The follow-up message is sitting in your drafts. Has been for three days. You wrote "Hey, just checking in!" and then stared at it until you felt vaguely embarrassed on your own behalf and closed the tab.
You know you need to send it. The client booked six weeks ago, came in once, and hasn't rescheduled. Or you finished a project and never heard back. Or someone said "let me think about it" and you genuinely do not know if following up makes you sound desperate or professional. So you do nothing. The message stays in the drafts. The client books somewhere else.
This is where people tell you to use AI follow-up message templates for your service business. And honestly, they're not wrong. But the way most people use them is the exact reason those messages sound like they came from a software company, not a person.
The Real Problem Isn't That You're Bad at Follow-Ups
You already know what you want to say. Roughly. You want to say: hey, I was thinking about you, how did everything go, I'd love to see you again, here's how to book. That's the whole message. It takes eight seconds to say it out loud to a friend.
The problem is that something happens between "what you'd say out loud" and "what you type into a text box." The formality creeps in. The hedging starts. You add "I hope this message finds you well" because you've seen it a thousand times and your brain defaults to it under pressure. Then you read it back and you don't sound like yourself anymore. You sound like a dental office.
So you delete it. Or you send it anyway and wince every time you open your phone for the next two hours.
That's not a writing problem. That's a translation problem. You need to get from "what you'd actually say" to "something you can actually send," and right now you don't have a system for that.
Why AI Follow-Up Templates Usually Make This Worse
Here's the thing about most AI follow-up message templates for service businesses: they're written for nobody in particular. They're written to be useful in any situation, which means they end up being perfectly wrong for your specific situation.
You've seen them. "Dear [Client Name], I wanted to follow up on our recent interaction and ensure your experience was satisfactory. Please don't hesitate to reach out if you have any questions or concerns." Cool. Love it. Very human. Very warm. Very much something a person who has never actually talked to another person would write.
Or the other kind: the ones generated by people who searched "AI follow-up templates" on Pinterest, copied a prompt, and published the output without reading it all the way through. You can tell because they use phrases like "circle back," "touch base," and "per my last email" within the same paragraph. Somehow simultaneously corporate and passive-aggressive.
The tutorial starts with "just plug in your client's name and hit send," and that's the exact moment the whole thing falls apart. Because your client isn't a variable. She's Maya, who drove 45 minutes for her first lash appointment and mentioned she had a work event coming up. Plugging "Maya" into a generic template doesn't make the message about Maya. It just puts her name on a message that could have been sent to anyone.
The AI isn't broken. The instructions were just wrong from the start.
What Actually Needs to Change
Stop treating AI like a vending machine for words. Put in keyword, get out message. That's not how this works, and it's why the output sounds like a press release.
AI is good at one specific thing that most people overlook: taking a bunch of raw, slightly disorganized human information and turning it into clean, usable language. Which means your job is not to ask it to write a follow-up message. Your job is to give it enough real context that it has no choice but to write something that actually sounds like it came from you.
This is the whole argument behind talking to ChatGPT like a business owner instead of a tech person. You don't need to learn to prompt. You need to learn to have a conversation. There's a difference, and it matters more than any template.
The shift is this: instead of asking AI to write your message, tell AI about your client first. Then ask it to help you write the message. Small change. Completely different output.
The Framework: How to Give AI Enough to Actually Work With
This is a five-part setup. It takes about two minutes the first time, and once you get used to it, it becomes second nature. You're essentially having a quick conversation with AI before you ask it to do anything.
Step one: Tell it who you are and how you talk. Not in a grand way. Just one sentence. "I'm a lash tech. I've been doing this for 17 years. I'm warm but not bubbly. I don't use a lot of exclamation points." That one sentence will change the entire register of what comes back to you.
Step two: Tell it who the client is. Again, one or two sentences. "Her name is Maya. She came in for a full set three weeks ago. She mentioned she had a work event coming up. She seemed nervous before the appointment but left happy." That's it. You're not writing a biography. You're giving AI enough texture to stop writing for a faceless placeholder and start writing for an actual person.
Step three: Tell it what happened last time. Did the appointment go smoothly? Was there something specific you talked about? Did she leave with a plan to rebook and then not? This is the detail that turns a generic check-in into something that makes the client feel remembered. "We talked about how the magnetics would hold up for her event" is a follow-up hook. Use it.
Step four: Tell it what you want the message to do. Not "write a follow-up message" but "I want her to feel like I remembered her and make it easy for her to rebook without feeling pressured." That's a goal. AI can write toward a goal. It cannot write toward a vague concept of "follow-up."
Step five: Tell it what to stay away from. "No corporate-sounding phrases. No 'hope this finds you well.' Keep it short, three sentences max, and make it sound like a text I'd send a regular." This is where most people skip, and it's also why most AI output sounds like it was proofread by a compliance department.
Put all five pieces together in a single paragraph, paste it into ChatGPT, and ask it to write two or three versions so you can pick the one that feels most like you. Then read each one out loud. If you'd never say it in your actual voice, ask it to try again.
What This Actually Looks Like in Practice
Here's the before-and-after, so this doesn't stay abstract.
The generic prompt: "Write a follow-up message to a client who hasn't rebooked."
The output: "Hi [Name], I wanted to reach out and check in after your recent visit. We'd love to have you back! Please feel free to book online or give us a call at your convenience. We look forward to seeing you soon!"
Fine. Inoffensive. Forgettable. Could have been sent by anyone, to anyone, at any time in the last fifteen years.
The context-first prompt: "I'm a lash tech, been doing this 17 years, I'm warm but not overly enthusiastic. My client Maya came in three weeks ago for a full set before a work event. She was nervous but left happy. I haven't heard from her and want to check in without being pushy. Keep it short, casual, like a text from a real person. No 'hope this finds you well.' No exclamation points unless it's one and it's earned."
