Every automation conversation starts the same way: how much? It is the wrong first question, and answering it honestly requires turning it around.
01 Start from the cost, not the quote
Manual process has a price. It does not appear on an invoice, which is why it runs for years unexamined. Your team is paid whether they are quoting or retyping a customer's address into a third system, so the cost hides inside wages you were already paying.
Until you put a number on that, no quote can look like value. Once you have it, most automation decisions answer themselves.
02 The sum that decides it
Take one ordinary week. For every task somebody repeats, write down the minutes it takes and how many times it happens. Do not estimate from memory, because memory undercounts the small ones badly, and the small ones are where the hours actually go.
- Hours per week across all repeatable tasks.
- × your loaded hourly rate, which is the wage plus on-costs, not the wage.
- × 52.
That is your annual cost of doing nothing. Across the businesses we work with, the recovered figure averages more than a thousand hours a year. At $50 an hour that is over $50,000 in capacity that goes back into the business.
Recovered hours are not the same as saved wages. You rarely reduce headcount; you stop paying good people to do work that does not need a person. The return shows up as more jobs quoted, faster follow-up and fewer things dropped, not as a smaller payroll. Judge the investment on that basis or you will be disappointed by it.
03 What a build actually involves
Cost tracks the number of systems involved and how well they talk. A single workflow between two platforms that both have a decent API is quick. Six workflows across a POS, an accounting package, a booking system and a payment terminal is a project.
The work itself divides into three parts, and the first is the one people skip:
- Mapping. Watching how the work is genuinely done, not how the manual says it is done. Automating a broken process gives you a faster broken process.
- Building. The workflows themselves, in Zapier, Make or n8n, or in code where the platform will not stretch.
- Handover. Documentation and training, so the business is not dependent on whoever built it.
04 Ongoing costs nobody mentions
Two of them, and they are easy to miss when comparing quotes.
The first is platform subscriptions. Zapier, Make and n8n all charge by task volume, so a busy workflow costs more to run than a quiet one. Our comparison of the three covers which one suits which job. Ask for the expected monthly run cost before signing anything, because a build that is cheap to make can be expensive to operate.
The second is maintenance. Platforms change their APIs. A workflow that has run for eighteen months will eventually break, usually silently. Either you hold that capability in-house or you keep someone on a support arrangement. Pretending it is zero is how businesses end up with automation nobody has checked since it was installed.
05 When not to automate
We tell people not to fairly often, and it is worth saying plainly here.
- The process is about to change. Automating something you are replacing in six months wastes the build.
- It runs a handful of times a year. Rare tasks rarely justify the maintenance burden.
- The judgement is the job. If a person is genuinely deciding something each time, automation moves the decision rather than removing it.
- The underlying process is broken. Fix it on paper first. Then automate the version that works.
06 Common questions
Do I need to replace my current software? Usually not. If a system has an API or webhooks, it can generally be connected as it stands. Replacing systems is a separate and much larger decision.
How long until it is running? A single workflow is days. A connected stack across several platforms is weeks. The mapping stage is what determines the rest, which is why we do it first and why the audit is free.
What if it breaks? It will, eventually, when a platform changes something upstream. What matters is whether anyone notices. Monitoring and alerting are part of the build, not an extra.
Automation costs less when the website is already collecting clean data. If yours is not, budget for the site first; our Albury Wodonga web design page has the price bands. Both services together are described on the Albury agency page, and the costs of AI assistants and agents specifically are on the AI automation agency Albury page.
The free audit is the mapping stage. We go through your workflows, put a number on what the manual version is costing, and tell you which parts are worth automating. If the numbers do not justify it, we will say so. See business automation or book a call.