + 0.4 x (minus 45) = 54 minus 1836
Expected value if you knew in advance, and bought only when it would work0.6 x 90 + 0.4 x 054
Most a test before buying can be worth54 minus 3618
Break-even chance of success45 divided by (90 + 45)33 per cent

So what does the plant head do with this? First, INR 18 lakh is the ceiling on what any test before buying is worth here. A test that costs less, and would genuinely show whether the cell will work, is worth a look. A test that costs more isn't. Second, 60 per cent is about 27 points above the break-even of 33 per cent, so a little more or less confidence won't flip the decision. At 40 per cent, it would.

Methodology

Every formula is here. Hand it to your finance team.

Every amount is in the currency you choose. Nothing is converted.

Annual net benefit if it works
Hours of manual work removed x full cost of one hour of work, plus defects caught earlier x (cost of a defect that reaches the customer minus cost of a defect caught inside the plant), plus extra stock or overtime you no longer need, minus annual running cost.
Simple payback
Capital cost divided by annual net benefit. Shown only when the annual net benefit is above zero.
Net present value if it works
Minus capital cost, plus annual net benefit x annuity factor. The annuity factor is (1 minus (1 + r) to the power of minus L) divided by r, where r is the discount rate and L the life in years. When r is zero, the factor is L.
Net present value if it fails
Minus (capital cost minus salvage value), minus disruption cost.
Expected value of committing now
p x net present value if it works, plus (1 minus p) x net present value if it fails, where p is your probability that it works.
Value of finding out first (called the expected value of perfect information in decision analysis)
p x the larger of (net present value if it works, zero), plus (1 minus p) x the larger of (net present value if it fails, zero), minus the larger of (expected value of committing now, zero).
Break-even probability
Minus net present value if it fails, divided by (net present value if it works minus net present value if it fails). Shown only when the project gains if it works and loses if it fails.
Fragility note
Shown when your probability is within ten points of the break-even probability.

Where you give a low and a high value, the cautious case takes the high end of every cost and the low end of every benefit, the hopeful case does the opposite, and the central case uses the midpoints.

The value of finding out first is a ceiling, not a price. No real test tells you everything, so a real test is worth less than this figure. The sum comes from standard decision analysis, first set out by Ronald Howard in 1966 and taught on most courses about deciding under uncertainty.

Your figures are kept in this page's web address, so you can bookmark or share the result. Anyone with that link can see your numbers. Nothing reaches us unless you press "Discuss your result with a founder" and then send the contact form yourself, after seeing exactly what it includes.

What the estimate assumes

  • The benefit is the same every year. Real benefits often build up over the first year, so counting full benefit from day one makes the result look better than it is.
  • Each year's net benefit arrives at the end of that year and is discounted at your rate.
  • Nothing is added for what the system is still worth at the end of a successful life.
  • A failure is treated as happening at the start: you lose the capital cost minus what you get back, plus the disruption cost. A system that fails after a few years of partial benefit lands somewhere between the two.
  • It either works or fails. Partial success isn't modelled, which is the main reason the result is a guide and not a forecast.
  • The saving from catching defects earlier is the gap between what a defect costs once it's outside the plant and inside it. Don't count the same defects again in hours removed.
  • Tax, depreciation, financing and inflation are left out. Your finance team's own model should include them before anyone decides.
  • The chance of success is your number. The calculator doesn't check it against any benchmark, because there isn't one behind it.

Formula version automation-payback.v1.

Talk to a founder

Bring your numbers and the vendor's quote.

Arjun will tell you which assumption he'd check first, and whether a test before the purchase order is worth what it costs.

A founder replies personally within one business day, Monday to Friday, India time.

General enquiries: support@eliteworks.ai, write using the message form

Visitor insights

Privacy choices

Read the privacy notice