But Keiko has been doing this work for seventeen years. She knows that a high completeness percentage can be either a sign of excellence or a symptom of dysfunction. The question is not what the number says. The question is what the number means.

She opens the detailed view for Site 023 at Westbrook Clinical Research in Portland, Oregon. The site shows 97% completeness, among the highest in the study. Yet when she drills into the data, something feels wrong. The site activated fourteen months ago and has enrolled 34 participants. They have completed six interim monitoring visits. But she notices that the expectation model shows only 312 expected documents while comparable sites with similar enrollment show expectations above 400.

Keiko picks up the phone and calls Thomas Henderson, the clinical operations lead for MERIDIAN-3. "Thomas, I am looking at Site 023. Can you walk me through how we set their document expectations?"

Twenty minutes later, Keiko understands the problem. When Site 023 was set up in the eTMF, the expectations were configured based on a template for a smaller site type. Safety reports that should have been expected were not flagged. Several monitoring visit report categories were missing from the expectation model. The 97% completeness rate was accurate against the configured expectations, but those expectations were wrong. The site was actually missing dozens of documents that no one knew should be there.

"We have been congratulating this site for outstanding TMF quality," Keiko says. "In reality, we have no idea what their actual completeness is."

This lesson is about what Keiko learned that afternoon: that completeness metrics are only as meaningful as the expectations they measure against. A TMF quality program that focuses solely on achieving high percentages without validating that expectations are correct, complete, and appropriate is building on sand.

But Keiko has been doing this work for seventeen years. She knows that a high completeness percentage can be either a sign of excellence or a symptom of dysfunction. The question is not what the number says. The question is what the number means.

She opens the detailed view for Site 023 at Westbrook Clinical Research in Portland, Oregon. The site shows 97% completeness, among the highest in the study. Yet when she drills into the data, something feels wrong. The site activated fourteen months ago and has enrolled 34 participants. They have completed six interim monitoring visits. But she notices that the expectation model shows only 312 expected documents while comparable sites with similar enrollment show expectations above 400.

Keiko picks up the phone and calls Thomas Henderson, the clinical operations lead for MERIDIAN-3. "Thomas, I am looking at Site 023. Can you walk me through how we set their document expectations?"

Twenty minutes later, Keiko understands the problem. When Site 023 was set up in the eTMF, the expectations were configured based on a template for a smaller site type. Safety reports that should have been expected were not flagged. Several monitoring visit report categories were missing from the expectation model. The 97% completeness rate was accurate against the configured expectations, but those expectations were wrong. The site was actually missing dozens of documents that no one knew should be there.

"We have been congratulating this site for outstanding TMF quality," Keiko says. "In reality, we have no idea what their actual completeness is."

This lesson is about what Keiko learned that afternoon: that completeness metrics are only as meaningful as the expectations they measure against. A TMF quality program that focuses solely on achieving high percentages without validating that expectations are correct, complete, and appropriate is building on sand.