1. Why feature checklists fail
Two products can list identical modules and behave completely differently, because the question is not which boxes exist but whether they share a record. A checklist cannot see that. It records that both products have scheduling, charting, payments and inventory, and scores them equal.
What separates them is what happens between the boxes. When a provider signs a prescription, does the vial come out of inventory, does the card on file get charged, and does the patient get told — or does a staff member do three of those four by hand? Both products truthfully claim prescribing and inventory. Only one of them connects them.
So the criteria below are written as things a vendor must demonstrate or commit to, not things it can claim. Every one has an evidence column, and the evidence is the point.
2. The twelve criteria, with the evidence to demand
Weight them before you see any demo. Weighting afterwards is how a clinic talks itself into the product with the best presentation. Weights below are a reasonable default for a prescribing cash-pay clinic; change them to fit, but change them first.
| # | Criterion | Evidence to demand | Weight |
|---|---|---|---|
| 1 | One record, not several | Watch one real event move through charting, inventory, payment and the patient app in a single demo without a hand-off | 3 |
| 2 | Prescribing that fits your drugs | Compose a script for something you actually prescribe, including a compounded item, during the demo | 3 |
| 3 | Controlled substances | See identity proofing and two-factor signing, and ask who may sign under whose identity | 3 |
| 4 | What the price is at your size | A written quote at your prescriber and staff count, including year two | 3 |
| 5 | Your data on exit | A written answer on format, scope, timing and fee — before you sign, not after | 3 |
| 6 | Payments and the money trail | See a charge land against a visit, an invoice and a membership in one ledger | 2 |
| 7 | Migration from your current system | What they have moved before, what they will not move, and who does the work | 2 |
| 8 | Labs, if you order them | An order placed and a discrete result plotted, not a PDF in an inbox | 2 |
| 9 | The patient’s side | Open the patient app and do the three things your patients do most | 2 |
| 10 | Access control and the audit trail | Two roles side by side, and an audit entry including a denied attempt | 2 |
| 11 | The BAA and where data sits | A signed BAA offered as standard, and a straight answer on hosting | 2 |
| 12 | Who answers when it breaks | A named channel, a stated response time, and who you reach at 7pm | 1 |
Score each one out of five on what you were shown, multiply by the weight, and total it. A product that cannot be scored on a criterion scores zero on it, not a blank — an unanswered question is information.
3. The four questions that decide most purchases
If you only ask four things, ask these. In our experience they separate products faster than any demo, because each one is answered with a document rather than a screen.
| Question | Why it separates products | What a good answer looks like |
|---|---|---|
| What does this cost at my size in year two? | Pricing models diverge as a clinic grows, and growth is the plan | A written quote with the model stated, not a starting price |
| What do I get on the day I leave, in what format, and what does it cost? | It is the only question a vendor has no incentive to answer well | A written commitment, agreed before signature |
| Show me one event moving through four modules | Separates a connected platform from a bundle of them | A live demo on their own system, not slides |
| Who signs the BAA and when? | You are the covered entity; this is not optional | A standard BAA, offered without being chased |
The exit question deserves a note. It is uncomfortable to ask during a sales process, which is exactly why it is informative — and a federal right to your electronic health information exists independently of what a contract says. We wrote about where that right applies and where it does not in what it costs to leave your EHR.
4. How to run the demo so it tells you something
Bring your own data and your own worst day. A demo on the vendor’s tidy sample clinic tells you the software works under ideal conditions, which was never in doubt.
| Bring | Because it reveals |
|---|---|
| Your three most common visit types | Whether the workflow matches yours or theirs |
| A drug you prescribe that is awkward | Whether the drug database covers your actual practice |
| Your real prescriber and staff counts | What the price is, rather than what it starts at |
| A messy patient | Allergies, a stopped protocol, an outstanding balance |
| The member of staff who hates change | The person whose objection will decide adoption |
Ask the vendor to drive as little as possible. A product that only looks good when its own staff are at the keyboard will look exactly that good in your clinic, which is to say not very.
5. What should worry you
None of these are disqualifying on their own. Two together usually are.
| Sign | What it often means |
|---|---|
| Pricing only on a call | The number depends on what they think you will pay |
| A roadmap answer to a feature question | It does not exist, and the timeline is not a commitment |
| “HIPAA certified” | No such certification exists; it suggests nobody internally corrects this |
| Reluctance to put the exit in writing | The leaving will be worse than the arriving |
| A demo that never leaves the sample clinic | Your data may not fit the shape they are showing you |
| Modules that each need their own login | The seams are where the work and the mistakes live |
The third one is worth stating plainly because it is widespread. There is no such thing as a HIPAA-certified product. A vendor can sign a business associate agreement and engineer to the requirements — which is what we do and what you should expect — but certification is not a thing that exists to be held.
6. Turning the sheet into a decision
Score every product on the same day, from the same evidence, with the weights you set before you started. Then do one thing that feels redundant and is not: write down, in a sentence, what would have to be true for the loser to have won.
That sentence is the most valuable artefact of the whole process. In six months, when something irritates you about the product you chose, it tells you whether you traded for it knowingly or missed it. Clinics that keep it rarely regret the decision; clinics that do not tend to relitigate it every time something goes wrong.
And treat a tie as a real outcome. If two products finish within a few points, the criteria did not separate them and more demos will not either. At that point decide on the two things no sheet captures well: who you would rather telephone when something is broken, and whose exit terms you would rather be bound by.
7. Frequently asked questions
How should a clinic evaluate EHR software?
Weight your criteria before any demo, score each one on evidence the vendor demonstrates or commits to in writing rather than on claims, and insist on seeing one real event move through several modules without a manual hand-off.
What are the most important criteria when choosing clinic software?
A single shared record, prescribing that covers the drugs you actually use, controlled-substance handling, the real price at your size in year two, and a written commitment on what you get when you leave.
What should I ask an EHR vendor in a demo?
What it costs at your size in year two, what you receive on exit and in what format, a live demonstration of one event crossing four modules, and who signs the business associate agreement.
Is any EHR HIPAA certified?
No. There is no HIPAA certification for software. A vendor can sign a business associate agreement and build to the requirements, but a product cannot hold a certification that does not exist.
How many vendors should a clinic shortlist?
Three is usually enough to calibrate. Beyond that the demos blur together and the decision tends to go to whichever was seen most recently rather than whichever scored best.