CONSULTATION WRITE BACK / EMIS WEB · NHS PRIMARY CARE
UX Designer · EMIS Web · discovery to three-phase rollout
For PCN Hub clinicians, receiving GPs and practice managers
THE WORKFLOW PROBLEM
When a consultation failed to write back to the patient’s registered GP practice, users got no clear signal. The sender assumed it had arrived. The receiver could not tell what was missing or which consultations came from elsewhere.
WHY IT MATTERED
A quiet interface created false confidence, which makes this a patient safety and continuity-of-care risk, not only a usability one. Manual workarounds filled the gap: calls, emails and printouts.
The product looked calm when the system was broken. Users read silence as success.
MY OWNERSHIP
I owned: the reframing of write-back as a clinical safety risk, the phased strategy, the provenance indicator, the resend patterns and the partner configuration flow.
With the team: Product and Engineering aligned scope within PI delivery constraints. Evidence came from focus groups, support tickets, heuristic review and workflow analysis.
THE DECISION I CAN DEFEND
Visibility first, then recovery, then configuration.
Without visibility, resend becomes blind retry, so I argued for sequencing: show what failed, then let users recover it. I chose manual resend with visible status over silent auto-retry, because auto-retry would hide persistent failures and recreate the original silent-failure problem.
Also rejected: a real-time sync rebuild, which needed architecture changes outside PI scope; full integration of all 15 to 20 partner apps in one phase, because contracts and authentication differed; and partner setup as a side panel or checkbox table, in favour of a focused modal in EMAS Manager.
DESIGNED FOR THE REAL SYSTEM
Legacy EMIS Web patterns and PI scope shaped every phase. Failures were split into temporary (network, session, server) and persistent (no sharing agreement, no usual GP), with retry locking and bulk resend limited to reviewed failures to protect server load. Partner setup moved to admins, away from clinicians.
EVIDENCE AND CHECKS
Focus groups, support tickets, heuristic review and workflow mapping for two user groups, sender and receiver. Design checks covered icon plus text, colour never the only signal, keyboard access for resend, and an audit trail.
WHAT THE DESIGN ESTABLISHED
WHAT I WOULD MEASURE NEXT
Post-release validation is still to do: failures shown with clear status, resend success and retry rate, related support tickets, and partner apps configured. [Confirm which phases shipped, and add any measured result.]
View detailed case study