Skip to content

pwc-20-item-checklist.md

Latest commit

 

History

History
59 lines (45 loc) · 2.28 KB

pwc-20-item-checklist.md

File metadata and controls

59 lines (45 loc) · 2.28 KB
Raw
title PwC 20-Item ROI Checklist (Do's and Don'ts)
source-id pwc-agentic-playbook-2026
wiki-page (02_References/enterprise-ai/wiki/frameworks/pwc-top-10-do-and-dont-checklist.md)
last-synced 2026-06-07

PwC 20-Item ROI Checklist

What this reference contains

PwC's operational pre-flight gate for any AI investment: 10 Do's and 10 Don'ts. Functions as a structured forcing device before greenlighting projects. An unanswered item is itself a red flag.

The 10 Do's

# Do
1 Ensure a customer-centric approach
2 Conduct thorough research on capabilities and limits
3 Start with small projects — pilot first
4 Monitor performance and iterate against business-aligned metrics
5 Build cross-functional teams (IT, ops, finance, marketing)
6 Invest in employee training
7 Invest in quality data
8 Prioritize data security and privacy
9 Invest in scalable AI platforms
10 Invest in continuous learning

The 10 Don'ts

# Don't
1 Ignore customer feedback
2 Underestimate complexity (AI is not plug-and-play)
3 Rush implementation
4 Neglect human oversight
5 Ignore user adoption
6 Overlook ethical considerations
7 Ignore change management
8 Underestimate costs (budget for infrastructure, maintenance, training)
9 Ignore partnerships
10 Overlook long-term sustainability

Source: PwC Agentic AI Playbook 2026, pp.16–17.

Critical pairs

  • Don't #2 + Don't #8 are the most load-bearing — AI complexity and total-cost underestimation are the #1 and #2 causes of failed projects per Stanford's invisible-costs research (77% of hardest challenges are change management, data quality, process redesign).
  • Don't #4 (human oversight) caps how far automation can go — even in autopilot mode.
  • Do #7 + Don't #2 = data quality gate — cannot skip.

When to consult

  • For skill general-roi-gate: this IS the core content of the ROI gate — load all 20 items, run each as a written Q&A before greenlighting any AI investment.
  • For skill process-pilot-design: Do #3 and Do #4 define the pilot philosophy.

Source

PwC Agentic AI Playbook 2026, pp.16–17. Banner: "Avoid the GenAI hype trap — focus on pragmatic steps that deliver real value."