Every vendor will hand you references, and every reference on that list is there because they are happy. That does not make reference calls useless. It makes the questions matter more. A curated reference will still answer honest questions honestly; the trick is asking questions that have textured answers instead of yes-or-no ones.
Here is the script we would use, and why each question earns its spot. It is the last diligence step in a switching evaluation, after demos and before signature.
First: ask for the right references
Before the call, shape the list. Ask the vendor for references at your level of care and your organization size. A glowing reference from a 12-bed residential program tells a 200-client outpatient group very little, and vice versa. If a vendor cannot produce a reference that looks like you, that is itself an answer about their customer base.
If you can, also find one customer the vendor did not hand you. A peer at a conference, a colleague from a previous role, a name from a webinar. One uncurated conversation is worth three curated ones.
The script
1. What broke in the first 90 days, and how fast was it fixed?
The most important question on the list. Every implementation has a first-90-days story, and "nothing broke" means you have not gotten past the pleasantries yet. What you are measuring is not whether something broke, it is the vendor's response: how it was triaged, who owned it, and how long it took.
2. Did billing output hold up through the transition month?
Revenue is where switches get scary. Ask directly whether claims kept going out and cash kept coming in through cutover, and how long payer enrollment actually took versus what was quoted. The gap between quoted and actual enrollment time is one of the most honest numbers a reference can give you.
3. How many implementations was your implementation manager carrying during yours?
You asked the vendor this in the demo. Now check the answer. A reference who says "our IM was great but clearly stretched" has just told you more than the sales deck did.
4. How long do form or workflow changes take now that you are live?
Demos show configuration speed with the vendor driving. Live customers know the real turnaround: what they can change themselves, what needs a ticket, and how long tickets take. This is the question that predicts your team's daily experience two years from now.
5. How does support actually work when something goes wrong?
Not the support model, the support experience. Who do they contact, how fast is first response, and do billing questions get bounced between teams? Ask for a recent example.
6. What do you know now that you wish you knew before signing?
The open door. References often save their most useful observation for the question that invites it. Let the silence sit until they fill it.
7. Would your clinicians go back to the old system?
End here. Not "are you satisfied," which invites politeness, but a concrete counterfactual. Clinician sentiment is where EMR satisfaction actually lives, and a reference who laughs at this question has told you everything.
Reading the answers
Three patterns to weigh as you compare calls across vendors:
Specificity beats enthusiasm. A reference who describes a real problem and a real resolution is more reassuring than one who describes nothing but delight. Textured answers are how you know you are hearing experience rather than loyalty.
Consistency across calls. One reference's complaint is an anecdote. The same complaint from two references is a pattern, and it will be your pattern too.
The vendor's answers, verified. Every claim from the demo that a reference can check, check: implementation staffing, support response, change turnaround, enrollment timelines. The deltas are your risk map.