The output: "Hey Maya, hope that work event went well and those lashes held up for you. If you're ready to refresh them, I've got some spots opening up next week."
That's a text a real person sends. That's a text Maya might actually respond to. And it took the same amount of time as the generic version, maybe less, because you didn't have to agonize over how to start it.
The AI didn't do anything magical. It just took the raw material you gave it and turned it into something usable. That's the whole job.
Can You Turn This Into a Repeatable System Without It Feeling Automated?
Yes, and this is where the real time savings live.
Once you've done this a few times, you'll notice your "who I am" section barely changes. Your voice is your voice. So save that part somewhere. A note on your phone, a sticky note on your computer, a saved message you can paste in. That's maybe 20% of the prompt, already done.
The client-specific part is what changes each time, and that's maybe three sentences. You're basically just making a few notes about the appointment before you send it, the same kind of notes you'd jot in a paper appointment book. "Nervous before, happy after. Work event. Magnetics. Three weeks out." Hand those notes to AI, and it handles the sentence-making.
This is the same thinking behind the AI follow-up workflow I've written about elsewhere. The system isn't "automate everything and forget about it." The system is "handle the part that costs you the most time and mental energy so the part that actually requires you still gets your full attention."
Because the part that requires you is knowing that Maya was nervous. Knowing she had a work event. Knowing she left happy. No template has that. Only you have that. The AI just helps you write it down in a way you'd actually send.
A Note on Volume: When You Have a Lot of Follow-Ups Stacking Up
If you have a backlog, which, if you've been avoiding this, you probably do, don't try to write thirty personalized messages in one sitting. You'll burn out by message four and the last twenty-six will sound like the dental office again.
Pick five. The five clients you most want back, or the five conversations you remember most clearly. Use the framework above for each one. Send them. See what comes back.
Turns out most service business owners aren't avoiding follow-ups because they're lazy. They're avoiding them because every message feels like it requires a fresh decision about tone, about timing, about whether they're being too much. Having a system removes most of those decisions. You still make the human call on the client-specific details. Everything else runs on a track.
And if you want a broader look at how this kind of thinking scales into your whole client communication stack, these exact prompts for client emails walk through the same approach applied to longer-form outreach.
The Proof Is in the Texts I Actually Got Back
I built a website for my nephew Richard, who fixes trucks out of a garage. No marketing degree. No SEO certification. I just paid attention to who Richard's actual customers were and what they actually needed, and I built the site around that instead of around what a website is supposed to look like. Random truckers now text him directly. He texted me: "Business is booming and we appreciate the Google website you created. We have had customers because of it."
Nobody handed me a template for that. What I had was a clear picture of the real human on the other end of the business, and the willingness to let that picture drive every decision.
Follow-up messages work the same way. The AI follow-up message templates that actually land aren't the ones with the most professional phrasing. They're the ones written with a real client in mind, by a real person who remembered something specific, using AI to get the words out of their head and into a form they'd actually send.
That's not a tech skill. That's a people skill you already have. The AI just removes the blank-page problem.
And if you've spent any time in the AI content world, you've probably absorbed the idea that prompting is this complicated, learnable art form that requires study and practice. I wrote a whole piece on why that skills gap is mostly invented. The only thing that actually matters is whether you can describe your situation clearly enough for AI to help. You can. You just did it every time you explained a client situation to a colleague.
Start This Week, Not Someday
Pick one client you've been meaning to follow up with. Open ChatGPT. Write one paragraph: who you are, who they are, what happened at the appointment, what you want the message to accomplish, and what you want it to avoid. Ask for three short versions. Read them out loud. Send the one that sounds most like you.
That's it. That's the whole system on day one.
Every week on Substack I break down exactly how I use AI in a real service business context, no jargon, no hype, no pretending any of this is more complicated than it actually is. If you want the plain-English version of AI for your actual Tuesday, subscribe at Simply Doing AI and it lands in your inbox weekly.
Frequently Asked Questions
Do I need a paid AI subscription to write good follow-up messages?
No. The free version of ChatGPT handles this kind of conversational, short-message task without any problem. Paid tiers add useful features for longer or more complex work, but for writing a few client follow-ups a week, free is completely sufficient.
How do I keep AI follow-up message templates for my service business from sounding generic?
The fix is giving AI something real to work with before you ask it to write anything. Tell it who the client is, what happened at the appointment, and what you want the message to feel like. A generic output almost always means a generic input. The more specific your context, the more specific the message.
What if I don't remember the details of every appointment?
Use what you do remember, even if it's small. "She seemed nervous" or "she booked for a special occasion" is enough to change the tone of the message. If you genuinely have no specific detail, keep the message short and low-pressure, and ask AI to make it feel like a casual check-in rather than a formal outreach.
How many versions should I ask AI to write?
Two or three is usually the sweet spot. One version might nail the tone but feel too long. Another might be the right length but a little flat. Having options lets you pick the best parts without starting over from scratch. You can also ask it to combine elements from two versions if neither one is quite right.
Can I use this system for email as well as texts?
Yes, and the process is identical. The only thing that changes is the final instruction at the end of your prompt. Tell AI whether this is a text (short, casual, conversational) or an email (a bit more room to breathe, but still not formal). The five-part context setup works the same way for both.
Is it dishonest to use AI to write messages to my clients?
You're using AI to help you say what you already mean, the same way people use spell-check, grammar tools, or ask a friend to read something before they send it. The thoughts, the details, the care for the specific client are all yours. AI is just helping you get it out of your head and into a message you'd actually send. That's a tool, not a deception.