Your Optometry CRM Should Track Prescriptions, Not Just Patients

Most eye care practices in Allen County run on two systems: an EHR that holds the clinical chart, and a scheduler that holds the calendar. Between them sits a set of facts nobody really owns — when a prescription was released, whether the patient signed for it, when it expires, and whether anyone has reminded them since.

In June 2025 the FTC sent 37 warning letters to contact lens prescribers over possible violations of its Contact Lens Rule. All of them came out of consumer complaints. The letters weren't findings of guilt, but they were a clear signal that the agency reads complaints and acts on them, and that violations can carry civil penalties of up to $53,088 each.

The underlying rules got stricter not long before that. The revised Eyeglass Rule took effect September 24, 2024. Since then, when you complete a refractive exam, you release the prescription and obtain signed confirmation that the patient received it — and you keep that confirmation for at least three years. Deliver electronically instead and you need verifiable affirmative consent on file first, retained for the same three years. The Contact Lens Rule works the same way.

That is not a clinical record. It's an operational one, with a retention clock attached. And it's exactly the kind of thing an off-the-shelf CRM has no field for.

Build the model around the prescription

The instinct is to put Patient at the center of the system. For an optometry practice, that's one level too high. The thing with an actual lifecycle — a start date, a clock, and an ending that should trigger work — is the prescription.

A data model that matches how the office really runs looks closer to this:

• Patient — contact info, insurance, household links so a parent booking for three kids is one conversation
• Exam — date, doctor, type: routine refraction, contact lens fitting, medical follow-up
• Prescription — a child of the exam, typed glasses or contacts, carrying an issue date and a real expiration date. Ohio expires a contact lens prescription one year after the exam and completed diagnosis under normal circumstances, and lets the optometrist shorten that based on ocular health, so expiration is a stored field, not an assumption
• Release record — how the prescription was handed over, the signed acknowledgment or the consent to email, and a retention date three years out
• Order — glasses or contacts, with stages: submitted, at the lab, arrived, dispensed

Once Prescription is a first-class record with a date on it, the automations write themselves. Ninety days before expiration, the CRM opens a recall task. Sixty days out, it texts. Thirty days out, it flags a human to call. Someone who bought a year of contacts in March gets a reorder nudge in January, not a generic postcard in June because that's when the postcards went out.

What the build looks like

Paglow's Custom CRM engagements start with a spec: a written description of your pipeline, your entities, and every automation, in plain language, before anyone builds anything. Then the data model — fields, relationships, and the rules that govern them. Then a build plan with sequencing, so the prescription-and-recall piece goes live first and order tracking follows, instead of one nine-month project that lands all at once.

The AI sits inside that model rather than bolted on top. Exam notes and lab emails get parsed into structured fields so your staff stops retyping the same data twice. Recall lists get ranked by who is actually likely to book — the patient two weeks past expiration who has come in four years running is a very different call from the one who no-showed twice. Draft recall messages arrive pre-written with the patient's prescription type and last order attached, and a person sends them.

Here's the practical test. Pick a patient at random and ask your current setup three questions: When does their contact lens prescription expire? Do we have their signed acknowledgment, and can we produce it? Where is their last order right now? If that takes more than a few seconds, or means opening three programs, the problem isn't your team. It's that nothing in the office was designed around the questions the office actually asks.