Match each service to the one or two products you would genuinely use on that client's hair, give the AI the product facts from the brand's own page, and have it write a short post-appointment text with the product name, how to use it and the price. Check the margin first, and keep health claims out.
Honesty is the part of the method that makes it work and the part AI will not do for you. A model asked for "persuasive" product copy adds benefits the brand never claimed, invents urgency and sometimes promises results, all of which clients notice and advertising rules can catch. So the facts go in, the claims stay as written, and a person who knows the client's hair chooses the product.
Build a product card for every item before any message
A product card is one line of facts per product: what it is, the size, what you charge, what it costs you, and the margin. The brand's own description goes beside it, copied word for word from the brand's product page or the label, because that wording is the only source the AI may draw on. Here are illustrative cards for five products. The last column takes the card fee into account, because most retail is paid for by card.
| Product | Retail | Cost | Gross margin | After 2.5% card fee |
|---|---|---|---|---|
| Colour-care shampoo, 250 ml | $24.00 | $11.50 | 52.1% | 49.6% |
| Toning shampoo, 250 ml | $22.00 | $10.20 | 53.6% | 51.1% |
| Leave-in curl cream, 150 ml | $26.00 | $13.40 | 48.5% | 46.0% |
| Heat-protection spray, 150 ml | $21.00 | $10.50 | 50.0% | 47.5% |
| Nourishing mask, 200 ml | $32.00 | $16.80 | 47.5% | 45.0% |
The margin is retail minus cost, divided by retail: for the shampoo, ($24.00 − $11.50) ÷ $24.00 = 52.1%. As a rough guide to card fees, Square's pricing page currently lists 2.6% plus 15 cents for in-person payments on its free plan and 2.5% and 2.4% on higher plans (October 2026); your own processor statement sets the real figure. The product-by-product margin tutorial shows how to run the same check across everything you sell, and the salon pricing calculator tutorial does it for services.
What is the margin for? Two uses only: comparing two products that are both right for the client, and testing an offer before you make it. It must never decide which product a client needs. If the client's hair needs the 47.5% mask and not the 53.6% toning shampoo, the mask is the recommendation. Selling someone the wrong thing costs trust, and that is worth more than the margin difference.
Copy the brand's wording from its own product page or the label on the bottle, and if the two disagree, use the label. If the page makes a claim you would rather not repeat, don't repeat it. If a claim is not on the page or the label, it is not yours to add, however natural it sounds.
Run any offer through the card before it goes into a message. A 10% offer on the colour-care shampoo makes it $21.60 and drops the gross margin from 52.1% to 46.8% (44.3% after card fees), which is $10.10 of profit instead of $12.50. A two-for-$44 offer costs you $23.00 and leaves $21.00, or 47.7% (45.2% after card fees). Neither is wrong, but each is a decision to take deliberately and to state in the message as a plain fact with an end date.
Map services to home care, with the stylist's signature on it
The map is the single most important document in the method, and it comes from your stylists, not from AI. It says, for each service, which product they would suggest and what to check before suggesting it. Fifteen minutes with the team produces the first version.
| Service done | Product the stylist would suggest | Check before suggesting |
|---|---|---|
| Permanent colour | Colour-care shampoo | The colour brand's aftercare sheet for when to first wash |
| Lightening or blonding | Toning shampoo, with the mask if budget allows | That the stylist actually toned; the brand's directions for how often |
| Curly cut | Leave-in curl cream | Curl type in the consultation notes |
| Blow-dry or styling | Heat-protection spray | That the client heat-styles at home |
| Smoothing or bond treatment | The treatment brand's own aftercare pack | Follow the brand's wash and product rules exactly; do not improvise |
| Child's cut | Nothing | Recommend only if the parent asks |
Two columns hide the value. The "check" column stops the AI, or a rushed receptionist, from suggesting something that conflicts with what the stylist did. And the "nothing" row gives staff permission not to sell. Consultation notes are the best raw material for the map; the hair salon consultation form tutorial shows how to record colour history and consent so the stylist's choice is based on what is actually written down.
The prompt: product facts in, a short message out
The prompt contains the product card, the service done and the rules, and nothing about the client except a first-name placeholder. Colour formulas, allergies and treatment histories are sensitive; keep them out of the chat and merge names in later with your booking or messaging tool.
Write a post-appointment text from a hair salon to a client.
Service done today: permanent colour.
