The classic case-processing chain is right-heavy: intake, data entry, duplicate check, MedDRA coding, seriousness and expectedness, causality, narrative, medical review, QC, submission. Quality sits at the end of the line, where QC and medical review clean up what arrived dirty.
A shift-left model moves the controls to where the data enters: E2B(R3) gateways, OCR and language models on unstructured sources, validation while the record is created, automated coding, probabilistic duplicate detection, completeness checks and a first draft of the narrative. QC becomes quality by design rather than quality by inspection.
The logic is not new. Quality professionals have long used the rule of thumb that preventing a defect costs 1, fixing it internally costs 10 and absorbing its consequences externally costs 100. In pharmacovigilance the scale is concrete. A missing field caught at intake costs one question to the reporter. The same gap found in medical review costs rework. Found after submission, it costs an amendment, a late report or an inspection finding. What is new is that we can finally apply the logic at scale, and the market has noticed: on 1 September, Veeva announced that Falcon Safety will work with any E2B-compliant safety system across any intake channel. Intake and the safety database used to be bought together. The front door is becoming a product category of its own.
One control deserves more attention than it gets: the follow-up question on day zero. A case has the least context at the moment it arrives, but that is also when the reporter is still reachable and still remembers. An intake that can generate the right follow-up question immediately is worth more than ten checks downstream.
