Get a demo

The Canadian pharmacist's guide to fax automation

Fax is still how the majority of prescriptions reach a Canadian community pharmacy. On a busy day that means a stack of inbound pages, each one retyped by hand into your pharmacy management system, each retype a chance for a wrong DIN, a misread sig, or a missed refill authorization. Fax automation exists to take that retyping away. This guide is the checklist we wish every pharmacist had when evaluating it: what the technology actually does, where it goes wrong, and the questions that separate a real solution from a demo that falls apart in week two.

1. Start with workflow fit

The best tool is the one that works inside your existing PMS, not around it. For most Canadian pharmacies that PMS is TELUS Health Kroll, and the single most important question you can ask a vendor is: how exactly does your system get a prescription INTO Kroll?

There are only three honest answers, and they are not equal:

  • Native integration. The software talks to Kroll through a supported agent or API, writing prescriptions into the intake queue the same way Kroll itself would. It can survive Kroll updates because the integration point is stable, not visual.
  • RPA (screen-scraping bots). The software drives Kroll’s screens like a very fast human: click, paste, tab, click. It works in the demo and breaks the day a Kroll update moves a button. Someone then has to remap every screen, and that someone is usually billed to you.
  • It does not actually write to Kroll. Some “fax automation” products only digitize and sort the fax, then hand your team a nicer queue to retype from. That is document management, not automation. It has value, but it does not remove the data-entry labour, which is where the payback lives.

Ask the vendor which of the three they are. Then ask how they handle write queues and retries: what happens when Kroll is mid-update, the terminal is busy, or two prescriptions arrive for the same patient at once? A production system has an answer with a dashboard behind it. A demo does not.

2. Read before you write

Writing into Kroll is only half the job. The difference between an automation you trust and one you babysit is whether it can read context before it writes.

A system that reads Kroll’s patient profile and drug catalog before writing can check that the patient exists (and which of the three “J. Smith” profiles is the right one), match the prescribed drug to the correct DIN and strength from the catalog rather than guessing from OCR text, and flag a mismatch as an exception instead of writing a wrong entry that a pharmacist has to catch at verification.

OCR alone cannot do this. OCR turns pixels into text; it does not know that the text “Metformin 500” maps to a specific DIN, that the prescriber’s license needs checking, or that this patient already has an active therapy that conflicts. When a vendor says “AI”, make them show you the step between the fax image and the Kroll write. If the answer is “we extract the text and fill the fields”, you are looking at OCR with better marketing.

Numbers to ask for, and what good looks like in practice:

  • First-attempt DIN match rate on YOUR fax mix, not a curated demo set. Strong systems run in the 90s; be suspicious of “99%+” claims with no exception process, because the honest number depends on fax quality.
  • Exception rate and handling. Some prescriptions SHOULD fail automation: illegible handwriting, unusual compounds, missing information. What you want is a clean exception queue a human resolves in seconds, not silent wrong writes.
  • Time from fax arrival to Kroll entry. Minutes, not hours, and measured end to end.

3. Compliance is not a checkbox

A faxed prescription is personal health information. The moment a third party processes it, you need answers to all of these, in writing:

  • PHIPA and PIPEDA alignment, stated plainly, with the vendor’s role (agent or service provider to the custodian, which is you) spelled out in the agreement.
  • Data residency. Where do the fax images and the extracted health data live? For Canadian pharmacies the defensible answer is Canadian data centres, full stop.
  • Subprocessors. Which other companies touch the data (cloud host, AI provider, fax carrier), and will you be told when the list changes?
  • Audit logs. When the college or your own investigation asks who saw and wrote what, can the vendor produce the trail, and can YOUR team see it without filing a support ticket?
  • Retention and deletion. How long are the source fax images kept, and what happens to them when you leave the service?

None of this is exotic. A vendor operating properly in Canadian pharmacy has these answers pre-written. Hesitation is the tell.

4. Questions that expose weak vendors

Take this list into the demo call:

  1. Show me a fax with bad handwriting. What happens, step by step, and where does a human enter the loop?
  2. What happens when Kroll releases an update? Who fixes the integration and how fast, and is that in the contract?
  3. Can my pharmacists see every entry the system made this week, with the source fax beside it?
  4. What is the onboarding actually like: days or months, and what does my team have to do?
  5. What does it cost when my volume doubles in flu season, and is there a per-page surprise?
  6. Who else in Canada runs this today at my volume, on Kroll, and can I talk to one of them?

A good vendor enjoys these questions. That is also a signal.

5. What the payback actually is

The labour being automated is real and measurable: a typical community pharmacy spends staff hours every day on prescription intake typing, and that time comes from your most-interrupted people. Automating it does three things at once: it returns assistant and technician hours to the counter, it shortens the time a patient waits between “the doctor faxed it” and “it is ready”, and it converts data-entry errors from something a pharmacist must catch into something the system flags before entry. When you model payback, count all three, not just the wage hours.

Frequently asked questions

Does fax automation replace pharmacy staff? No. It replaces the retyping portion of intake. Verification remains a pharmacist’s clinical act, exceptions still need human judgment, and the freed hours go to the clinical services that actually grow a pharmacy.

We also get digital prescriptions. Does the same automation cover those? It should. Fax is the dominant channel but a good intake system treats faxes, scans, and digital Rx as one queue with one audit trail. Ask the vendor to show both paths.

How long does implementation take? With a native Kroll integration, days. If a vendor quotes months, the integration is heavier than they are letting on, or the “integration” is a person.

Is it safe under PHIPA to let AI read prescriptions? It can be, when the processing happens under the custodian’s authority with Canadian data residency, proper agreements, and audit trails (see section 3). The compliance question is not “AI yes or no”, it is the same question you would ask of any processor handling PHI on your behalf.

What does it cost? Pricing models vary by vendor and volume. What matters is that the model is flat and predictable rather than per-page metered, and that a pilot lets you verify accuracy on your own fax mix before you commit.

When you’re ready

Book a demo and we’ll map your real fax and digital volume to an onboarding plan, inside a Kroll sandbox, using your own workflow. Prefer to read further first? See how fax intake works and the Kroll integration details.