Most people ask for automation too early. A messy inbox, a spreadsheet that someone copies by hand, a form that never reaches the right person. All of that is annoying. Not all of it is worth a custom system.
The test I use is boring on purpose: does this task cost more than an hour a week, every week? If it does not, the build will cost more than it saves. I say that on the first call. If it does, we can usually put a number on the return before anyone writes code.
What “an hour a week” really means
It is not one dramatic hour. It is twelve minutes after every enquiry, plus the follow-up you forget, plus the tool that does not talk to the other tool. Added up over a month, that is a day of someone who should be selling, coaching, or shipping.
For a coach, that is often the path from “someone filled in a form” to “they have a time in the calendar.” For a shop or a small team, it is orders, invoices, or the same three emails sent in a slightly different wording.
What I actually build
Not a chatbot for the sake of a chatbot. The useful work is quieter:
- A website form that creates the right record, in the right tool, with the right next step
- An agent that drafts the first reply from facts you already have, not from a generic script
- A webhook between two systems you already pay for, so nobody copies and pastes
The website and the automation belong together. A fast site that still dumps every lead into a private inbox is only half the job.
When to wait
If the offer is still changing every fortnight, automate later. You will build the wrong path. If only you know how the work is done, write the steps down first. Automation does not replace a process. It repeats one.
I work from Vienna with businesses that want a direct builder, not an agency layer. If you can describe the hour you keep losing, that is enough to start.