Whether it’s powering recommendation engines, pricing strategies, or automating customer service, artificial intelligence is pitched as the next great leap in commerce. But behind the pitch lies a sobering truth: many AI initiatives fail. And when they do, it’s not because of the algorithms. It’s because of the data.
According to a Forbes article, as much as 85% of AI and machine learning projects fail to deliver on their promises, and poor data quality is often the culprit. In 2016, Harvard Business Review reported that bad data costs businesses in the U.S. over $3 trillion annually
Distributors and e-commerce businesses are sitting on a goldmine of data: SKU-level details, transaction histories, competitors’ online pricing data, customer touchpoints, supplier data, and more. Every click, quote, order, return, and reorder adds to the pile. On the surface, this seems like an ideal foundation for AI.
But many are discovering that abundance isn’t the same as quality and readiness.
Much of the data is siloed, unstructured, outdated, or inconsistent. Product descriptions are incomplete. Customer records are duplicated or mismatched. Pricing data lacks hierarchy or logic. Sales data may be riddled with overrides and exceptions. When companies try to feed this into machine learning models or pricing engines, the results are unreliable. AI ends up amplifying the flaws instead of solving them.
In this context, data becomes both an asset and a liability. It’s the raw material for transformation, but it also highlights years of neglect and fragmentation. Companies often find that before they can implement AI solutions, they have to clean house.
Digital transformation begins with data discipline. That means defining standards, cleaning historical records, rationalizing product catalogs, and enforcing data governance. It’s tedious, often thankless work. It doesn’t get the same attention as AI pilots or shiny dashboards. But it’s the precondition for everything that follows.
Distributors that succeed in using AI and analytics typically have a strong culture of data stewardship. They treat data as infrastructure. They invest in cleansing and classification, and they apply consistent rules across functions, sales, operations, pricing, and marketing. They don’t just use data; they manage it.
While traditional digital commerce platforms have focused on product availability, order speed, and convenience, the next wave of innovation is being built around AI in pricing.
The companies that get it right approach data not as a one-time cleanup, but as a long-term capability. They don’t outsource the responsibility to IT alone. Instead, they involve business leaders, functional experts, and data professionals in shaping the structure and rules that define what “good data” looks like.
They also embed data practices into everyday workflows. For example, when pricing teams override a system recommendation, that feedback loop becomes a data input. When product managers improve descriptions or categorization, they contribute to future automation efforts.
Every improvement in data quality compounds over time and enables smarter decisions tomorrow.
Before a distributor can even think about adopting AI, there’s a groundwork that often takes longer, and requires more cross-functional effort, than initially expected. This isn’t about perfection. It’s about creating enough consistency and structure in your data to make it usable, trustworthy, and scalable. The reality is that most distribution businesses underestimate just how much time and effort this phase demands.
Starts with a detailed audit that identifies where the data lives, who owns it, how it’s used, and where the gaps or inconsistencies are. This includes everything from customer master data to product attributes, transaction history, pricing records, vendor terms, and channel performance.
Once the audit is complete, the real work begins. Cleaning and rationalizing product data alone can take months: Standardizing descriptions, applying categories consistently, aligning units of measure and packaging formats. Often, legacy product hierarchies no longer reflect how customers actually buy or how sales teams actually quote.
Customer data is no less challenging: Resolving duplicate, mapping parent-child relationships, linking buying behaviors to the right accounts. In many cases, this requires manual review and validation, especially for key accounts that have evolved through years of acquisitions or informal sales practices.
Transactional data also needs structure. Distributors often find that the same SKU is priced differently for no clear reason. Or that price overrides and exceptions dominate the history, making it hard for any AI model to learn meaningful patterns. Without clear rules and guardrails around how pricing decisions are captured, models will struggle to interpret what’s valid versus what’s noise.
For most distributors, this process is a multi-quarter effort involving pricing, sales ops, IT, category managers, and finance. It requires dedicated resources, clear accountability, and an understanding that no AI investment will pay off unless this work is prioritized and funded first.
A look at the most impactful ways organizations can use generative AI to improve customer experience and how they can get started.
Getting the data ready isn’t about slowing things down; it’s about giving AI a chance to work. Skipping or rushing through this foundational phase is one of the biggest reasons why AI fails to deliver value in distribution. The companies that take the time to do it right are the ones who ultimately see a return that justifies the hype.
AI isn’t magic; it’s math powered by data. And that data must be trustworthy, structured, and continuously maintained. Distributors and e-commerce firms already have the data volume. Their challenge now is to create the data quality.
Investing in AI before investing in data readiness is like building a skyscraper on sand. If modern commerce wants to unlock the promise of smart systems, smarter pricing, smarter inventory, smarter selling, it needs to begin with the unglamorous but essential work of getting the data right.