Product to recommend (use only these facts):
- Colour-care shampoo, 250 ml, $24. Brand's description: "[paste the
brand's own sentence here]". Directions on the label: "[paste]".
Rules: one product only; under 60 words; start with "Hi [first name]";
say how to use it, copying the label; give the price; say where to get
it; no claims about health, damage repair, hair loss, scalp conditions,
safety or results; no urgency or stock claims; no discounts; plain
English with straight apostrophes and no emoji; end with "[opt-out line]".An illustrative first reply from an unconstrained assistant might read: "Hi [first name]! Your gorgeous new colour deserves the best care! Our colour-care shampoo repairs damage, stops fading and is dermatologist approved. Only 3 left, grab yours before it's gone! $24." Look at what it did. "Repairs damage" and "stops fading" are claims the brand never made, "dermatologist approved" is an endorsement from nowhere, "only 3 left" is urgency you haven't checked, and the message gives no directions and no opt-out. With the rules above you get something closer to this:
Hi [first name], thanks for coming in today. For the colour we did, I'd use the colour-care shampoo (250 ml, $24) for your washes at home, following the directions on the label. It's at reception, or reply and we'll put one by for your next visit. [opt-out line] [stylist name], [salon name]
It is plainer and shorter than the first draft, and it contains nothing the brand or the stylist hasn't said. The sender's name matters: a text from [stylist name] reads as advice from the person who did the hair, while one signed only by the salon reads as promotion.
Check the length before anything goes out. Filled in, that message runs to about 280 characters, which most text services bill as two segments, because anything over 160 characters is split into parts of about 153. That is why the prompt asks for straight apostrophes and no emoji. Assistants often produce curly apostrophes, and a single one switches the whole text to an encoding that allows only 70 characters a segment, so the same message would bill as five. Some messaging tools convert curly quotes for you; check whether yours does before relying on it.
Draft a day's messages in one go, and four more finished examples
Once the first message reads right, stop writing them one at a time. Give the assistant the map, the cards and a list of the day's finished services, with no names:
Here is our service-to-product map and four product cards [paste].
Today's finished appointments, service only: 1 permanent colour,
2 curly cut, 3 blonding, 4 child's cut, 5 blow-dry.
For each one, write the text using the map. If the map says nothing,
write "no message". Use only facts on the cards. Same rules as before.A good reply produces four messages and "no message" for the child's cut. If it writes a message for the child, the "nothing" row wasn't clear in the prompt, so fix the prompt rather than editing the message. Read each draft against the card, then merge in names from your own system. These are four more finished examples (illustrative) that follow the same rules.
After blonding, where the stylist toned the hair:
Hi [first name], thanks for coming in today. After the lightening we did, I'd use the toning shampoo (250 ml, $22) on wash days, following the directions on the label. It's the one I used on your hair. It's at reception, or reply and we'll put one by. [opt-out line] [stylist name], [salon name]
After a curly cut:
Hi [first name], good to see you today. For your curls I'd suggest the leave-in cream (150 ml, $26), used on damp hair as the label directs. It's at reception, or we can have one ready for your next visit. [opt-out line] [stylist name], [salon name]
When budget is tight, with two products on the map:
Hi [first name], thanks for today. If you're choosing between the two, start with the toning shampoo ($22) and leave the mask ($32) for another time; there's no need to buy both. Both are at reception. [opt-out line] [stylist name], [salon name]
And a reorder line, added to the appointment reminder for clients who bought a product before:
Hi [first name], your appointment is on [date] at [time]. If you'd like another colour-care shampoo (250 ml, $24), reply and we'll have one waiting. [opt-out line] [salon name]
No health claims: a ban list, safer wording and a pass-or-fail test
Keep a short ban list beside the prompt, so every draft can be checked against it in seconds. The right-hand column shows what to do instead.
| Phrase to remove | The problem | Do this instead |
|---|---|---|
| "Repairs damaged hair" | A result claim the brand may not make | Quote the brand's own description, or just name the product and how to use it |
| "Stops hair loss" or "promotes growth" | A health claim | Leave it out; suggest a doctor if the client raises a concern |
| "Treats dandruff, dry scalp or eczema" | A medical claim | Send no product message; suggest a doctor or pharmacist |
| "Dermatologist tested" or "clinically proven" | An evidence claim | Only if the brand's page says so, quoted exactly |
| "Safe for sensitive scalps" or "non-toxic" | A safety claim | Point to the label and ingredient list |
| "Guaranteed shine" or "lasts eight weeks" | A promise of results | "Use as directed on the label" |
| "Only 3 left!" | Pressure, and true only if checked | Delete it |
Then apply a three-question test before any message is sent. Does every claim appear on the brand's page or label, in the brand's words? Does the message promise a result? Does it mention a condition such as irritation, hair loss or an allergy? The first answer must be yes and the second and third must be no. If either of the last two is yes, rewrite the message, or, for a condition, don't send a product message at all and call the client. Rules about advertising claims for cosmetics differ by place, so when you are unsure whether a phrase is allowed, ask the brand's own support team or an adviser, not the chat assistant. For a business where the same line applies to skin, the tattoo aftercare card tutorial covers aftercare wording without medical claims.
