AI for Optometrists: Contact Lens Aftercare and Reorder Messages

An optometry practice can use AI to draft contact lens aftercare check messages and reorder reminders for each lens type, then send them through its recall or messaging system. AI writes the wording; clinicians set the timings, stop rules and escalation, and any reply about discomfort goes to a person at once.

The split matters more here than in most small businesses, because a lens wearer who writes "my eye is a bit sore" is describing a possible clinical problem, not a customer-service query. The design below keeps the machine on the repeatable parts, which are the timed nudges and the reorder prompts, and keeps every judgement with the practice's staff.

Follow me on Instagram@sagnikteaches

What to automate, what to hand to a person

Contact lens work generates four kinds of message, and they carry different levels of risk.

Connect on LinkedInSagnik Bhattacharya

This is a narrower task than answering general patient queries. For that wider job, AI receptionist for optometrists: eye test bookings and recalls looks at the calls side, and how independent opticians can use AI for recalls and bookings covers recalls overall. Here the focus is contact lens wearers only.

Subscribe on YouTube@sagnikteaches

Step 1: write the schedule rules before any wording

A message sequence is a schedule with sentences attached. The schedule belongs to your practitioners. Sit down with them and write a one-page table per lens type with three columns: the date offsets for each message, the purpose and the stop rule. The timings below are illustrative placeholders only, since your own clinical protocols and local professional guidance set them.

Say a four-room practice fits about 30 new wearers a month and keeps 900 established wearers on file. If each aftercare invitation currently depends on someone running a report and writing individual texts, a rules-based sequence replaces about 8 hours a month of manual chasing. The saving is only real if the schedule is clinically approved, so get the table signed off in writing by a named practitioner before you build.

Step 2: have AI draft the messages from approved facts

Give the AI the schedule row and a short list of facts you approve, and ask for a set of messages in your voice. Do not paste patients' details into the prompt. Merge fields such as the first name are inserted later by your recall software.

Write a four-message text sequence for a new wearer of daily disposable contact lenses at an independent optometry practice.
Messages: (1) day 2 check-in, (2) day 10 handling tip, (3) day 14 invitation to book the aftercare check, (4) a supply reminder about three weeks before the lenses run out.
Facts you may use: the practice name is [practice]; aftercare checks are 20 minutes; booking link is [link]; phone [number]; daily lenses are worn once and thrown away.
Every message under 300 characters (two standard text segments at most), straight apostrophes only, no emoji. Friendly, plain, British English. Start each with "Hi {first_name},".
In message 1 include: "If your eyes feel sore, red or your vision is blurred, take your lenses out and call us on [number]."
Do not give medical advice, diagnoses or lens-care instructions beyond the facts above. Add "Reply STOP to opt out" to every message.

An illustrative output for message 1 reads: "Hi {first_name}, how are you getting on with your new lenses? If your eyes feel sore, red or your vision is blurred, take your lenses out and call us on [number]. Reply STOP to opt out." Message 3: "Hi {first_name}, it's time for your two-week lens check at [practice]. It takes about 20 minutes: [link]. Reply STOP to opt out."

Check each for three problems: invented advice ("rinse with water" is a classic, and wrong: public-health guidance says to keep lenses away from all water), length (a standard text holds 160 characters, or 153 per segment once it splits, so anything over 306 becomes a third billed segment, and one curly apostrophe or emoji cuts the limit to 70 characters, or 67 per segment), and any promise ("your lenses will feel comfortable in no time"). Delete the advice and the promise, and trim anything too long. Your practitioners read the final set once, and that is the sign-off you keep.

Step 3: build the stop and escalation rules

The reply rule is the point of the whole design. Write it as code-free plain English that your recall or messaging tool can implement, or that a receptionist can follow if the tool is simple.

1. Any reply containing a symptom word (sore, pain, red, blurry, blurred, sensitive, light, sting, discharge, watery, itchy, can't see, worse) halts the sequence for that patient and creates a task for a named member of staff, marked same-day.
2. The system sends only one fixed reply: "Thanks for letting us know. A member of the team will call you today. If your eye is painful, your vision has dropped, or you're worried, take your lenses out and call [number] now, or seek urgent care." The wording is approved by a practitioner; it is not generated on the fly.
3. A staff member reads the reply and calls within the practice's stated time. They record the outcome in the patient record.
4. The sequence stays off until a clinician says it may resume. No automated message arrives for a patient with an open symptom task.
5. Anything ambiguous ("not great", "meh", a thumbs-down emoji, an unrelated question) goes to staff too. When in doubt, a person reads it.
6. Opt-out words (STOP, unsubscribe, no more) end all reminder messages at once and are recorded.

Test this before going live. Ask a colleague to reply to a test message with "eyes feel a bit gritty", with "all fine thanks", with "STOP", and with a photo. Confirm the first creates a task and stops the sequence, the second does nothing harmful, the third unsubscribes, and the fourth lands in a staff queue. An untested escalation rule is a hope.

