August 28th, 2025
By Dr. Kevin O’Flynn | Supply Chain Planning
Artificial intelligence is revolutionizing supply chain planning for consumer packaged goods (CPG) companies and their retail partners. But, as with any disruptive technology, the promise comes with real risks. Here’s what you need to know—and what to do about it.
Using AI Right
AI can optimize demand forecasting, automate inventory management, and enable real-time decision-making, helping brands and retailers stay ahead in volatile markets. In a world where consumer preferences shift rapidly and supply disruptions are common, the ability to react quickly is crucial for CPG companies and their retail partners.
As more companies adopt AI, those who lag behind risk losing shelf space, market share and customer loyalty. Still, the road to AI-powered supply chains is paved with pitfalls. Ignore them at your peril. Without careful planning and risk management, AI initiatives can backfire, causing more harm than good.
Risk #1: Data Quality and Integration Nightmares
The reality: AI is only as smart as the data you feed it. Most CPG and retail supply chains are still plagued by:
- Siloed data: Sales, inventory and production data often live in separate systems, and often in different excel workbooks, making it tough for AI to see the big picture. When data isn’t unified, AI models can’t accurately predict demand or optimize inventory across the supply chain.
- Inconsistent data: because the data is in silos, the naming convention for the same SKU can vary from system to system. This makes it slow to map between systems.
- Legacy tech: Outdated platforms can’t support timely data flows or advanced analytics. Many CPGs and retailers still rely on old ERP systems that struggle to keep up with modern AI tools.
- Garbage in, garbage out: Poor-quality or incomplete data leads to bad forecasts, missed opportunities, and costly mistakes. If your data isn’t clean, AI can have hallucinations that hurt your bottom line instead of helping it.
What it means for you:
AI that can’t access or trust your data will underperform—sometimes spectacularly. That’s a risk to service levels, inventory, and customer trust. Inaccurate AI-driven decisions can result in empty shelves, overstocked warehouses, and frustrated customers. Start by doing an honest assessment of your data quality. Where appropriate, you may be able to use AI to fill in the gaps, but be deliberate in where you apply it.
Smart moves:
Invest in data modernization and integration before scaling AI. Upgrading your systems and connecting data sources ensures AI has the foundation it needs to deliver results. Also, break down silos and get your teams and systems talking. Last, prioritize data governance and quality at every stage.
Risk #2: Use the right AI for the job
“AI” is used as a catch-all term that sounds like one solution but in reality is a myriad of different techniques from large language models (LLMs), to advanced analytics to machine learning.
Not all AI was created equal: LLMs, like ChatGPT, have transformed our opinion of AI’s capabilities. But use the right tool for the job. LLMs can write your next piece of marketing content but it can’t do complex calculations like route optimization in Google Maps.
Your board is saying “we need to do AI”: Great! But start with the problem you’re trying to solve and choose the right AI solution, rather than starting with a piece of AI and trying to fit it to a problem. And often AI isn’t the solution, it might just be a part of it.
Best in Class users think about edge cases: This is where hallucinations can become blurred with correct answers and it can be hard for anyone not an expert user to know what to do.
The risk? Using the wrong approach means the project will never get off the ground. Want to use Machine Learning? You need a lot of clean historic data, ideally well tagged. Want to use LLMs? Great for writing text and supporting critical thinking.
What it means for you / Smart moves
Start with your business objectives and challenges, then review what tools you have available right now internally. Once you’ve identified the gaps, look at the market to see how market leaders are solving and the types of solutions they’re using. If you’re being sold “AI,” critically evaluate whether that type of AI is the right approach for your challenge or opportunity.
Risk #3: Overreliance and “Black Box” Decisions
The reality is that AI can automate complex decisions—sometimes too well.
Opaque logic: Many AI models are “black boxes,” making it hard to explain or challenge their recommendations. When teams can’t understand AI decisions, it’s difficult to justify actions to stakeholders, regulators or auditors.
Unexpected outcomes: AI can make mistakes, especially when market conditions shift or if the data is biased. Relying solely on AI can lead to errors if the underlying assumptions no longer hold true.
Human expertise still matters: Blindly trusting AI can lead to stockouts, overstocking, or missed demand signals. Human oversight is essential to catch issues AI might miss and to interpret results in context.
What it means for you:
Unchecked AI can amplify errors, create compliance risks, or alienate customers if decisions can’t be explained or justified. Transparency and accountability are critical for maintaining trust with customers, partners, and regulators.
Smart moves:
Keep humans in the loop for critical decisions, demand transparency from AI vendors—insist on explainable AI and continuously monitor, test and adjust AI models as markets evolve.
What CPGs Can Do—Now
Start smart, scale wisely:
- Pilot, don’t plunge: Test AI in focused areas (like demand forecasting or inventory optimization) before rolling out enterprise wide.
- Train your teams: Upskill staff to work alongside AI and spot when human judgment is needed.
- Choose partners carefully: Work with vendors who understand CPG and retail realities—and who offer flexible, secure, and customizable solutions.
The Upshot
AI is transforming supply chain planning for CPGs and retailers. But the risks are real—and growing. By tackling data, security and transparency head-on, brands and retailers can unlock AI’s potential while protecting their business and customers.
The Bottom Line
The future of supply chain planning is AI-powered—but only the prepared will win.
The first step is to understand there are both huge opportunities and huge risks for brands moving into applying AI to their supply chain planning processes. The second step is to contact an expert in the field like Demand Chain AI.
