Cleveland Kitchen — UAT Acceptance Criteria
Version: 1.0 Date: March 16, 2026 Contract Type: Phased — billing starts after Phase 3 completion
Overview
Section titled “Overview”This document defines measurable acceptance criteria for each phase of the Cleveland Kitchen integration. It is designed for use during UAT with Ben, the Philippines CS team, and Dan.
The companion UAT Feedback Sheet (Google Sheet) is used to capture per-order feedback during testing.
This file is the canonical Cleveland Kitchen UAT acceptance criteria (phased metrics and sign-off). A duplicate, less formal summary previously at docs/ck-uat-acceptance-criteria.md was removed to avoid conflicting expectations.
Phase 1: Email Extraction & Validation
Section titled “Phase 1: Email Extraction & Validation”Duration: Weeks 2–4 | Sign-off Target: Week 4
1.1 Extraction Accuracy
Section titled “1.1 Extraction Accuracy”| Metric | Target | How to Measure |
|---|---|---|
| Customer identification | ≥ 95% correct | Match extracted customer name → WhereFour customer ID |
| SKU/UPC matching | ≥ 90% items matched | Count items with valid erp_item_id vs total items |
| Quantity extraction | ≥ 95% correct | Compare extracted qty to source document |
| Price extraction | ≥ 90% correct | Compare extracted unit price to source document |
| PO number extraction | ≥ 95% correct | Compare to source document |
| Address extraction (all 3 types) | ≥ 90% correct | Ship-to, bill-to, location each scored separately |
Test corpus: 10 real CK documents (Armada, Four Seasons, FreshPoint, Giant Eagle, Heinen’s, Misfits Market, Rainforest, Sirna & Sons, Sysco)
1.2 File Format Support
Section titled “1.2 File Format Support”| Format | Test File | Expected Result |
|---|---|---|
| four-seasons, misfits-market, rainforest, sirna-sons, sysco | Full extraction | |
| Excel (.xlsx) | giant-eagle | Full extraction |
| Excel (.xls) | heinens | Full extraction (xlrd fallback) |
| Email body text | freshpoint-ct | Full extraction |
| PDF (Armada/Darden) | armada | Full extraction |
1.3 UPC Normalization
Section titled “1.3 UPC Normalization”| Scenario | Expected |
|---|---|
| 12-digit UPC input | Normalized to match CK master SKU |
| 14-digit UPC input (leading zeros) | Normalized to match CK master SKU |
| Customer-specific product codes | Resolved via cross-reference aliases |
1.4 Visual Indicators
Section titled “1.4 Visual Indicators”| Feature | Acceptance |
|---|---|
| Discount/allowance chip | Yellow badge visible when discount line items detected |
| Price verification alert | Shown for CK org when prices differ from ERP |
| Validation badges | Customer ✓/✗, Shipping ✓/✗, All Items ✓/✗ visible |
Phase 1 Sign-off Criteria
Section titled “Phase 1 Sign-off Criteria”- All 10 test documents processed successfully
- Extraction accuracy meets targets above
- Ben + Philippines team confirm quality is “production-ready”
- Demo video delivered and reviewed
Phase 2: WhereFour Integration + Agentic Validation
Section titled “Phase 2: WhereFour Integration + Agentic Validation”Duration: Weeks 5–10 | Sign-off Target: Week 10
2.1 Agentic Validation Pipeline
Section titled “2.1 Agentic Validation Pipeline”The agentic validation system performs parallel checks and computes a weighted confidence score. Auto-submission occurs only when ALL checks pass and confidence meets the threshold.
| Validation Check | Weight | Description |
|---|---|---|
| Customer match | 25% | Customer identified in WhereFour |
| Item matching | 30% | All items matched with ≥ 90% confidence |
| Address validation | 15%* | Shipping address found in WhereFour |
| Pricing validation | 15%* | Prices within 10% tolerance of ERP pricing |
| Item validity | 15% | All matched items are active (not discontinued) |
*Weight redistributes proportionally if WhereFour doesn’t support the capability.
