What is that task actually costing you?

Name one job somebody repeats every week. See what it costs in a year, and whether it is worth automating or better left alone.

Only used to label your result. Tap one to fill the box, or write your own.

Time actually spent on it, not time it sits waiting.

Count everyone who does it regularly, including you.

In dollars: wage plus overhead. 35 is a starting point, not a benchmark.

An assumption, and yours to adjust. The real figure is only known once someone has looked at the task.

Not 48 working weeks a year? Change it

48 allows for four weeks of holidays and leave. Use 52 if the task gets done every week regardless of who is away.

Calculated in your browser. Nothing you type is sent to us or stored anywhere.

The calculator needs JavaScript to do the sum with your numbers. The arithmetic is four lines and all of it is written out below, with a worked example, so you can run it on paper — or send the task to hello@ataday.net and we’ll work it through with you.

Worked example

Worked example — round numbers, not a benchmark

Invoicing and payment follow-ups

About $20,000 a year. That is worth automating.

If an automation takes over 60% of it, $12,000 and 300 hours come back every year, not once. That is enough to pay for building it properly, testing it, and still come out ahead.

  • Worth automating $12,000 a year recoverable
  • Hours a year

    500

    Done by hand

  • Cost a year

    $20,000

    At $40 an hour

  • Hours back a year

    300

    At an assumed 60%

  • Dollars back a year

    $12,000

    At an assumed 60%

  1. Worth automating Cost

    Done by hand, it costs about $20,000 a year

    That is 500 hours a year of paid time going into one task. Nobody sees the bill, because it arrives as a few hours a week rather than as an invoice.

    5 h a week × 2 people × 50 weeks = 500 hours a year
    500 hours × $40 an hour = $20,000 a year
  2. Your assumption Recoverable

    At 60%, $12,000 and 300 hours a year come back

    The 60% is an assumption, and the whole result leans on it. Nobody can know the real share until they have watched the task being done: the exceptions, the judgement calls, the customer who always does it differently.

    500 hours × 60% = 300 hours a year
    $20,000 × 60% = $12,000 a year
  3. Worth automating Break-even

    A build costing up to $12,000 pays for itself within 12 months

    That figure is a ceiling, not a quote. Spend less than it and the automation has paid for itself inside a year; everything after that is saving. Running costs come off the top, so leave room for them.

    What we would doWatch the task done once from start to finish, then build one narrow, tested automation around it using the tools you already have, with a person checking the steps where a mistake would be expensive.

This is exactly what we fix

Twenty minutes on a call. Bring the task and we’ll tell you whether the share you assumed is realistic, what building it properly involves, and whether it still pays once the real numbers are in.

This is an example with invented round numbers. Yours will look different, and may well say “leave it alone”.

The arithmetic

Four lines, no secret sauce

There is no model behind this and no industry average. Every figure is one you typed, or multiplication on the ones you typed.

01 · HOURS

What the task takes in a year

hours a week × people × working weeks. Five hours a week sounds like nothing. Across two people and a working year it is 500 hours.

Where it goes wrong People underestimate. Count the checking, the chasing and the fixing afterwards, not just the typing.

02 · COST

What those hours cost you

hours a year × cost of an hour. The cost of an hour is what it costs the business: wage, payroll taxes, benefits, and a share of the desk and the software.

Where it goes wrong Using take-home pay. An hour of somebody’s time always costs the business more than the number on the payslip.

03 · RECOVERABLE

What an automation gives back

hours and cost × the share it takes over. That share is the one number here that is a guess. It starts at 60% so the box is not empty, and it is yours to move.

Where it goes wrong Assuming 100%. The exceptions and the judgement calls stay with a person, and in most businesses they should.

04 · BREAK-EVEN

What a fix can cost and still pay

The dollars recoverable in a year are also the most a build can cost and still pay for itself within twelve months. It is a ceiling to hold a quote against, not a price.

Where it goes wrong Forgetting that an automation has running costs and needs looking after. Leave room under the ceiling for both.

The verdict

How it decides, in the open

One number decides it: dollars recoverable per year. The two lines are our own rule of thumb, stated here so you can disagree with them.

  1. Worth automating $10,000 a year or more

    Enough to build it properly

    At this size the saving covers a real build: watching the task, testing against the awkward cases, documenting it, and keeping it running. This is where automation earns its keep.

  2. Borderline $2,000 to $10,000 a year

    Worth fixing, if the fix is small

    Real money, but only enough to justify something simple. Often the answer is a feature you already pay for and have never switched on.

  3. Leave it alone Under $2,000 a year

    Not worth automating

    Anything built has to be tested and looked after. Below this line that costs more attention than the task does, so the tool says no. We would say the same on a call.

This is exactly what we fix

The sum is the easy part. The task is the job.

A calculator can tell you a task is expensive. It cannot tell you why it takes five hours, which parts need a person, or what breaks when a customer does something unexpected. We start there: one named task, built around the tools you already have, tested against real edge cases before it touches real data, and handed over documented. Not a chatbot bolted onto your homepage.

Questions people ask

Where do the numbers come from?

From you. Every figure in the result is either something you typed or arithmetic on what you typed. The boxes start with placeholder values so the page isn’t empty; they are not benchmarks and they are not averages. Nothing you type is sent to us or stored.

What should I put for the cost of an hour?

What the hour costs the business, not what lands in the pay packet: wage, plus payroll taxes and benefits, plus a share of the desk, the software and the management time. If you do the task yourself, use what an hour of your time is worth to the business. That is usually the highest number in the building.

How much of a task can really be automated?

Nobody can say honestly without watching it being done. It depends on how many exceptions and judgement calls the task contains, and those never show up in the description. That is why the share is your assumption to move, not a figure we supply.

Why does it sometimes tell me not to automate?

Because sometimes that’s the answer. An automation has to be built, tested and looked after. When less than $2,000 a year is recoverable, that usually costs more attention than the task does. A calculator that always says yes is an advert.

Is this about putting a chatbot on my website?

No. It’s about one named, repetitive task inside the business: retyping, chasing, copying, scheduling. Most small businesses don’t need a chatbot. They need one manual task to stop eating hours, and that is the work we do.