What’s consuming most of your time at work?

Here’s The Workflow Automation Starter Plan: What to Automate First (And What to Leave Alone)

Have you heard the phrase “there’s an AI for it”? If you haven’t, that’s okay. All it really means is that whatever you’re toiling over on your own probably already has a solution out there, some AI platform or program built to handle exactly that.

Let’s be honest, AI can be genuinely great at saving time, cutting costs, and helping you brainstorm through complex problems faster than you’d get there alone. But…

blindly throwing everything at AI, especially the tasks that run on autopilot, will eventually backfire.

 So the question worth answering is this: what should you actually automate first, and what should you leave alone, at least for now? Let’s learn from our friend Marcus. He runs a bookkeeping side hustle out of his apartment in Edmonton, six small clients deep, squeezed in around a day job that still covers most of the bills while the business finds its feet.

Marcus caught the mistake at 11:40 on a Tuesday night. He’d entered the same inv

Let’s learn from someone who’s in your very shoes. His name is Marcus. He runs a bookkeeping side hustle out of his apartment in Edmonton, six clients deep, squeezed in around a day job that pays the rent while the business finds its feet. That night he’d entered the same invoice into two different places — his accounting software and the spreadsheet he used to track client status — and typed two different amounts. Off by forty dollars. Not a disaster. Just a crack.

He sat with it for a minute before fixing it. Forty dollars wasn’t the problem. The problem was that he was the person clients paid to catch exactly this kind of error, and he’d just made one, at midnight, because his process demanded more attention than a tired brain had left to give.

That’s usually the real trigger for automating a small business, and it’s rarely the one people expect. It’s not a productivity article or a tool recommendation from a podcast. It’s a near-miss.

A moment where the gap between how much you’re doing by hand and how much attention you have left becomes visible, just for a second, and it’s unsettling enough that you finally do something about it.

Marcus’s first idea was the obvious one: hire it out. A virtual assistant a few hours a week to handle the repetitive parts — chasing invoices, replying to routine emails, managing his calendar. He got as far as running the numbers. A reasonable VA rate, even part-time, would have consumed most of the revenue from two of his six clients. He didn’t have the margin for a person. He had margin, just barely, for a system.

This is the fork every solo operator hits eventually, and it’s worth naming directly: automate or delegate. They solve the same underlying problem — too many small tasks eating time that should go toward client work — but they cost different things. Delegating costs money you may not have yet and management time you’ll need to spend training and checking someone else’s work. Automating costs setup time upfront and a willingness to sit with imperfect systems while you tune them. For a business Marcus’s size, with no cash cushion built in yet, automation was the only fork that didn’t put him further behind.

So instead of trying to fix everything over one exhausting weekend, he gave himself three weeks. One process at a time.

Week #1: Email

Marcus didn’t automate his email relationship with clients. He automated the one percent of it that was purely mechanical: the first response to a new inquiry. Someone finds him through a referral or his website, sends a message asking about his services, and used to wait anywhere from a few hours to a full day for a reply, because Marcus was busy doing actual bookkeeping for actual clients.

Now that first reply goes out within minutes. It confirms he received the message, explains what happens next, and sets expectations for when he’ll personally follow up. It’s not trying to sound like Marcus having a conversation. It’s doing the job a receptionist would do if he had one: acknowledge, orient, buy time.

The distinction matters. Marcus wasn’t automating judgment. He wasn’t automating the decision about whether to take on a new client, or the tone of an email to someone who’s frustrated about a mistake in their books. He automated the part where nothing was being decided at all — just acknowledged.

Week #2: Invoicing

This was the fix Marcus felt the most stupid about, in his own words, because the tools were already sitting inside software he was already paying for. His accounting platform had recurring invoicing and automatic payment reminders built in. He’d just never turned them on. Turning things on felt like a project, and projects got pushed to some slower month that never actually arrived.

It took him forty minutes on a Sunday afternoon. He set his recurring clients up on automatic monthly invoices, and set reminders to fire at three, seven, and fourteen days past due, before he’d have to personally follow up.

