The automation that builds a new client's folder did not fire. So I built the whole pack by hand in about 5 minutes.
The coach asked for the file 4 minutes into that. He had it 84 seconds later. The call went fine. And as far as I can tell, the only reason that call had its file is that a person was awake, watching the automation not run.
That is not a system yet. That is me covering for one. I bring it up because if you're trying to work out how to automate a coaching business, the first workflow you pick decides whether you end up with a system or with a person watching a system. Here's how I choose now.
TLDR
- Most coaches and consultants are 1 workflow away from AI paying off. The problem is picking which 1.
- The order of steps after someone buys matters more than which tool runs them.
- An automation nobody watches fail is worse than a task nobody automated.
- 4 questions pick the first workflow. If it passes all 4, build it this week.
- After the first one works, the second is easier for a reason that has nothing to do with the tech.
Most Coaches Are 1 Workflow Away From AI Paying Off
Here's what I see on almost every first call. The coach has tried AI. Usually for content, usually in a chat window, usually with results that were fine and then forgotten. And they've concluded that AI is a content thing that sort of works.
Meanwhile their business has 3 to 5 workflows eating most of their non-coaching time. Onboarding. Follow-up after a call. The nurture emails that go out to people who haven't bought yet. Content, yes, but content is usually fourth on the list by hours and first on the list by attention.
You can probably automate 3 of those 5. But you can only build 1 at a time, and the first one sets your expectations for all the others. Pick the wrong one and you'll spend 6 weeks on something fragile and conclude the whole idea is fragile.
The Order of Steps After Someone Buys Matters More Than the Tool
I got this wrong last month in a Slack thread, and the correction was better than my original position.
We were fixing the order of 2 steps after someone signs up. My position: the enrollment forms go first, because people forget to fill them in before the onboarding call. I'd already told a colleague to build it that way. Reasonable. Based on a failure I'd watched happen.
The coach's counter was about where the buyer actually is at the moment of purchase. People sign up during a live session. They should be able to book the onboarding call straight from the checkout page and get the forms after. The call is the number 1 priority, because it sets the urgency and the timeline for everything else. Get the call on the calendar while the buyer is still in the room.
I switched it back in the same message I was typing. He was right. And 5 minutes later I offered to add an automated chase email for anyone who hadn't filled in the form by the day of the call. The reply: "That all already exists."
2 lessons in 1 thread. The order of steps is a business decision, not a tooling decision, and mine came from watching what people forget while his came from watching where the buyer is. His should win. And I proposed a fix for a system I hadn't fully inventoried, which is the other way automations go wrong: you build the thing that already exists.
When the Automation Didn't Fire, a Person Was the System
Back to the folder that didn't build itself.
The setup is simple. A client submits their onboarding details, and a folder plus a worksheet plus a welcome packet get created from those details. That night the client hadn't submitted, so nothing fired. I flagged it 2 hours into my shift and gave the coach 2 options: run the call without the details while I build everything by hand, or ask the client to reschedule. Then I stopped waiting and built it. Folder at 10:04pm. Worksheet at 10:05. Welcome packet at 10:07.
Here's the thing. It worked because I was there. Until I know whether the form failed or the trigger failed, I'm treating that automation as optional and budgeting 5 minutes of manual work per new client. That's the honest state of it.
An automation that fails silently is worse than no automation, because you've stopped watching. So the first workflow you build needs 2 things a chat window never gave you: a way to know it ran, and a way to know it didn't. I've written about the wider version of this in what broke in a year of production. Silent failure is why gates exist.
Which Workflow Goes First: A 4-Question Test
For coaches and consultants choosing their first build, 4 questions. Score each workflow.
- Does it happen the same way every time? Onboarding does. "What should I post today" does not. Same-way-every-time is the whole precondition for automation.
- Can you tell when it failed? If a folder doesn't appear, you notice on the call. If a nurture email doesn't send, you might never notice. Pick the one where failure is visible, for the first build at least.
- Does it already exist somewhere? Inventory first. Ask the person who built your current setup what's already running. I didn't, and pitched something that existed.
- Will it save you time this month? Not in year 2. This month. The first build has to pay back fast or you won't build the second.
Onboarding passes all 4 for almost every coach. Follow-up after a call passes 3. Content passes 2, which is why I think people should not start there, even though it's where everyone starts. The content system is worth building, and it's what I do, but it's a better second build than first, because by then you know what a gate looks like.
What Happens After the First One Works
Something shifts, and it's not technical.
Once you've watched 1 workflow run without you for a month, you stop asking "can this be automated" and start asking "why am I still doing this by hand." That's a different person running the business. The second build takes half the time because you already know what the gate, the reference file, and the failure check look like. The third one you sketch on a call.
It's also where the content system finally makes sense, because you understand that a system is a standard plus a check, not a prompt. That's when the 3 Sources and the Encode-Verify-Ship method stop sounding like framework vocabulary and start sounding like the thing you built for onboarding, pointed at your recordings.
Look, you already know which workflow it is. You've known for a while. It's the one you did by hand yesterday and resented. The part you're stuck on is whether it's really same-way-every-time, and the 4 questions will tell you in 5 minutes.
If you want a second opinion on which one to pick, book a free call. 30 minutes, no pitch. And if the first build turns out to be content after all, send me 1 piece of your work and you'll have a finished asset back within 48 hours.
FAQs about automating a coaching business
Which workflow should a coach automate first?
The one that happens the same way every time, fails visibly, does not already exist in your setup, and saves time this month. For most coaches that is client onboarding: the folder, worksheet, and welcome pack built from a form submission. Content is a better second build, once you know what a gate looks like.
Should new clients fill in forms before or after booking the onboarding call?
Book the call first, straight from checkout, while the buyer is still in the room. The call sets urgency and the timeline for the pre-work. Forms come after, with an automated chase on the day of the call for anyone who has not completed them. Ordering by where the buyer is beats ordering by what people forget.
What makes an automation safe to rely on?
2 things a chat window never gave you: a way to know it ran, and a way to know it did not. An automation that fails silently is worse than a manual task, because you stop watching. Until you can see failures, treat the automation as optional and budget the manual fallback.
How long does the first automation take to build?
For a well-chosen first workflow, hours rather than weeks. The onboarding pack that failed in this post takes about 5 minutes by hand and is a small build to automate. The second build is faster than the first because you reuse the same gate and failure-check pattern.