AI Is Ready, Your Workforce Isn’t: Why AI ROI Falls Short

High-level thesis

AI capability is accelerating faster than organizations’ ability to capture value. The primary bottleneck is human readiness — workforce skills, role design, change management, processes and data — not the technology itself.

Key metrics / KPIs

• By end of 2025:
  • ~20% of AI initiatives will have achieved ROI (about 1 in 5).
  • ~2% of initiatives will have delivered truly disruptive/transformative value (about 1 in 50).
• 74% of CFOs report observing productivity gains from AI (time saved, faster decisions) — often non‑financial.
• Only ~11% of CFOs report clear financial ROI from AI today.
• To capture financial ROI (i.e., move from ROE to ROI by extending cases), expect to re‑engineer ~30–60% of an end‑to‑end process.
• Training and change costs are materially higher than with past enterprise tech:
  • Training effort increases to ~25% of the implementation/preparation effort.
  • Change management/process redesign effort can be ~200% higher versus prior projects (ERP example).
• Job‑impact breakdown (recent analysis):
  • ~1% job loss due to productivity‑only automation.
  • ~17% due to repositioning (headcount reduced in some areas and hired in others).
  • The majority of other job losses were driven by non‑AI factors.

Frameworks, playbooks and concepts

Value portfolio framework (three layers of AI value)

• Defend (ROE — Return on Employee)
  • Augmentation, time savings, better quality/throughput.
  • Usually non‑financial and incremental.
• Extend (ROI)
  • Re‑engineer processes end‑to‑end for competitive differentiation and measurable financial returns.
• Upend (ROF — Return on the Future)
  • Longer‑term disruptive bets that create new products, services or markets.

Investment portfolio approach

• Allocate AI investment across defend / extend / upend rather than single pilots.
• Example guidance: ~20% defend, ~50–60% extend, remainder to upend (adjust by risk appetite).

AI maturity mapping for workflows

• Example (software engineering): Levels 1–5 from static chatbots to autonomous, agentic systems.
• Organizations should decide target maturity level for each workflow and define human vs machine responsibilities at each level.

Role heat‑mapping

• Visualize which tasks in a role will be automated versus remain human (darker = more human required; lighter = more automation).
• Use heatmaps to plan role evolution and reskilling.

Concrete operational recommendations / playbook items

• Clarify desired value up front: align IT, CFO, HR and business leaders on whether an initiative is intended as ROE, ROI or ROF to avoid mismatched expectations.
• Adopt a portfolio investment strategy across defend / extend / upend use cases rather than one‑off pilots.
• Prioritize extend (process re‑engineering) to drive financial returns — be prepared to redesign a large share (30–60%) of processes to reach ROI.
• Treat role/job redesign as a major program — far larger in scope than simple hiring or layoffs (an order‑of‑magnitude larger effort per analysts).
• Budget for hidden human costs: allocate substantially more resources to training, change management and process re‑engineering than for past large IT projects.

Build workforce skills systematically:

• Use‑case identification: teach people to spot opportunities that move beyond ROE to ROI/ROF.
• Technology fluency: ensure a basic understanding of AI limits and capabilities.
• Prompting: train staff to craft context‑aware prompts that encode organizational constraints and risk considerations.
• Discernment: teach evaluation of outputs for usefulness and actionability (avoid information overload).

Additional operational guidance:

• Map workflows to AI maturity levels and define the human/machine split — not all workflows should aim for full autonomy.
• Create role‑specific transformation plans (reskilling, new responsibilities). Example: software engineers may see core tasks automated earlier, shifting toward broader responsibilities; people managers are more likely to be augmented than automated.
• Communicate a clear vision to employees about how roles will evolve to reduce fear and support repositioning/hiring.

Examples & evidence cited

• Gartner has produced role‑specific examples and heatmaps (15–20 IT‑related role examples referenced), including a software‑engineer maturity ladder (levels 1–5).
• Market behavior observed: companies (for example, headlines about Salesforce and IBM) are reducing headcount in some functions while hiring new capabilities elsewhere — described as repositioning rather than pure automation‑led job loss.

Actionable next steps for leaders

1. Align executives on the type of value sought (ROE / ROI / ROF) and set portfolio targets for spend.
2. Run a role heatmap exercise for major functions to identify automation exposure and reskilling needs.
3. Estimate training and change‑management budgets at 2–3x prior large enterprise tech rollouts; explicitly provision for up to ~200% more process redesign effort.
4. Prioritize 30–60% process re‑engineering where ROI is the goal; measure process change coverage as a KPI.
5. Build capability programs around:
   • Use‑case identification
   • Prompt engineering
   • Technology fluency
   • Discernment

Presenter / source

• Alicia Mullery, VP Analyst, Gartner (presenter)
• Karen Stokes Lockhart, host (Gartner ThinkCast)