The effect was almost immediate. Clients who used to pay two or three weeks late, not out of malice but because a manual invoice email got buried in their own inbox, started paying closer to on time, because the system was more consistent than Marcus’s memory ever could be. He wasn’t chasing money anymore. He was reviewing a dashboard.

Week #3 Scheduling

The old process for booking a new client conversation was seven emails deep before anyone agreed on a time. Marcus proposes a slot, the client counters, Marcus checks his calendar, proposes again, and so on, for days, while the actual work waited.

He replaced it with a booking link. Client picks a time that already matches Marcus’s actual availability. No back-and-forth. No mental math about time zones or which evenings he’d already blocked for client work. The calendar filled itself in, and more importantly, it stopped being a task Marcus had to think about at all.

The Line Marcus Almost Crossed

Here’s where the story gets more interesting than a simple productivity win, because Marcus almost made a mistake in the other direction.

Riding the momentum of three fast wins, he found a tool that could generate an automatic monthly summary for each client, pulling numbers from their books and sending a templated recap by email, no manual work required. It looked like the natural fourth step. He tried it on one client.

Those monthly summaries were supposed to replace the calls Marcus has with every client, walking through their numbers together, one on one. He noticed the problem almost immediately. On a call, when a client mentions their revenue dipped and their voice tightens slightly on the word “dipped,” Marcus asks a follow-up question. He catches a slow month before it becomes a slow quarter. He hears the thing they didn’t say directly. A templated summary can’t hear anything. It can only report.

He turned it off after one cycle and went back to the calls.

This is the part of automating a small business that doesn’t get talked about enough:

knowing what to leave alone is just as valuable as knowing what to fix. The tasks worth automating are the ones with no judgment in them, no relationship attached, and no cost to getting it slightly wrong. The tasks worth protecting are the ones where a human noticing something is the entire point.

A Simple Way to Sort Your Own List

If you’re standing where Marcus was, staring at a list of everything that eats your week, here’s the filter that would have saved him some trial and error.

Ask three questions about each task.

• Does it require a judgment call specific to this client or this situation?
• Does a relationship depend on a human being the one who does it?
• Is the cost of getting it wrong high, or is it a minor, easily fixed error?

If you answered yes to any of those, that task probably needs to stay human, at least for now. If you answered no to all three, it’s very likely safe to automate, and you’ll probably wonder why you waited.

Client-facing conversations where trust is being built: stays human. 
First-response acknowledgments with no decision attached: automate. 
Anything involving reading between the lines of what a client actually means: stays human. 
Recurring paperwork with a fixed, predictable format: automate.

What Marcus Didn’t Do

He didn’t buy a bundle of five different automation tools. He didn’t overhaul his entire workflow in one sitting. He didn’t hire the VA he couldn’t yet afford. He picked one task a week, for three weeks, starting with the one that was purely mechanical and had the least room for anything to go wrong if the automation wasn’t perfect on day one.

By the end of the third week, he’d freed up roughly four hours, which doesn’t sound dramatic until you remember that four hours is nearly a full extra workday inside a business that runs on evenings and weekends squeezed around a day job. More than the time, though, he’d gotten his margin for error back. The kind of mistake that happens at 11:40 pm because a tired person is doing a task a machine could have done more reliably — that mistake got a lot less likely.

Where to Start This Week

You don’t need Marcus’s exact three tasks. You need his method: pick one purely mechanical task, automate it, watch what changes, then pick the next one. Start with whichever task on your list has the least judgment attached to it. For most solo operators, that’s either the first-response email, the invoice reminder, or the scheduling back-and-forth — the same three Marcus started with, because they tend to be the most universal time drains in a small service business.

Give yourself a week per task instead of a weekend for all of them. The goal isn’t a perfect system by Friday. It’s one less thing standing between you and the work only you can actually do.

Is this useful to you every WEEK?

make sure you’re subscribed to Digital Equity Notes so you don’t miss the next one. And if you’re a founder or entrepreneur who wants to stay ahead of what’s coming, Signal is built for you: https://www.wilglobo.com/subscribe

Have a task in your business you keep doing by hand that’s probably safe to automate? That’s your starting point, let’s chat.

---