Why a keyword list is the floor, not the ceiling

A keyword list misses phrases such as "it feels like something's stuck". If your messaging tool offers AI classification of replies, you can use it as an extra net that flags anything that might be a symptom, but never as the only net, and never to answer. The classification's only permitted action is "send to a person". A false alarm costs a staff member a thirty-second read. A missed symptom is the failure that matters. For that reason, accept a high false-positive rate: staff would rather read ten harmless replies than miss one that matters.

Consent for texts: record it, keep it, respect it

Reminders about a booked appointment and promotional or reorder messages are often treated differently, so ask your adviser which rules apply to your practice and where your patients are. A sensible baseline is the same everywhere. Ask at the exam, in the form or at the desk, for permission to text about lens checks and reorders. Record the date, the patient's yes, and the channel. Offer an easy opt-out in every message. Never send a first message to a number that has not agreed. If you have patients in the EU, data-protection law treats a recorded consent and a clear privacy notice as essential for this kind of use, and you should have your notice reflect that AI helps to write, but not to decide, the messages.

A wording that works on the exam form: "May we text you reminders about lens checks and reordering? You can reply STOP at any time." Record it in the patient record as a field, not a note, so the recall tool can read it. The practical tip is to make "consent given" a required filter on the list that feeds the sequence: no consent, no message.

Step 4: reorder messages that stay useful and honest

Reorder messages are the commercial part of the sequence, and they are where wording can drift into pressure. Keep them useful: a link to reorder, the date the supply is estimated to run out, and an honest note that wearing lenses beyond their replacement schedule is not recommended. Never state an expiry or supply date that your system has not calculated.

Write two reorder reminder texts for a contact lens practice.
Version A: for a patient whose supply of monthly lenses runs out in about 3 weeks. Version B: the same patient, 7 days later, no reorder yet.
Facts: reorder link [link]; phone [number]; the aftercare check must be up to date to order, which we will tell them if it is not.
Under 280 characters each. Friendly, no urgency tricks, no discounts unless stated. Include "Reply STOP to opt out".
Do not mention lens brands or health benefits.

An illustrative version A: "Hi {first_name}, your monthly lenses should run out in about three weeks. Reorder here: [link]. If your lens check is due we'll let you know when you order. Reply STOP to opt out." Version B is a softer repeat with the phone number. Stop after the second reminder. A third nudge to someone who has chosen to buy elsewhere erodes trust, and the practice will see opt-outs rise.

A rule worth having: the reorder link is only live for patients whose check is in date. If a patient's check has lapsed, the same sequence swaps the reorder message for the aftercare invitation. That gives the practice a natural, non-pushy route to booking and keeps supply linked to clinical oversight. Your recall software must support the condition; if it does not, a weekly list pull by a receptionist does the same job.

One practice's first month of lens messages, in numbers

Say a practice with two optometrists and 600 contact lens wearers switches on the sequence for all patients with recorded consent. Suppose 420 have consented. In month one, each of the 420 receives a first supply or aftercare message, spread over the first fortnight, and the timed sequence runs from there. The numbers below are illustrative.

• Reply volume: about 60 replies in the first month, of which most are "thanks" or a booking confirmation.
• Symptom-flagged replies: 5. Each created a same-day task. Three were mild dryness and were advised by a clinician by phone, one was booked for the next morning, and one was a different issue entirely. No automated message was sent after the fixed safety line.
• Opt-outs: 6 (1.4 per cent of those messaged).
• Bookings from aftercare invitations: 38 across the month, tracked by the booking link.
• Staff time: about 2 hours across the month (60 replies read at under a minute each, plus five calls of 10 to 15 minutes with notes), against roughly 7 to 8 hours a month of report-running and individual texts before.

The point of listing these is to decide in advance what you will measure. Review the symptom-flag count first, since it tests the safety net, then the opt-out rate, then bookings. If opt-outs climb above a few per cent, the messages are too frequent or too pushy, and you cut them. For reminder timing that holds up, how to reduce no-shows with AI reminders has useful spacing ideas that apply to check bookings too.

A cheerful lens text, rewritten without the sales pitch

Before (illustrative AI first draft): "Hi [first name]! Hope you're loving your new lenses! Remember, great eye health starts with regular check-ups, so book your aftercare today. Dry eyes? Try our recommended drops from the shop!"

After: "Hi {first_name}, it's time for your two-week lens check at [practice]. It takes about 20 minutes: [link]. If your eyes feel sore or red, take your lenses out and call us on [number]. Reply STOP to opt out."

Three edits: the product pitch and the implied advice about dryness are gone, a symptom line replaces the upbeat guess, and the personalisation uses a merge field so that a real name never passes through the AI tool. Your edit list will look similar: delete enthusiasm, delete advice, add the safety line and the opt-out.

A filled-in escalation log, and the wording standards that go with it

Staff handling flagged replies need somewhere to write what happened, and the practice needs a record that the safety net worked. A one-line-per-event log does the job. Here is an illustrative week from the pilot above, with no names and no clinical detail beyond what a log needs.

The second row is the one to learn from: no keyword fired, yet a person read the reply and caught it, which is the reason the "ambiguous goes to staff" rule exists. If you see a pattern of phrases that the keyword list missed, add them, and note the date you changed the list.