When to send, who to send to, and the consent wording
Send the message the same afternoon or the next morning, before the first wash at home, because that is when the advice is useful and the visit is fresh. One message per visit is plenty. Send only to clients who agreed to receive messages of this kind, and check whether your booking system records that agreement, for example in a marketing-consent field. Rules on the difference between service messages and marketing, and on opt-outs, vary by place, so ask an adviser if you are not sure which side an aftercare text falls on.
Capture agreement at booking or check-in with a plain question, and record it:
Can we text you aftercare tips and product suggestions after your visit? You can opt out at any time.
Provide an opt-out line in every message that your messaging tool actually supports, and act on opt-outs at once. A sale made to someone who didn't want the message is worse than no sale.
Honest selling: seven rules the whole team agrees to
The messages only work if the conversation at the chair matches them. Write these on one page and have every stylist sign it.
- Recommend only what you would use on that client's hair.
- No more than two products in any recommendation, and say which one matters if money is tight.
- Give the price in the message, and the size.
- Tell the client it is fine to say no, and mean it.
- Check stock before the message goes out.
- Never let margin decide what a client needs.
- Log what was recommended, so you can see what sells and what comes back.
Rule 7 is the easiest to skip, and without it you cannot tell a good message from a lucky Saturday. A log with date, service, product and outcome is enough. If a product doesn't suit a client, take it back or exchange it, and say that in your shop policy so nobody has to ask.
A week in a three-chair salon, with the arithmetic
This is the main worked example, with illustrative figures. A three-chair salon does about 90 appointments a week. Sixty of those clients agreed to messages, and 36 had a service on the map. So 36 messages go out, each from the stylist who did the hair, within a day of the visit. During the following seven days four of those clients buy: two colour-care shampoos at $24, one curl cream at $26 and one toning shampoo at $22. Sales come to $96.
The gross margin on those four sales is 2 × $12.50 plus $12.60 plus $11.80, or $49.40. The card fee on $96 is about $2.40, leaving about $47.00. Four buyers from 36 messages is 11%. If the salon had been selling 14 retail items a week to 90 clients (15.6%), it is now selling 18 (20%). That is an illustration of how to calculate, not a forecast: it assumes none of the four would have bought anyway, your conversion may be higher or lower, and the point is to measure it. The cost side is small. A paid assistant is about $20 a month at list price, and the free plans are enough to start. The real cost is the half hour a week to check the drafts and send them.
If the same week had gone worse, with 36 messages, one sale, $24 of sales and $12.50 of gross margin, the sums would show that the messages cost more in time than they earned, and the right response would be to test a different message or product, not to send more. If you want to keep client touchpoints sensible, the upsell triggers tutorial lists five triggers worth setting up first.
What to do when the client replies
A reply is the best data the system gives you, and how you handle it decides whether the next message is welcome. Keep three answers ready, written once and edited by the stylist. For a yes:
Lovely, we'll put one by at reception for [day]. You can pay when you collect it, or we can add it to your next visit. [stylist name]
For a question about use, copy the label rather than improvising, and offer to show them in person:
Good question. The label says: [copy the directions]. I'm happy to show you at your next visit if that helps. [stylist name]
And for a no, which is a perfectly good outcome:
No problem at all, thanks for letting me know. See you on [date]. [stylist name]
After a no, note it in the log and don't send that client a message about that product again. Two refusals of the same kind, such as price, tell you something about the product or the match, not about the client. If someone replies with a skin or scalp concern, treat it as the earlier ban list says: no product advice, a kind suggestion to see a doctor or pharmacist, and a note on the record.
Five slips to look for in your first batch of messages
Five realistic ones, with how each one appears.
- The product isn't in stock. The message names a shampoo the salon sold out of last week. It shows up as clients asking at reception for something that isn't there. Fix it by checking stock before sending, and removing that product from the map when it is out.
