Supply chains are racing to adopt AI, but one issue keeps blocking progress: data quality. In a recent session with Intent Group’s Ed Lawson, we explored why AI success depends less on models—and more on clean, contextual data. The biggest risk? Skipping foundational work in the rush to scale.
In a recent conversation with Ed Lawson, Managing Director at Intent Group, we explored AI’s ROI and the Killer App for Supply Chains. We also ran a live poll with supply chain professionals to understand their real-world blockers to AI adoption. The #1 result? Data quality. 57% of respondents flagged it as the primary obstacle - above ROI, talent, or scalability concerns.
The poll results confirmed something we're seeing in the market: the disconnect between AI ambition and foundational readiness. While companies chase AI capabilities, they often skip the first step - ensuring their data is usable, reliable, and integrated.
Despite the AI hype cycle, one truth holds: AI is only as good as the data it’s fed and trained on. If the data is messy, siloed, or outdated, even the most advanced models will underperform - or mislead entirely. If the models ingest good operational data, they will be able to train better and deliver good insights.
We’ve been talking about data quality for decades, and AI brings it into sharp relief. This legacy issue has now become a frontline problem, with AI amplifying the risks of bad data rather than masking them.
This isn't just theoretical. IDC’s research shows that 90% of enterprise data is unstructured, and yet less than half of it is ever analyzed to extract value. IDC calls it what it is: a "gold mine" that’s routinely wasted.
But quality alone isn’t enough. As anyone in supply chain knows, the right number at the wrong time - or without context -is still the wrong decision. It’s not just about clean data. It’s about context-rich data.
Supply chain leaders aren't just looking for systems that can analyze, they need AI that understands. Missing context leads to misleading outputs, especially in forecasting.
Humans still outperform AI in decision-making when critical nuances aren’t captured by data alone. Ultimately, humans must provide the context. But that’s where AI agents can help by jumping between silos and integrating fragmented information.
AI agents are really good at jumping between silos. But that still requires context to be codified and fed in – and that’s where LLMs come in, with their capacity for structuring, normalizing, mapping, and validating data, as well as reasoning and prompting to action based on trained data.
Without both clean and contextualized data, AI pilots hit walls fast. We’ve seen this repeatedly with organizations eager to prove value but stuck in data chaos. Typical symptoms?
This erodes trust both in the AI systems and the teams that champion them.
IDC backs this up with hard data: organizations with fragmented, siloed unstructured data face double the cost of data breaches - averaging $4.5M annually, compared to $2.2M for companies that manage their data through centralized, secure platforms.
The cost of failure isn’t just technological - it’s operational, financial, and regulatory. Fragmented data slows AI adoption and bleeds productivity:
Meanwhile, productivity loss from data hunting, duplicated effort, and version mismatches hurts operational agility - especially in supply chain contexts where timing is everything.
The good news is all these are solvable problems. But they require a strategic shift.
Data readiness must precede AI readiness. Without putting the work in structuring processes, transferring context, and cleaning data, AI won’t just under-deliver - it will mislead.
Vertical-specific AI agents beat generic platforms because they understand nuances in processes, logic, and decision-making.
We’re at a stage where we only see AI succeeding when it solves real, costly inefficiencies. However, chasing one-off ROI creates more silos. Poor data quality is a silent cost center and a clear place to start. But clean data and structured processes are the healthy foundation as well as the steady path to scalable AI and survival over the next decade.
Data quality isn’t an IT hygiene issue. It’s the #1 success factor for AI. AI transformation begins with data transformation, which relies heavily on restructured processes and clean data. No shortcuts. No exceptions.
Subscribe to our newsletter, and we will send AI automation insights like this straight to your inbox on a regular basis.
.png)
.png)