AI Counseling Forms and Follow-Up Emails for Esthetic Salons: A Practical Workflow

It is 7 p.m. on a Friday. The owner of a small esthetic salon a friend of mine runs has just waved off her last client of the day, and she still can’t go home. She has six clients’ charts to write up cleanly, a follow-up email to send to a regular starting an intensive body course next week, and a thank-you note with a rebooking reminder for a first-time visitor. Only after all of that can she finally turn off the lights.

“The paperwork is harder than the treatments,” she told me. Her handwritten intake notes run together so she can’t read them later, and for follow-up emails she copies an old message and just swaps the name. Last week that meant she sent a facial promo to someone who had booked a body treatment. She noticed three seconds after hitting send.

Can we shrink these “wiped-out-by-paperwork” evenings? I used her salon as a test bed and handed half of the intake-form and follow-up-email work to AI. Instead of leading with the verdict, let me start with what actually changed.

Key takeaways

• The two biggest time sinks in salon admin, writing up intake forms cleanly and drafting follow-up emails, drop to a few minutes per client once AI builds the templates.
• AI handles the rough draft and formatting. A human always makes the judgment calls about skin and health, checks for any claims that cross legal lines, and presses the send button.
• The golden rule for client data: never hand real names or contact details to the AI. Replace them with codes on the way in, and fill the real names back in only on your own machine.
• I’ve included a copy-paste intake-form prompt and a runnable script that mail-merges follow-up drafts from a spreadsheet.
• Before AI, six clients a day meant about 60 minutes of admin. After a short ramp-up, it’s around 15. Over 20 business days a month, that’s roughly 15 hours of breathing room.

Why salon admin gets tangled in the first place

Let me name the reader first. This article helps owners and therapists at solo-to-five-person independent salons, or small chains, doing facials, body work, hair removal, or slimming. They have booking software, but counseling and email are still done by hand. That is the situation I’m writing for.

A salon day usually flows like this:

1. Confirm bookings, prep the bed and supplies
2. Client arrives, changes, counseling (concerns, condition, medical history)
3. Treatment
4. After-care counseling (home-care guidance, next-visit suggestion)
5. Checkout, rebooking
6. After close, write up charts and send follow-ups

The trouble shows up in steps 2 and 6. During counseling, a question slips your mind and you forget to ask “are you pregnant?”; the notes are scrawled and unreadable later; and the depth of questioning varies from client to client. In step 6, you’re squeezing out email copy with a tired brain, which is exactly when misfires and cold-sounding messages happen.

The classic rework looks like this:

• You miss a medical-history item during counseling, and on treatment day the client mentions “actually, I have a metal allergy,” forcing a menu change.
• A copy-paste slip in a follow-up email mixes in the wrong course promo or another person’s name.
• The thank-you to a first-timer keeps getting bumped, so it goes out three days late and feels awkward.

None of these happen because anyone lacks ability. They come from fatigue and a string of repetitive tasks. That is precisely the part AI is good at taking off your plate.

Three ways to use it in an esthetic salon

Use case 1: Build a complete, no-gaps intake-form template

The first thing to fix is writing from scratch every time. Have the AI produce a few versions of a one-page template that covers concerns, condition, medical history, home care, and the next-visit suggestion, then pick the one that fits your salon. Ask for both a paper version you fill in by hand and a digital version you tap into on a tablet, and the comparison goes fast.

Because the items change by menu, make separate templates for facial, body, and hair removal. To make the form easy to scan, here is a checklist of the must-have fields:

• Chief concern (the one thing bothering them most today)
• Medical history, ongoing treatment, current medications
• Allergies (metal, plant, cosmetics)
• Pregnant or breastfeeding
• Today’s condition: sleep, alcohol
• Usual home care and products used
• Post-treatment home-care instructions
• Suggested next menu and recommended interval
• Consent signature line

Use case 2: Clean up treatment notes into a proper chart entry

Turning a hasty scribble into a polished chart entry you’d be comfortable showing a client is another AI strength. Hand it fragments like “T-zone dry, redness on cheeks, bad shoulder tension, suggested hot stone,” and it returns proper prose covering what you did, the client’s state, and the next suggestion. The key here is that a human verifies the accuracy. AI will sometimes rewrite “redness” as “inflammation” on its own, and you want to avoid that kind of medical-sounding assertion, both for accuracy and because health claims can cross regulatory lines.