- The advice conflicts with the stylist's. A toning client gets a message recommending the wrong shampoo for the service. It shows up when a stylist reads the sent log. Fix it with the "check before suggesting" column and a stylist reading the first ten drafts.
- The price is stale. A supplier raised costs and the message still quotes the old price. Keep prices in the product card and paste from the card every time.
- A client who never agreed gets a message. It shows up as a complaint or a reply asking to stop. Send only from the consent list and honour the opt-out immediately.
- Two messages land in a day. The aftercare text and a rebooking reminder arrive together. Space them and allow one product message per visit.
Measure it without fooling yourself
Track four figures each week: messages sent, replies, purchases within seven days, and retail sales per appointment across the whole salon. Judge the result over at least four weeks, because one weekend of weddings or a bad forecast can move a single week more than your messages do. To compare two wordings, alternate them by day of the week and keep every other rule the same; the winner is the version with the higher purchase rate over a comparable number of messages, and a handful of sales is not evidence. To see whether the messages sell anything at all, rather than catching purchases that would have happened anyway, hold the message back on one fixed day each week and compare the seven-day purchase rate of consenting clients from that day with everyone else's. Say that over four weeks the held-back clients buy at 8% and the messaged ones at 11%: the messages are adding about three points, not eleven, and that is the figure to weigh against the half hour a week they take. For the rebooking side of the same visit, the rebooking and diary-gaps tutorial covers rebooking messages.
Finally, retire what doesn't earn its place. If a product on the map is returned more often than it sells, or clients stop replying to it, the problem is the product or the match, not the wording. Take it off the map and use the AI's time on a better one. The goal of the whole system is not to sell more of everything; it is that the right product reaches the right client with enough information for them to say yes or no freely.
Questions salon owners ask about retail messages
Which number should aftercare messages come from?
Send from the salon's business number or booking-system messaging, not a stylist's personal phone, so replies are logged and anyone can cover for a colleague. Sign each message with the stylist's first name, because it is advice from the person who did the hair. Personal phones also mix client data into private accounts, which is hard to unwind later.
Can I send these through WhatsApp or another messaging app?
Check that platform's own rules and charges for business-initiated messages before relying on it, because they differ from ordinary texts and change. Template approval, opt-in requirements and per-message charges may apply. Whatever the channel, the consent, opt-out and no-health-claims rules in this tutorial still apply in full.
What if the product I want to recommend is out of stock?
Do not imply stock you do not have. Either recommend a similar product your stylist would also use, or say you can order it and give a realistic date. Check stock before the message goes out, and remove sold-out products from the service map until they return, so the AI cannot recommend them.
What if a client says they can buy it cheaper online?
Be honest and brief. You may not match the price, so say what you offer: the stylist chose it for their hair, directions in person and no waiting. Do not criticise other sellers, and let the client decide. If you do match prices, state the rule in your shop policy so staff apply it consistently.
Further reads
- Can AI Manage Salon Inventory? Colour Stock and Reorder Points — Keeping the products you recommend in stock.
- How Much to Charge for Hair Extensions: Fitting, Hair and Aftercare — Aftercare as part of a priced service.
- AI Assistant for Hair Salons: ChatGPT, Claude, Gemini or Copilot? — Which assistant suits a hair salon, and for which jobs.
- Salon Patch Test Form Checklist: Reminders and Records With AI — Reminder and record wording for patch tests.
- AI Marketing for Salons: Fill Diary Gaps With Posts, Emails and Offers — A four-week plan for posts, emails and offers.
- Chatbot for Hairdresser Marketing: Offers, Win-Backs and Ad Replies — Offers and win-back messages from a hairdresser's chatbot.
- Chatbot for Hairdresser Training: Onboard New Stylists With AI — A training bot that answers new stylists only from your manual: standard cards, tool choice, a 25-question test, a four-week plan and a Friday spot-check.
- Gift Voucher Terms and Conditions for a Salon: Draft Them With AI — Gift voucher terms for a salon: nine decisions, clauses to adapt, an AI drafting prompt and the checks a local professional should make.
- Nail Salon Receptionist Job Description: What AI Can Take Over — An owner's duty-by-duty split of a nail salon front desk: what AI handles, what a person keeps, and the handover rules for walk-ins and complaints.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Square's pricing page (in-person card rates, checked October 2026). Product names, prices, costs and results are illustrative; copy product wording from each brand's own page.