The clearest sign that a practice is about to replace its electronic health record is almost always a tiny excel spreadsheet.
Walk into almost any clinic that has soured on its software and you’ll find one. Sometimes it lives on the biller’s desktop. Sometimes it’s a shared sheet the front desk updates between patients. It tracks what the system was supposed to track anyway, such as, which prior authorizations are still pending, which claims bounced and why, who needs a callback before Friday. Nobody approved it. Staff just built it, because the software couldn’t do the job and arguing about that had stopped being productive.
That spreadsheet is a diagnosis. Once it appears, the practice usually has about eighteen months left with that vendor. But the decision that produced it was made long before anyone opened Excel. It was made in a forty-five minute sales demo, years earlier, when nobody in the room knew what to ask.
Quick answer
Most clinics don’t switch EHR systems because the software broke. They switch because the system was evaluated against a rehearsed demo instead of against their own worst week. So, the fix is developing a harder purchase process, plus contract terms that make leaving cheap before you ever need to leave.
How common is the three-year churn, really?
More common than vendors like to admit, and the numbers hold across practice sizes.
Black Book Market Research has tracked practice dissatisfaction for over a decade, and its surveys have consistently found roughly 23% of practices unhappy enough with their current system to consider moving to a different vendor. Discontent clusters are hard by specialty. Nephrology, urology, and ophthalmology have reported dissatisfaction rates in the 80% range, while internal medicine, family practice, and pediatrics sit near the opposite end. That gap tells us that most systems were designed around primary care workflows, and everyone else has been adapting ever since.
Satisfaction isn’t great even in the friendly specialties. In one national survey of family physicians, only 27.2% said they were very satisfied with their EHR, and higher satisfaction tracked with lower burnout. Smaller and rural organizations are restless too. Black Book’s rural outlook found nearly half of critical access hospitals intending to reassess their systems by 2026, citing functionality, cost, cybersecurity, and workflow fit.
Then there’s the part nobody budgets for. Productivity commonly drops somewhere between 10% and 40% in the weeks after cutover, and in some settings the disruption stretches on for months. For a three-provider clinic with no IT department and no project manager, that dip takes a heavy toll on the payroll.
The switch was decided during the demo
Now, let’s move on to the part most write-ups on this topic skip entirely. They treat switching as a migration problem and hand you a checklist for the transition. But by then you’ve already lost the money.
A vendor demo is a performance. It runs on clean seed data, a cooperative patient, an uncomplicated payer, and a sales engineer who has driven that exact click path several hundred times. Nothing in it resembles a Tuesday in November with a double-booked schedule, a fax machine full of records, and a medical assistant who started last week.
So stop watching the demo and start breaking down the nitty gritty. A few requests that reliably separate real software from a good deck are:
- Ask them to load one of your genuinely ugly charts. A twelve-year patient history with scanned outside records, three active specialists, and a medication list nobody has reconciled since 2019.
- Ask to see a denied claim end to end. Not the clean submission. The denial, the work queue it lands in, who touches it next, and how many clicks it takes to correct and resubmit.
- Take the mouse away from the sales engineer. Let your own front desk person attempt a new patient registration with insurance verification, unaided, for ten minutes. Watch where they stall.
- Ask what happens when the internet goes down mid-clinic, and get the answer in writing.
- Ask for three reference customers in your specialty who are within their first year. Not the polished five-year advocates. The ones still in the hard part.
- Ask which features on the roadmap slipped last year, and why.
Any vendor worth signing with will handle these calmly. The ones who deflect toward a follow-up call with a solutions architect have told you what you needed to know.
If you’re switching for the second time, the software probably isn’t the problem
This is uncomfortable but worth saying plainly. Practices that replace their system twice in five years are usually not unlucky. They’re misdiagnosing their problem. For instance…
- Slow charting is frequently a template configuration issue, not a platform issue.
- Rising denials often trace back to front-desk eligibility capture rather than the billing module.
- Staff resistance is sometimes a staffing problem.
When a practice migrates without ever isolating the actual failure, it simply reproduces the same problem inside a new interface, having paid five figures and lost a quarter of its productivity for the privilege.
Therefore, the honest question before any evaluation begins should be, can we name the specific workflow that’s broken, without using the vendor’s name? If the answer is no, the search hasn’t started yet. Fix the diagnosis first. This is the same discipline that’s reshaping other back-office corners of healthcare, where the operational bottleneck usually turns out to be compliance and process rather than the matching tech everyone assumed.
What changed in 2026: the AI trigger
There’s a newer reason practices are heading for the exit, and it’s worth naming because it’s producing a lot of hasty decisions right now.
Ambient documentation went mainstream fast. By 2026, 70% of physicians in the UCSF health system were using AI scribes daily, and Kaiser Permanente clinicians had run more than 2.5 million encounters through them over fourteen months. Another multicenter study reported burnout falling from 51.9% to 38.8% over thirty days. Meanwhile more than 70% of large delivery networks have deployed or are piloting an ambient scribe, and independent practices have followed as prices dropped.
Naturally, physicians at clinics without it want it.
Naturally, some are willing to change platforms to get it.
That’s where the mistake happens. Ambient AI is increasingly available as an integration rather than a reason to rip out your core system, and the honest first question is whether your current vendor supports one. Replacing an entire records platform to acquire a documentation feature is an expensive way to solve a solvable problem.
Negotiate your exit before you sign the entrance
If you take one practical thing from this piece, take this…
Federal rules already moved in your favor and most practice owners don’t know it. Under the 21st Century Cures Act, certified health IT that stores electronic health information has been required since the end of 2023 to support EHI export for both a single patient and the full patient population. Critically, fees charged to export data for the purpose of switching systems, when performed through that certified capability, do not qualify for the information blocking Fees Exception. In plain terms, your vendor generally cannot bill you for leaving.
Portable is not the same as usable, though, and this is where practices get ambushed. The rule only requires that exports be computable and documented, so nowadays some vendor exports arrive as more than 6,500 distinct tab-separated table formats with separate downloads for documents and images, while others ship newline-delimited JSON and HTML. That means, you will receive your data. You may then need to pay someone competent to make it mean anything.
So write your exit into the contract on day one, while you still have leverage:
- Confirm in writing which certification the export runs through, and ask for a sample export file before signing.
- Fix the notice period and any conversion assistance rate now, not during a dispute.
- Get clarity on who holds your data if the vendor is acquired, which happens constantly in this market.
- Ask how long historical records stay accessible after termination.
Vendors that build around smaller independent clinics tend to answer these questions directly, partly because platforms serving one to five provider practices compete on transparency rather than switching costs. Enterprise contracts are a different beast altogether.
The three habits that actually prevent the three-year switch
Practices that stay put for a decade tend to do three unglamorous things.
First, they configure deliberately in the first ninety days, rather than living permanently with default templates built for somebody else’s specialty. Second, they assign one internal owner who actually knows the system well enough to fix small annoyances before those annoyances become a shadow spreadsheet. And, third, they re-evaluate the fit annually against measurable things, like days in accounts receivable, first-pass claim acceptance, and time from encounter to signed note, instead of relying on how everyone happens to feel that quarter.
Worth adding, much of what physicians blame on their records system is really scheduling, eligibility, and collections. Which means the honest evaluation often belongs to practice management software for medical practices rather than the clinical documentation layer everyone stares at all day.
None of this is dramatic. That’s rather the point. The clinics that avoid the three-year churn aren’t the ones that chose a perfect vendor. They’re the ones who asked harder questions before signing, diagnosed their own problems honestly, and made leaving cheap enough that they never felt trapped into staying.
