You've seen the dashboards. Utilization climbing. Member satisfaction in the 90s. Cost-per-visit trending down. The ROI spreadsheet looks like a work of art. But there's a problem nobody in the benefits world is talking about: we're measuring the wrong thing.
The entire industry evaluates telehealth based on isolated episodes: a single 15-minute video visit, a quick prescription, a satisfied member. Each one gets counted as a win. The bigger picture stays hidden.
The Ghost Chart problem
In traditional primary care, your doctor sees the whole person. They know your A1c is creeping up, you missed your mammogram, and you recently lost your job. They connect the dots. Now consider a typical telehealth encounter:
A member logs on for a sinus infection. The provider prescribes an antibiotic in eight minutes. The member rates the experience five stars. The health plan saves $50 compared to an urgent care visit. What did we miss?
- Was that sinus infection a side effect of a new immunosuppressant prescribed by a specialist last week?
- Is this the third upper-respiratory visit in six months, signaling undiagnosed sleep apnea or diabetes?
- Did the member mention anxiety in passing, but the visit was coded as acute sinusitis?
The ghost chart is the patient's real medical history: the one that lives on the PCP's EMR, the specialist's system, and the pharmacy database. It never touches the telehealth platform. When we evaluate telehealth effectiveness solely on the episode, we celebrate the transaction while ignoring the fragmentation. One caveat keeps this fair: telehealth delivered by a member's own health system or primary care group, where the visit lands in the same EMR the PCP already uses, doesn't carry this risk. The problem concentrates in standalone, episodic vendors whose chart never leaves their platform.
Cost avoidance vs. cost attribution
This is where the model breaks down. The standard ROI model for telehealth assumes that avoiding one ED visit is a net positive. But what about the downstream costs of poor coordination? A concrete example:
- Telehealth visit: Member uses a vendor for a sore throat. Prescribed steroids.
- Result: Steroids spike blood glucose.
- Outcome: Member, a diabetic, ends up in the ER for hyperglycemia two weeks later.
- Accounting: The ER visit is charged to the medical bucket.
- The reality: The telehealth vendor's report shows a "successful diversion." The medical carrier sees a $5,000 ER claim. No one connects the dots.
From a systems architecture perspective, the health plan is paying for a telemedicine solution that is functionally independent from its care management and disease management platforms. Until the data flows in a closed loop, we're not evaluating telehealth effectiveness. We're evaluating digital triage effectiveness: a much lower bar.
A new framework: The 3-Point Integration Score
To evaluate a telehealth solution, set aside the vendor's dashboard and open your own claims data and member behavior after the visit. One question leads: did this visit leave the member's care more coordinated, or less? Three numbers answer it. Few plans have measured any of them, so treat the first pass as a baseline rather than a target.
1. The Care Disruption Rate
What it is: The percentage of telehealth visits that resulted in a break in continuity of care: a canceled PCP visit, a medication change that contradicted a chronic care plan, or a failure to follow up on a specialist referral.
Why it matters: If a telehealth visit prevents a diabetic from seeing their PCP for a routine check-up, you've increased long-term risk, even if you saved $50 that day.
What to look for: Can the vendor schedule a follow-up within the member's medical home? Or do they create an isolated, disconnected chart?
2. The Diagnostic Churn Rate
What it is: The rate at which a telehealth visit leads to a new misdiagnosis or a redundant test or visit within 30 days.
Why it matters: Classic example: prescribing a Z-Pak for a viral infection. The member feels worse in three days, sees their PCP, gets the correct diagnosis, and incurs a new copay. The system paid twice. The pattern is measurable: a University of Pittsburgh study highlighted by the National Institutes of Health found antibiotics prescribed for respiratory infections at 52% of telemedicine visits, compared with 31% of primary care visits.
What to look for: A high diagnostic churn rate means the vendor is prioritizing "member satisfaction scripts" over clinical accuracy, and that hurts your medical loss ratio.
3. The Data Return on Investment
What it is: Does the vendor push structured data (not a PDF) back into your health plan's core health information exchange or care management platform? Or do they hoard it?
Why it matters: A vendor that refuses to provide a Continuity of Care Document or an HL7 FHIR feed is creating a silo. Your wellness coach, your disease management nurse, and your analytics team can't see what happened in that visit.
What to look for: The vendor is generating intelligence about your population. If that intelligence stays locked in their platform, they are stealing data from your population health strategy.
Data return is now a compliance question
Until recently, a telehealth vendor that refused to share a structured feed was a nuisance. Today it is closer to a regulatory liability. The information blocking rules that took effect in April 2021 under the 21st Century Cures Act prohibit health IT developers, networks, and providers from interfering with the access, exchange, or use of electronic health information. Enforcement has caught up. In June 2023, HHS's Office of Inspector General finalized penalties of up to $1 million per violation for developers and health information networks, and in July 2024 CMS and ONC published disincentives for providers who block data. The network itself has scaled: TEFCA, the national interoperability exchange, moved from roughly 10 million exchanged records in January 2025 to nearly 500 million by February 2026.
For a benefits leader, this changes the QBR question. When a vendor cannot return a Continuity of Care Document or a FHIR feed into your systems, you are looking at a data practice that may sit on the wrong side of federal interoperability rules. That third metric becomes a screening question: a vendor that cannot show live, structured exchange with your claims and care management platforms is a compliance risk you can spot before the contract is signed.
What to ask in your next QBR
Replace the standard dashboard with these requests:
- "Show me the percentage of visits that resulted in a closed-loop referral to a PCP in our network."
- "Show me the specific diagnosis code for the visit and the pharmacy claim generated. Now show me if that same member has a primary care relationship. Did the PCP agree with the treatment?"
- "What is your data integration protocol with our claims administrator? Can your platform update a member's risk profile in real time, or does it sit in a separate data lake?"
What to measure instead
Telehealth only pays off when it feeds the rest of the care system. Years of vendor scorecards have graded the speed of the connection and the polish of the member app, while the visit's data evaporated at the end of the call.
The best telehealth vendor is the one you stop noticing, because its notes, diagnoses, and referrals land in your care management and claims platforms without friction. What it leaves behind is a coordinated, documented member journey.
Track that handoff. It will show you the real ROI.