AI E-Commerce Projects Risk Failure Without Clean, Centralised Data: Report

New Delhi: E-commerce companies investing heavily in artificial intelligence may struggle to generate meaningful returns unless they first establish clean, structured and centralised data systems, according to a report by global technology firm Nisum.

The report identified data readiness as a critical factor determining whether AI initiatives progress beyond pilot projects and deliver measurable business outcomes.

It said product information, inventory records and customer data are often scattered across multiple systems, preventing AI platforms from accessing a complete and reliable view of business operations.

Poor-quality information can also magnify existing problems. Inaccurate inventory counts or duplicate customer records could be replicated at scale by AI systems, reducing the reliability of forecasts and automated decisions.

The risks become particularly significant when retailers deploy AI for personalisation, dynamic pricing, demand forecasting and inventory management.

For instance, an AI-powered marketing platform disconnected from inventory systems could promote products that are unavailable, potentially affecting customer experience and sales.

Nisum said retailers require an integrated architecture where generative AI, predictive analytics and automation operate together rather than as disconnected applications.

Centralising information across physical stores, websites and mobile applications could also help businesses move from broad customer segmentation towards more personalised shopping experiences.

The report recommended businesses identify fragmented data sources, establish ownership of data quality and implement governance frameworks before expanding AI deployments.

Nisum CEO Anurag Chauhan said only 5.5 per cent of organisations using AI currently generate tangible financial returns from their investments.

Companies addressing data readiness can move faster from experimentation to production, while businesses overlooking these foundations risk unreliable forecasts, disconnected insights and AI projects failing to achieve full-scale commercial deployment.