Skip to Content

AI is Business

Why IT Alone Cannot Solve the Problem
April 17, 2026 by
AI is Business
INNOCARUS AG, Olivier Gemoets
| No comments yet
This article is for CEOs, CIOs, COOs and transformation leaders who are investing in AI — or are about to — and are wondering why results are not materialising.


AI is Business: Why IT Alone Cannot Solve the Problem


The Uncomfortable Truth


95%. That is the share of enterprise generative AI pilot programmes that fail to deliver any measurable return on investment — according to MIT's State of AI in Business 2025 report, based on 150 executive interviews, 350 employee surveys and an analysis of 300 publicly documented AI deployments.

30 %. That is the share of generative AI projects Gartner predicts will be abandoned after proof of concept — due to poor data quality, unclear business value, escalating costs or inadequate risk controls. (Gartner, July 2024)

40 %. Shat is the share of agentic AI projects Gartner expects to be cancelled by end of 2027 — "often misapplied, driven by hype, and blind to the real cost of scale". (Gartner, June 2025)

And yet: nearly 9 out of 10 organisations say they are already regularly using AI. (McKinsey State of AI, 2025)

The gap between usage and value creation is not a technology problem. It is a leadership, governance and organisational problem.

The answer is uncomfortable — but clear: AI is not an IT project. It is a leadership issue.


Why AI Initiatives Fail — The 4 Root Causes


1. Strategy without value: use cases without prioritisation


Most organisations start with a long list of AI ideas. Few have a structured, prioritised portfolio with clear business cases and value hypotheses. The result: energy is scattered, investments are duplicated, and no initiative reaches the critical mass to deliver real impact.

The question to ask yourself: Do you have a use case portfolio with explicit prioritisation criteria — and do you systematically track value realisation after go-live?


2. Data, governance and risk: the underestimated foundation


Data quality is the most frequently cited failure reason — and it is legitimate. But the deeper problem goes further: Who is responsible for which data? Who approves access? Who governs quality standards? Without clear data ownership, even good data becomes inaccessible — or unreliable — in practice.

Gartner (2025): 63% of organisations do not have AI-ready data infrastructure.

The same logic applies to governance and risk. In regulated industries — and Switzerland is full of them — McKinsey estimates that 30 to 50% of AI development time is spent on compliance. This is not a sign of excessive bureaucracy. It is a sign that AI risk governance was not thought through from the outset, but bolted on afterwards. Responsible AI frameworks, model risk management and privacy-by-design are not obstacles to speed. They are preconditions for sustainable scale.


3. Operating model: roles, ownership, decision chains


Who owns an AI initiative after the pilot? Who monitors the model in production? Who decides when to retrain, redeploy — or retire?

Most organisations never answer these questions before launching. The result: successful pilots with no owner in production. Models drift. Decisions become opaque. Trust erodes.


4. Change management, adoption and operations: the human factor — through to production


MIT (State of AI in Business 2025) shows: organisations that make line managers — not central AI labs — the driving force of adoption are significantly more successful. Prosci's decades of research confirms the same pattern: without Awareness, Desire, Knowledge, Ability and Reinforcement at the individual level, even the best technology goes unused. Most AI programmes invest 95% of the budget in technology — and 5% in the human factor. The ratio should be closer to 50/50..

The same question arises in operations. The most dangerous phrase in the AI context is: "The pilot was a success." McKinsey (2025) confirms: "The transition from pilots to scaled impact remains a work in progress at most organisations." MLOps infrastructure, monitoring, drift detection and clear operational ownership are not secondary details — they determine whether a pilot ever generates real value.


What Successful Companies Do Differently


BCG's analysis of over 1,250 companies across 68 countries reveals a consistent pattern among AI leaders:

  1. AI strategy is business strategy — not an IT initiative running alongside the business
  2. IInvestment governance is explicit — with clear decision rights, portfolio management and ROI tracking
  3. Line management drives adoption — not the IT department or a central AI lab
  4. Responsible AI is built in from the start — governance, ethics and compliance are architected in from the beginning, not added afterwards
  5. Data is treated as a strategic asset — with ownership, quality standards and access policies defined before the first model is trained

What distinguishes companies is not the quality of the AI model. It is the quality of the organisation around it.


