One of the key messages we are using in the market is “Everyone’s ready for AI except your data.”The funny thing about this statement is that we didn’t come up with it. We heard versions of this sentiment from customers, and there is truth to it, so we decided to just tell it back to them.
As AI becomes a growing business imperative, management teams need to ask: Is our data quality good enough? Is our data well-governed and secure? What are the risks of inaccurate data?
Many organizations still have work to do to get their data into the necessary state of readiness, but not because they’ve neglected data management. They haven’t. It’s just that AI is more demanding and less forgiving. In the past, a data error might have meant a missed business opportunity. Now, faulty data can lead to AI bias or hallucinations—a more serious misfire that harms customer relationships and brand value.
This helps explain why “data and AI” has become the new catch phrase; the two are inseparable. But that means data quality must meet a higher threshold.McKinsey recently pointed to “at-scale data quality checks” as a way to build confidence in AI and ensure readiness for governance tools and development protocols as part of its AI Trust Maturity Model.
I’m having frequent conversations with CEOs and other business leaders who are anxious to understand how to proceed, and it’s top of mind in the board meetings I attend at Informatica, Box, and The Tech Interactive. As board members, we have a responsibility to assess both strategic opportunities and risks, which means enterprise AI is often on the agenda.
Here’s the rub: You can’t have trusted AI without top-quality data. That implies clean, accurate, and timely data, but it’s more than that. There must be comprehensive governance, cross-functional committees for effective integration and implementation, and training, education, and ethical guidelines.
In other words, execution requires a top-down commitment, which is why data for AI is rising to a board-level concern. That wasn’t the case even a few months ago. To start the year, there was barely a mention of data-readiness for AI in reports on board priorities from consulting firms like PwC and EY. But the pace of change is forcing corporate management to take notice.
REVIEW, ALIGN, DEPLOY
What can the board and other executive leaders do to ensure their companies go above and beyond traditional data management practices to produce “AI-grade” data?
A good starting point is a clear-eyed assessment of an organization’s data landscape, including which assets are most essential to operations, the platforms and infrastructure used to manage those data stores, and how much of it has been migrated to the cloud, which is a factor in AI-readiness.
C-level executives should also ensure their AI strategies and business goals are well-aligned. Whether the objective is to expand into new regions, drive product development and revenues, or lower costs through supply-chain efficiencies, data-driven AI solutions can be brought to bear—but only if data is high-caliber.
The first steps in AI implementation are development, testing, and pilot projects, to be followed by wider rollouts. This is where data curation and quality management get put to the real-world test with customers and partners. That’s true whether it involves generative AI, private models based on retrieval-augmented generation (RAG), or agentic AI.
Other data and AI essentials are the ability to monitor, measure, and improve. Many organizations are still in the early stages with AI, but as they scale up and out, the executives responsible for oversight will need metrics and key performance indicators (KPIs) to assess progress and evaluate the effectiveness of their investments. These AI dashboards can also be invaluable for risk management.
CONFRONTING COMPLEXITY
A word of warning to business leaders who decide to get involved with AI data: Don’t be surprised if you encounter a mess. Corporate data systems are notoriously fragmented and complex. A survey by analyst firm BARC and sponsored by Informatica found that data quality and lineage were the most-cited challenges and obstacles to AI, identified by 45% of respondents.
Yet, given the growing role that AI has for many organizations, there’s no turning back. My advice is to follow best practices in modern data management, and you will be headed in the right direction—toward trusted data that is well-tuned for AI.
Like so much of what we do as stewards for the companies we build and guide, business performance will depend largely on how we handle AI’s opportunities and challenges. That’s why every AI strategy must also be a data strategy.