Use case 3: Draft follow-up emails tailored to each client’s situation

This is where it pays off most. Next-day thank-yous, post-treatment home-care reminders, next-course suggestions. Template these with the recipient’s situation as variables, and you get a personal-feeling message just by dropping in the name, course, and next date. Later in this article there’s a runnable script that batch-drafts these from a spreadsheet of clients.

Below is the split between “what to delegate to AI” and “what a human must decide.” When in doubt, come back to this table.

Step	OK to delegate to AI	A human must decide
Intake template	List the fields, format the wording	Final call on fit with your menu
Chart write-up	Turn fragments into readable prose	Medical judgment on skin/condition, wording validity
Follow-up email	Draft copy, adjust tone	Check the recipient isn’t mixed up, then send
Suggested menu	Offer a general interval	Whether it truly suits this client
Client data	Format already-coded data	Manage real names/contacts, decide if OK to send

For more on how to phrase your requests, Advanced prompt engineering for Claude Code goes deeper, so if you want to push the quality of the copy up another notch, read that alongside this.

A copy-paste intake-form prompt

Start with the intake form. Paste the instruction below as-is and just change the menu name. The trick is to keep real names and contacts out and have it produce a blank template.

You are an experienced esthetic salon counselor.
Create a paper-fillable intake-form template for a first-time "Facial" client.

Conditions:
- Must include the fields: chief concern, medical history/medications, allergies,
  pregnancy/breastfeeding, today's condition, usual home care, post-treatment
  instructions, next-visit suggestion, and consent signature.
- Write each field as an easy-to-answer question, with checkboxes where options apply.
- Do not use wording that could be mistaken for medical treatment or that asserts results.
- Keep it to a single A4 page.
- Output headings and bullet points only. Do not include any real personal information.

For the follow-up email draft, ask with the situation as variables. Keep the name as a code here too.

Write 3 versions of an esthetic salon after-care follow-up email, in a polite,
non-pushy tone.

Context:
- Address the recipient as "{{client}}" (no real name)
- Menu received: {{menu}}
- Visit: first time
- Purpose: a thank-you plus a light note on the recommended timing for the next visit
- Avoid: wording that asserts results, medical phrasing, overly long body text
- Give each version a subject line; keep the body around 200-300 characters

Pick one of the three, swap {{client}} and {{menu}} for the real values on your own machine, and send. This “a human does only the final swap” flow prevents misfires and data leaks at the same time.

A runnable script to batch-draft follow-ups from a client list

You don’t need to be an engineer here; following the flow is enough. This is a minimal script that reads a CSV you built in a spreadsheet (names as codes, only menu and next date filled in), drops the values into a template, and mass-produces drafts. It runs with Node.js. The safety device is that it never sends anything, it only writes out draft files.

First, prepare the data. Don’t use real names; use an internal code like A001.

code,menu,next_date
A001,First-time Facial,2026-06-21
A002,Body Intensive 90min,2026-06-23
A003,Facial Hair Removal,2026-06-28

Put the draft-mails.mjs below in the same folder and run it. All it does is “read each row, fit it into the template, save as text.”

import { readFile, writeFile, mkdir } from "node:fs/promises";

// Code-to-copy template. Never place real names or contacts in here.
function buildMail({ code, menu, next_date }) {
  const subject = `Thank you for your recent visit (${menu})`;
  const body = [
    `Dear ${code},`,
    "",
    `Thank you for coming in for your ${menu} today.`,
    "After your treatment, please drink a little more water than usual and avoid rubbing the area.",
    `Your next visit is best around ${next_date}. Reach out whenever it suits you.`,
    "",
    "We look forward to seeing you again.",
  ].join("\n");
  return { subject, body };
}

const csv = await readFile(new URL("./customers.csv", import.meta.url), "utf8");
const [head, ...rows] = csv.trim().split(/\r?\n/);
const keys = head.split(",");

await mkdir(new URL("./drafts/", import.meta.url), { recursive: true });

let count = 0;
for (const line of rows) {
  const cols = line.split(",");
  const rec = Object.fromEntries(keys.map((k, i) => [k, cols[i]]));
  const { subject, body } = buildMail(rec);
  const out = `Subject: ${subject}\n\n${body}\n`;
  await writeFile(new URL(`./drafts/${rec.code}.txt`, import.meta.url), out, "utf8");
  count++;
}

console.log(`Wrote ${count} drafts to drafts/. Always review them by eye before sending.`);

