Why will AI expose data weaknesses more effectively than traditional systems?

Traditional systems often rely on human intervention to compensate for data gaps or inconsistencies. AI systems, especially those built on machine learning, require structured, consistent, and complete data to learn and operate accurately. Flaws that humans can intuitively address become direct points of failure for AI.

What types of data issues are most problematic for AI in ecommerce?

Key issues include inconsistent product descriptions, incomplete attribute sets, ambiguous category assignments, fragmented customer data, and conflicting inventory information. Any data point that lacks standardisation or accuracy will hinder AI's ability to make reliable recommendations, answer queries, or automate processes.

How can a PIM system help prepare ecommerce data for AI?

A Product Information Management (PIM) system centralises, standardises, and enriches product data. It ensures a single source of truth for all product attributes, descriptions, and media, making the data consistent and accurate across all channels. This clean, structured data is ideal input for AI algorithms.

What is data governance, and why is it crucial for AI projects?

Data governance involves defining data ownership, establishing data standards, and implementing processes for data cleansing, validation, and enrichment. It is crucial because AI's effectiveness depends entirely on data quality. Good governance ensures the data AI consumes is reliable, thereby enabling trust in AI-driven outcomes.

What should be the first step for a company looking to improve data for AI?

The first step should be a comprehensive data audit. This involves evaluating the current state of all relevant data-product, customer, transactional-to identify inconsistencies, gaps, and areas needing immediate attention. This assessment will inform a targeted clean-up and improvement strategy.

Talk to an expert

Issue 036PositionAI for CommerceRef 125

Why AI will expose weak ecommerce data faster than ever

The shift towards AI adoption in ecommerce will swiftly highlight foundational data weaknesses. Poor data quality, often a silent drain on operations, will become a bottleneck for any AI initiative.

Heddwyn Coombs· Creative and Digital Director

23 Jun 20255 min readReviewed June 2025

"AI is not a magic bullet for poor data; it is an accelerator, exposing and magnifying existing issues."

Written by

Heddwyn Coombs

Creative and Digital Director

29 years at iWeb

LinkedIn →Speak with Heddwyn about your brief →

Heddwyn co-founded iWeb and leads creative and digital direction across the agency. Adobe Commerce architect since the Magento 1 era, he writes the strategy notes on platform choice, headless, agentic checkout, AI for commerce and the trade-offs leadership teams hit when modern tooling meets operational reality. Opinionated about data structure, design systems and catalogue size, cautious about unnecessary frontend complexity, and focused on work that stays commercially manageable over time.

Next step

Take this conversation private.

If this note matches the brief on your desk, the version we'd write for your platform, sector and trade-account model is the work itself. Send a short brief and a senior lead replies inside 48 hours.

Talk to an expert Back to The Record →

Operational notes worth reading next.

Conversational commerce in B2B: the use cases that pay back inside one quarter

Where merchandising teams should refuse to hand work to an LLM

Evaluation harnesses for commerce LLM features: the bit nobody ships

A regular round-up of new notes.No content marketing.

Take this conversation private.

A regular round-up of new notes.
No content marketing.