Confidence threshold: ≥ 95% (configurable per org via feature flag)
2.2 Auto-Submission
Section titled “2.2 Auto-Submission”| Metric | Target | How to Measure |
|---|---|---|
| Auto-submit rate | 60–80% of orders | Orders with zero validation issues / total orders |
| False positive rate | 0% | Auto-submitted orders with actual errors |
| Integration log created | 100% of submissions | Every auto-submit has audit trail in WhereFour |
2.3 Escalation / Manual Review
Section titled “2.3 Escalation / Manual Review”| Scenario | Expected Behavior |
|---|---|
| Unmatched customer | Escalate → manual review queue |
| Unmatched SKU | Escalate → show which items need attention |
| Address not found | Escalate → flag for CSR to verify |
| New pricing combo | Escalate → flag for pricing review |
| Confidence below threshold | Escalate → show confidence breakdown |
2.4 Safety Controls
Section titled “2.4 Safety Controls”| Control | Verification |
|---|---|
| Feature flags (4 flags) | Confirm each flag toggles correctly in Flagsmith |
| Dry-run mode | Orders score + log but do NOT submit to WhereFour |
| Idempotency | Duplicate submissions blocked by key |
| Fail-closed defaults | Disabled when flag service unreachable |
| Inventory blocking disabled | Orders submit regardless of stock status |
Phase 2 Sign-off Criteria
Section titled “Phase 2 Sign-off Criteria”- Orders successfully created in WhereFour via auto-submit
- Pricing validation working (flags new customer/SKU combos)
- 60–80% auto-submit rate achieved on test corpus
- Zero false positives in auto-submitted orders
- Integration log visible in WhereFour for every submission
- Error rate < 2% measured against manual entry baseline
- Philippines team reports measurable time savings
- Dry-run mode tested and verified
Phase 3: Production Deployment
Section titled “Phase 3: Production Deployment”Duration: Weeks 11–12 | Billing starts after Phase 3 completion
3.1 Production Metrics
Section titled “3.1 Production Metrics”| Metric | Target |
|---|---|
| Orders requiring minimal human editing | 60–80% |
| Validation time per order (Philippines team) | 5–30 seconds for auto-validated |
| Cross-reference table growth | New mappings captured automatically |
| System uptime | ≥ 99.5% during business hours |
Phase 3 Sign-off Criteria
Section titled “Phase 3 Sign-off Criteria”- Production deployment stable for ≥ 1 week
- Philippines team validates orders in Ordermatic → submits to WhereFour
- Positive feedback from Ben + Philippines team
- ROI analysis completed (time savings calculated vs. baseline)
- Cleveland Kitchen approves transition to paid agreement
UAT Feedback Process
Section titled “UAT Feedback Process”How to provide feedback
Section titled “How to provide feedback”- Open the CK UAT Feedback Google Sheet
- For each order processed during testing:
- Rate extraction accuracy (1–5)
- Rate validation correctness (1–5)
- Note any issues in the comments column
- Mark whether auto-submit decision was correct (Yes/No)
- Flag critical issues in the Blockers tab
Feedback cadence
Section titled “Feedback cadence”- During Phase 1 testing: Daily review of 5–10 orders
- During Phase 2 UAT: Full batch of 20+ orders, 2 review sessions
- During Phase 3: Continuous monitoring, weekly summary
Contacts
Section titled “Contacts”| Name | Role | Responsibility |
|---|---|---|
| David Boone | ERP Unlocked, Lead | Architecture, integration, delivery |
| Ben | CK, Customer Service Mgr | Quality approval, Philippines team oversight |
| Dan | CK, Management | Business sign-off, WhereFour access |
| Philippines Team | CK, Order Entry | Day-to-day UAT testing, order validation |