The GenAI gold rush and why most CIOs are still stuck at basecamp

The momentum behind Generative AI (GenAI) is undeniable. Across industries, CIOs are being asked to deliver transformational impact, from streamlined operations to hyper-personalised customer experiences. However, despite widespread experimentation, initiatives often stall before they scale.

According to Gartner, over 60% of AI projects are expected to fail due to poor data readiness. What separates projects that realise business value from those that falter isn’t access to better algorithms; it’s due to a lack of trustworthy, AI-ready data. Poor quality, inconsistent, or siloed data undermines trust in outputs, delays time-to-value, and increases compliance and cost risks.

To unlock business value, data must go beyond being “clean.” It must be accurate, complete, consistent, timely, and contextually relevant. These five dimensions of data quality form the foundation for GenAI readiness. The highest-performing GenAI initiatives begin with trust, which starts at the data layer.

Where GenAI projects lose momentum

Despite growing investment, many GenAI initiatives underperform due to foundational data challenges:

  • Manual data wrangling slows delivery – 53% of organisations see data preparation as a major cost and ROI challenge. Manual data preparations lows delivery, increases labour costs, and burdens technical teams. Automating data classification and discovery early can simplify operations and improve focus on delivery.
  • Unstructured and siloed data remains untapped – This is exacerbated by the nature of enterprise data itself. Up to 90% of enterprise data is unstructured, and 73% identify data silos as the primary barrier to quality. This shows that much of the valuable data remains hidden from AI systems. Successful GenAI projects utilise discovery tools to uncover and manage unstructured data at scale.
  • Inconsistent data increases costs and delays – With inconsistent data, costs rise rapidly. Frequent retraining often reflects weak data foundations, adding computing costs and delays each cycle. With only half of organisations satisfied with data accuracy and reliability, it’s evident that models can’t compensate for bad inputs. Strong data management practices are essential to minimise rework and keep GenAI efforts on track.
  • Sensitive data without governance poses risk – Compliance is another risk factor alongside cost and quality. Unclassified sensitive data, such as PII/PHI, can enter model training, increasing regulatory violations. Businesses incorporating policy-driven governance and automated classification lower their exposure and reduce their audit footprint.
  • Poor data quality leads to model hallucinations – Only 53% of organisations are satisfied with their data quality. GenAI models are only as good as the data they’re trained on. Incomplete or biased data leads to dreaded hallucinations. Successful GenAI leaders prioritise data discovery and quality from the outset, not just for cleanliness, but to ensure that the content is representative, complete, and relevant.

What the highest-performing projects do differently

Across industries, organisations achieving meaningful ROI from GenAI share a common playbook:

  • They prioritise AI-ready data from the start – According to Gartner, AI-ready data must be fit for purpose, not just ‘clean’. It’s data that’s contextual, complete, governed, and aligned to the use case. Successful teams align data early, assess quality and risk before training, and embed governance frameworks from day one.
  • They automate discovery and classification – A global professional services firm used EncompaaS to automate the classification of over 500 million documents and delete 10 million over-retained files, reducing the scope of audits. Automation accelerates GenAI readiness, enhances legacy content management, and reduces manual overhead.
  • They bring unstructured data into the fold – Unstructured data makes up most enterprise content but remains largely untapped. Successful organisations use automated solutions to analyse and extract insights from contracts, emails, and archives. By implementing tools that can discover, enrich, and normalise unstructured content across various repositories, businesses can expand the usable data for GenAI applications.
  • They embed governance in the pipeline, not after the fact – Retrofitting compliance after deployment leaves risk on the table. Leading teams integrate policy-based governance and lifecycle rules into their data strategies, ensuring sensitive information, like PII or PHI, is securely managed before GenAI is introduced.

Why data prep is the first GenAI use case

Data preparation is often seen as a bottleneck. However, leading organisations are flipping that narrative, using AI to prepare for AI.

As shown in our recent report, discovering, classifying, and de-risking the data that will eventually power downstream models is one of the best ways to begin using AI. By scaling tools for discovery and classification early, organisations reduce risk, accelerate readiness, and turn data preparation from a sunk cost into a strategic advantage.

Shifting from experimentation to execution

GenAI is a priority for many CIOs, but ambition is currently surpassing data readiness. Nearly 60% of executives believe their data is not prepared for GenAI, indicating that a lack of confidence in their information assets hinders growth.

Gartner predicts that 30% of GenAI proofs-of-concept will be abandoned by 2026, due to unclear business value or poor data quality. The risk of stalling after initial experimentation is as real as the opportunity.

Moving from POC to production means reframing the roadmap. Data readiness isn’t a follow-on task; it’s the differentiator. Successful organisations ensure their data is visible, accurate, and governed before GenAI begins.

Leading organisations treat data readiness as an ongoing capability, not a one-off milestone. With discovery, classification, and policy-driven governance embedded into the lifecycle, they’re building an intelligence system that supports GenAI as it evolves.

Build on a foundation you can trust

The GenAI advantage doesn’t start with algorithms; it starts with data.

Projects succeed when they invest early in the fundamentals: discovering what data exists, understanding its value, and governing it with confidence. In a landscape shaped by compliance, complexity, and rapid change, trust is earned through data readiness.

The highest-ROI initiatives aren’t gambling on generative capabilities. They’re investing in secure, scalable information foundations that support innovation without compromise.

Explore how the EncompaaS intelligent information management platform helps enterprises prepare their data for GenAI.