Message wording has its own standards. Aim for short sentences and common words, and avoid abbreviations such as "CL" or "AC" that patients do not use. Do not rely on colour, emoji or images to carry meaning, since some recipients use screen readers or read on a dim screen. Never include prescription details, powers or brand names in a text, as a phone left on a table is visible to anyone. Check that the first line of each message, which shows in the notification preview, makes sense on its own and does not reveal that the recipient wears lenses to someone glancing at the screen if you think that could matter to your patients. Read each message aloud at the desk before approving it. If it sounds like a script, shorten it.

Where lens message sequences misfire

Wrong lens type. A patient who wears monthlies receives a dailies tip because the recall list was built by recall date, not lens type. Build the segments from the lens field, and spot-check twenty records before each launch.

Quiet failure of the stop rule. A patient replies with a symptom, the task is created but sits unread over a weekend. Give the task an owner, a same-day target and a Monday-morning check for anything left open.

Message after a clinical contact. A patient had an urgent visit on Tuesday and receives a cheery reorder nudge on Wednesday. Suppress the sequence whenever a clinical task is open or an unscheduled visit is logged.

Copying competitors. A model asked for "lens reminder text" may borrow sales language from online retailers, including discount urgency. Keep your voice by pasting two or three of your own best-sent messages as style examples in the prompt, and ask the model to match them.

If you are writing the standing instructions for all of these, store them in a ChatGPT Project or Claude Project so every team member drafts from the same rules. Both let you keep project instructions, and neither requires a patient detail. For further context on what these tools can and cannot do at the desk, see ChatGPT for optometrists: recall letters, reviews and front-desk replies, and for the hub of related tutorials, the AI guide.

A rollout plan for the first six weeks

1. Week 1. Practitioners write the schedule table and the symptom words. Reception checks that the consent field exists in the recall system.
2. Week 2. Draft the messages with AI, edit them, and have a practitioner approve in writing.
3. Week 3. Build and test the stop rules with the four test replies. Fix gaps before any real patient receives anything.
4. Week 4. Pilot with one lens type and 30 to 50 consenting patients. Read every reply yourself.
5. Week 5. Review the symptom flags, opt-outs and bookings, then edit wording.
6. Week 6. Extend to the next lens type. Keep a monthly check of the first three numbers.

The order is deliberate: safety rules before volume, and one lens type before four. If you would rather work through the set-up with someone who can see your recall system, that is what my AI implementation consultation is for, and the 1:1 call route is the way in.

Questions about automated contact lens messages

Should AI reply to a patient who says their eye is sore?

No. Any reply mentioning pain, redness, blurred vision, light sensitivity or discharge goes straight to a clinician or trained staff member. The automated system's only job is to detect the words, stop the sequence and alert a person, with a standard safety line sent at once.

Do I need consent before sending text reminders?

Yes, in most places you need a recorded, specific permission for marketing and reminder texts, and a clear way to opt out. Reorder prompts can count as marketing. Capture consent at the exam, record the date, and have your adviser check the rules where you operate.

Can AI decide when a patient is due an aftercare visit?

It should not. Your practitioners set the intervals for each lens type and patient. AI can draft the messages and your booking or recall system can send them on the dates you set. Keep the schedule rules in a clinician-approved table.

What data should the drafting tool see?

Only the message template and generic lens-type details. Names, prescriptions and clinical notes stay in your practice system. Merge fields insert the patient's first name inside the recall software, not inside the AI tool.

Further reads

• AI Chatbot for Opticians: Appointments, Repairs and Lens Questions — A chatbot for bookings, repairs and lens questions.
• AI Scribe for Optometrists: Exam Notes, Templates and a Test Plan — Exam notes with an AI scribe and how to test one.
• Optometry Referral Letter Template: Draft With AI, Then Check — Referral letters drafted with AI, then checked.
• AI Receptionist for Opticians: Cost, Bookings and Frame Queries — Cost and call handling for an optician's front desk.
• Is AI Frame Advice and Virtual Try-On Worth It for Opticians? — Whether AI frame advice earns its cost.
• Can AI Handle Vaccination and Check-Up Reminders for a Vet? — Reminder logic from another clinical setting.
• Can AI Write a Tattoo Aftercare Card Without Medical Claims? — A tattoo aftercare card written with AI from your artist's own routine, with the wording limits that keep it clear of medical advice.
• Optometry Insurance Verification: A Front-Desk Workflow With AI — A front-desk routine for checking vision and medical eye insurance before the visit, with a call script, note template and safe uses of AI.
• AI in an Optometry Practice: A Week-by-Week First-Month Plan — A four-week plan for an optometry practice: phones, recalls, exam notes and marketing, with what to measure and what to leave out.
• GDPR in an Optical Practice: Which AI Tools May Touch Records — A scored 12-point checklist for vetting any AI tool before it touches eye-test records, with four tools worked through and the paperwork behind each.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Twilio's SMS character-limit glossary; public-health contact lens care guidance (keep lenses away from water, no sleeping in lenses unless directed); OpenAI and Anthropic project help and plan pages; your own practitioners' aftercare protocols and your recall software's documentation set the schedule.