The Pragmatic First Step — A KI-Readiness Assessment


Good news: you do not need to solve all four problems at once.

What you need is clarity on where your organisation stands today — and what to address first to unlock disproportionate value.

This is precisely the purpose of a structured KI-Readiness Assessment.

What it covers

The INNOCARUS KI-Readiness Assessment evaluates your organisation across 10 dimensions — from strategy and leadership to data, technology, ethics and ecosystem:

Dimensions
Strategy & Value Case
Leadership & Governance
Data Foundation
Technology & Architecture
Ethik, Risk & Compliance
Culture & Change Management ⚡
Talent & Skills
Operating Model & Processes
Finance & Investment
Ecosystem & Partnerships


The framework integrates leading methodologies from McKinsey, Deloitte, BCG, Bain, Accenture, EY and Umbrex — and translates them into practical, interview-based assessment criteria.

The assessment is comprehensive by design — 30 core elements, 300+ structured questions ready to be used if needed, a 6-level maturity scale. But it is pragmatic in application.

You do not need to answer every question in a first pass. We start where the pressure is greatest: your top 3 to 5 strategic use cases, your most critical data and governance gaps, your capacity to act over the next 12 months — not the next 5 years.

The result is not a score. It is a prioritised action roadmap: where are your quick wins, your medium-term investments, your structural gaps — ranked by impact and feasibility.

"The goal is not a perfect score — it is an actionable roadmap."

The three deliverables you walk away with:

  1. Readiness Profile — a scored, evidence-based assessment across all 10 dimensions
  2. Gap & Risk Map — the 3–5 structural blockers most likely to prevent scale
  3. Prioritised AI Roadmap — a 100-day action plan with ownership, timeline and KPIs


Conclusion: AI is a Leadership Decision


An AI initiative managed primarily as an IT project has a high probability of failure. MIT, Gartner, McKinsey, BCG and three decades of transformation experience confirm this — independently of one another.

What separates the 5% of companies that scale from the 95% that stagnate is not better algorithms. It is better leadership: clearer strategy, more deliberate governance, more targeted change management — and the organisational readiness to move from pilot to production.

The first step is not a pilot. It is the understanding of where you stand today.

Those who underestimate this complexity invest in pilots — and wait in vain for impact. A sensible first step is therefore not another use-case ideation session, but an honest stocktake: what is genuinely in place today — in strategy, data, governance, organisation and culture? That is precisely what a structured Readiness Assessment provides.

Have you experienced an AI initiative that succeeded — or failed — because of organisational factors? I look forward to the exchange in the comments.

Olivier Gemoets is the founder of INNOCARUS AG. For three decades, he has guided companies through complex transformations — from strategy through to operational anchoring. Areas of focus: Digital Transformation, AI Strategy, Change Management & Interim Leadership.


Sources

  • MIT, State of AI in Business 2025mlq.ai
  • Gartner, GenAI Projects Abandoned After PoC, Juli 2024 — gartner.com
  • Gartner, Agentic AI Projects Canceled by 2027, Juni 2025 — gartner.com
  • Gartner, Lack of AI-Ready Data Puts AI Projects at Risk, Februar 2025 — gartner.com
  • McKinsey, The State of AI 2025mckinsey.com
  • McKinsey, Overcoming Two Issues Sinking Gen AI Programs, 2025 — mckinsey.com
  • BCG, AI Adoption Puzzle: Why Usage Is Up But Impact Is Not, 2025 — bcg.com
  • RAND Corporation, via WorkOS AI Failure Analysis, 2025
  • S&P Global / Medium AI Analytics Diaries, 2026


Share this post
Tags
Archive
Sign in to leave a comment