Replace Streamlit and JupyterLab commands with Mercury for serving interactive notebooks as web apps. Update README to reflect new architecture where notebooks are the primary deliverables, utilizing Mercury input widgets for live client tuning. Add export_report.py script to generate LLM-readable HTML/Markdown reports from the notebooks. Update corrected business case notebook to include Mercury dependency and usage instructions.
Genesys CX Cloud TEI — December 2025
Source: Forrester, The Total Economic Impact™ Of CX Cloud — Cost Savings And
Business Benefits Enabled By Genesys And Salesforce (commissioned by Genesys
and Salesforce, December 2025). PDF in docs/.
Headline (published, 3-yr risk-adjusted PV @ 10%)
| Metric | Value |
|---|---|
| Benefits PV | $14,840,638 |
| Costs PV | $4,057,170 |
| NPV | $10,783,468 |
| ROI | 266% |
| Payback | ~4 months (computed; not headlined in the study) |
Composite: global supply company, $2.5B revenue, 10,000 employees, 600 CX agents (400 concurrent licenses), 80,000 weekly interactions @ 12 min.
Structure
4 benefits (legacy retirement ↓5%, self-service savings ↓15%, agent
efficiency ↓10%, agent-assist sales ↓5%) and 3 published costs (licenses ↑5%,
implementation ↑10% — initial-only, ongoing management ↑10%), plus one
Palladium addition: genesys_ai_tokens, an AI Experience token consumption
line the published study omits (it models $0 AI cost while three of four
benefits depend on AI). Stored exactly as Athena stores it — a single annual
cost value, entered from the Genesys quote in 01_business_case.ipynb (which
includes a sensitivity sweep), with quote details kept in the field notes.
Seeded at $0 to reproduce the published totals.
Study quirks (documented, handled)
- p.14 prints implementation initial as $1,304,600; correct figure is $1,309,000 (= 1,190,000 × 1.10) per the detail table and cash-flow analysis.
- B7's printed formula cites B2 (15%) where the 12-minute interaction length is meant; the result (40 FTEs) is correct.
- The initial cost is ~32% of cost PV, so Athena's discount-initial-as-Year-1
behaviour shifts ROI to ~277%. Verification matches
ATHENA_EXPECTEDtightly, then reconciles toPUBLISHEDwith this explained delta.
Notebooks
| Notebook | Purpose |
|---|---|
00_provision.ipynb |
Create template + fields + tool in Athena (client/proposal selection), seed, calculate, verify |
01_business_case.ipynb |
Working business case + Genesys AI token quantity × price sensitivity |
Env keys are study-scoped: PALLADIUM_GENESYSCX_REPORT_PUBLIC_ID,
PALLADIUM_GENESYSCX_TOOL_PUBLIC_ID, PALLADIUM_GENESYSCX_PROPOSAL_ID /
PALLADIUM_GENESYSCX_ENGAGEMENT_ID.