⌘K
Present review
MR
Maya Reynolds
Brightwell Pharmacy
Rubric & trust
Built from Brightwell Pharmacy's own material: playbook, policies, voice guide and historic tickets, not a generic template. A score is only worth what the rubric behind it can defend.
Pillars
5
weighted into one Prism Score
Source documents
7
ingested from Brightwell Pharmacy's own library
Scores challenged
14
9 upheld · 5 adjusted. Challenges improve the rubric
Calibration
Weekly
sampled against your QA lead's judgment
Accuracy & policy30% of score
- ✓Order, refill and insurance facts match the system of record
- ✓Policy quoted matches the current published version (returns policy v4.2, Oct 2026)
- ✓No invented commitments, credits, shipping dates, coverage promises
- ✓Correct use of current promotions and pricing
Process adherence20% of score
- ✓Identity verified before any account or Rx detail is shared (two identifiers)
- ✓Required systems checked in order (BrightOS → ClaimsHub → Zendesk)
- ✓Contact reason coded to the correct category taxonomy
- ✓Escalation path used where the SOP requires it
Resolution completeness25% of score
- ✓Root request addressed, not just the surface question
- ✓Next steps and timelines stated explicitly
- ✓No avoidable follow-up contact created (drives the reopen metric)
- ✓Confirmation sent where the workflow requires it
Empathy & tone15% of score
- ✓Acknowledges impact before process (medication delays are health events)
- ✓Plain language, no internal jargon or blame-shifting
- ✓Pace matched to the customer's distress level
- ✓Brightwell voice guide (2026) followed
Compliance & safety10% of score
- ✓HIPAA: PHI only after verification, never over an unverified channel
- ✓Zero clinical advice, dosage, interactions and symptoms route to the pharmacist line
- ✓Adverse-event or misuse signals escalated within the interaction
- ✓Payment data handled through the tokenized flow only
What Prism ingested
- Brightwell Support Playbook v7SOP · 148 pagesingested 2026-10-02
- Returns & refunds policy v4.2Policy · effective Oct 1, 2026ingested 2026-10-02
- HIPAA handling runbookCompliance · 36 pagesingested 2026-10-02
- Brightwell voice & tone guideBrand · 2026 editioningested 2026-10-03
- Historic tickets (Jan–Sep 2026)41,286 interactionsingested 2026-10-05
- Macro & snippet library214 macrosingested 2026-10-05
- Escalation matrixSOP · pharmacist + clinical routingingested 2026-10-06
How trust is built
- 1Every score opens its transcript. A number a client can't audit is a claim; a number they can read behind is evidence.
- 2Anyone can challenge a score, agents included. Disputes make the rubric better, not weaker.
- 3Reopens keep the score honest. If scores drift up while reopens drift up, the rubric is grading too generously. Both live on the same dashboard on purpose.
- 4Scores coach before they ever touch pay. Agents see their scores and the full rubric. Compensation models come only after the measurement is trusted.
- 5Every change is a version. The rubric that graded a conversation is frozen with it, so a score from March still means what it meant in March.