Running it is just this:

node draft-mails.mjs

A text file per code lands in the drafts/ folder. From there you check the contents, replace the codes with real names, and paste into your email client. If you want the AI to write more natural body copy, swap the inside of buildMail for the output of the prompt in the previous section and the template quality jumps a level. If you’re stuck on Claude Code’s initial setup, it’s faster to get your environment ready first with the Claude Code getting-started guide and then come back.

Privacy and security notes

This part can’t be skipped. An esthetic salon handles not just names and contacts but sensitive information like medical history and allergies. The line to hold is simple:

• Never write real names, phone numbers, email addresses, or home addresses directly into your AI input. Replace them with a code like A001.
• For health-related notes like medical history, abstract them where you can, to something like “has an allergy,” before passing them along.
• Fill personal data into a generated draft outside the AI, only on your own computer.
• Eyeball the recipient and body every single time before sending. Always keep one buffer step before feeding anything into a bulk-send tool.
• If staff use it, write “what may and may not go into the AI” onto a single-page rule and share it.

The larger the salon gets, the more that data-handling rules getting locked inside one person’s head invites accidents. For building team operating rules, the thinking in CLAUDE.md best practices applies directly, and if you want to get adoption to stick across several staff members, Claude Code productivity tips is worth a look too. For the basics of handling personal data, the U.S. Federal Trade Commission’s privacy and security guidance is clear and worth a read once for peace of mind.

Before, after, and a rough ROI

Numbers make the decision easier, so here’s an estimate close to what I actually measured at my friend’s salon (one owner plus one part-timer, six clients a day on average).

Item	Before	After
Chart write-up (per client)	~6 min	~2 min
Follow-up email (per client)	~4 min	~1 min
Daily admin total, 6 clients	~60 min	~18 min
Misfired emails	1-2 / month	near zero

That’s roughly 40 minutes saved a day, about 13 hours over 20 business days a month. At a notional $15/hour, that’s close to $200 a month in cost saved, or that time freed up for service and retail. The setup cost is a few hours of building the first templates. At that scale it paid for itself inside a month.

FAQ

Q. Is it OK to send what the AI wrote straight to a client? A. Avoid sending it as-is. Use it as a draft and have a human read it last every time. In particular, check each time that no result-asserting language or medical-sounding phrasing has slipped in.

Q. Can I adopt this even if I’m not good with computers? A. If all you do is paste the template prompt and generate copy, no special knowledge is needed. The batch-draft script raises the bar, so start with just the prompts and move on to automation once you’re comfortable. The non-engineer’s starting point is laid out in Claude Code for non-engineers.

Q. Can I have it read my client roster and send automatically? A. I don’t recommend handing the AI real names and contacts from the roster. The safe shape is to code them, let the AI handle up to the draft, then swap in real names and send on your own machine.

Q. Can it connect to my existing booking system? A. If your system can export to CSV, you can use it as the mail-merge source just like the script in this article. Even without a direct integration, the export to code to draft flow is plenty practical.

What I confirmed in practice

I checked three things. First, I had it build intake templates for facial, body, and hair removal; the gaps in fields disappeared, and even a new part-timer could ask questions at the same depth. Second, cleaning up scrawled notes worked exactly as the “hand it fragments, get tidy prose” flow promised, and write-up time felt like it dropped to a third. That said, there was one case where it rewrote “redness” as “inflammation,” which confirmed a human can’t be removed from the wording check. Third, the batch-draft from CSV generated ten people’s drafts in under a minute. I deliberately did not automate sending, keeping the one step of turning codes back into names, and the misfires I mentioned were zero throughout the trial.

The Friday evenings that used to wipe her out with paperwork now end a little earlier. That is the real result. When you reach the stage of making this stick as a system across the whole salon, talk to us about operating rules and training from the training and consulting page.