Clinical Trial Technology Doesn't Fail – It Fails At The Handoff

By John Oncea, Chief Editor, Clinical Tech Leader

Much has been written about which platform has the better dashboard, the cleaner integration story, or the smarter AI layer. But after talking with Tinatin Jashi, I’m convinced we’ve been asking the wrong question. The issue isn’t whether a given system works. It’s what happens the moment information leaves that system and lands in someone else’s hands.

Jashi is unusually positioned to make that case. Over nine years, she’s worked as a study coordinator, clinical trial assistant, CRA, clinical project manager, medical writer, cognitive debriefing specialist, and regulatory affairs specialist, often on the same trials, just at different points in their lifecycle. Most people in clinical research build depth in one or two of those roles. Jashi has stood in nearly all of them, watching the same protocol, the same data point, the same deadline looks completely different depending on who’s holding it.

“Clinical trial technology is not only about systems, features, or dashboards,” she told me. “The real question is whether technology supports the way information, responsibility, and decisions move between people.” A tool can be well-built and still create friction if it ignores the pressure on site staff, the sponsor’s expectations, or what regulatory and medical writing teams will need six months later.

Where The Chain Breaks

Ask Jashi where clinical trial technology does the most damage, and she doesn’t point to a specific platform. She points to a transition: the moment a sponsor’s strategy becomes a CRO’s execution plan, and that plan becomes a site’s daily reality.

“The site is where all of this becomes practical: patient availability, physician engagement, institutional limitations, regional culture, workload, staffing, and competing clinical priorities,” she said. “If feasibility and capability assessment do not capture that reality, the handoff is already weak before the first patient is enrolled.”

That’s a blunt way of saying something the industry doesn’t like to admit: a lot of downstream technology failure is actually an upstream assumption failure. Feasibility assessments built on optimism rather than operational reality don’t get corrected by better software later, they get inherited by it. “This is why feasibility can sometimes feel like a lottery,” Jashi said. Win it, and the sponsor saves time. Lose it, and that savings comes back later as recruitment shortfalls, deviations, and escalations no dashboard fully explains.

The Integration Promise Vs. The Integration Reality

If there’s one word that sells clinical trial technology, it’s “integration.” EDC talks to eTMF, which talks to CTMS, which feeds safety systems, and everyone gets a single source of truth. Jashi’s read on that promise, based on what she’s actually seen in the field, is more measured.

“Many clinical trial systems are strong within their own function, but they are not always meaningfully connected across functions,” she said. “We may have digital systems, but we do not always have a truly digital process.” The workflow connecting those systems, in her experience, is still frequently maintained manually, by people copying, checking, and reconciling across platforms that were never designed to speak the same language.

The deeper problem is interpretive, not technical. The same data point can mean five different things depending on who’s looking at it. “A data point can be a query issue for data management, a compliance or source verification issue for the CRA, a potential reporting concern for safety, a timeline or escalation risk for the project team, and another practical task for the site during an already busy clinical day,” Jashi explained. Integration moves the data. It doesn’t automatically move the meaning or clarify who owns what happens next.

Digitizing A Broken Process Doesn’t Fix It

Jashi is also candid about a pattern she’s watched play out repeatedly: taking a manual process and simply making it electronic, without asking whether that process was sound in the first place. “A broken manual process does not become better because it is digital,” she said. “Sometimes digitalization only makes the broken process faster, more expensive, and harder to see.”

Her caution isn’t theoretical. She’s seen digital workflows that functioned well on paper until a technical issue hit a site with no backup plan, no offline fallback, no contingency for a delayed vendor response. The problem got solved, she said, because people at the site and CRO level acted proactively. But it shouldn’t have required improvisation. “The delay could have been avoided with a simple approved backup process.”

What Sites Actually Need

Ask a coordinator what would make their day easier, and the answer is rarely “more technology.” Jashi, who has done that job, says what sites need most is simplicity: fewer logins, fewer redundant trainings, tools sized to the actual patient volume they’re managing. “Sites do not necessarily need more technology,” she said. “They need technology that is combined, user-friendly, intuitive, and proportional to the actual study workload.”

She also raises a point sponsors rarely put in a budget line: the time sites spend learning and maintaining access to a dozen separate systems is real work, and it’s frequently uncompensated. If sponsors want genuine engagement from sites, she argues that workload needs to be acknowledged, not assumed as a courtesy.

Downstream, The Bill Comes Due

By the time a data point reaches regulatory affairs or medical writing, the decisions that shaped it – protocol design, deviation documentation, how a site worked around a system limitation – are long since made. “Regulatory and medical writing outputs are only as strong as the upstream process that produced them,” Jashi said. When those upstream decisions weren’t made with documentation and inspection readiness in mind, the result is amendments and rework driven not by science but by workflow gaps that could have been caught earlier.

That’s also where she sees the real risk in AI-assisted regulatory writing. AI can generate a polished, coherent document quickly. What it can’t do is fix a fragmented upstream process on its own. “If upstream workflow problems are not addressed, AI-based tools may make document generation faster, but they will not necessarily make the underlying content more accurate, more meaningful, or more inspection-ready,” she cautioned. A clean-looking document built on incomplete or poorly contextualized information is, in her words, an illusion of efficiency, one that tends to surface its costs later, during audit or regulatory review.

Starting With Reality, Not The System

When I asked what she’d tell a sponsor serious about fixing this, Jashi didn’t start with a procurement checklist, she started with observation. “Clinical trials do not fail in dashboards first,” she said. “They fail earlier, in unclear ownership, unrealistic expectations, weak communication, and workflows that do not reflect site and patient reality.” Her recommendation: understand the process before digitizing it, identify where ownership is unclear, and only then decide what technology should support.

It’s a sequence the industry tends to reverse, and Jashi’s cross-functional vantage point is exactly why she notices it. She’s lived on both ends of enough handoffs to know that the seam between functions, not the functions themselves, is where clinical trials quietly go wrong. That same dynamic, as it turns out, is also what shapes the patient experience with eCOA and other digital tools further downstream, which is where we turn next.