- 40% of B2B buyers already use AI agents for procurement decisions (Deloitte).
- Only 24% of suppliers have AI agents on the sell side.
- 51.3% product data completeness industry-wide (Distribution Strategy Group).
Experts agree that manufacturers must urgently adopt AI-driven product data management to avoid exclusion from AI-assisted procurement processes.
The Invisible B2B Shelf: How AI Agents Are Exposing Broken Product Data
BOSTON – November 18, 2026 — For decades, dominance in industrial manufacturing and B2B distribution relied on a familiar triad: historical brand equity, massive installed bases, and deep-rooted relationships with distributors. A buyer needed a part, called their sales rep, and cut a purchase order. Today, that deeply human process is being rapidly dismantled by lines of code.
According to a new research paper published today by Product Experience Management (PXM) provider Salsify, a decisive and dangerous gap is opening between B2B buyers and suppliers. The report, titled "Product Data Is the Bottleneck: How Manufacturers Clear It to Grow," outlines a stark reality for the industrial sector: artificial intelligence is taking over the procurement process, and most manufacturers simply do not have the data infrastructure to participate.
The numbers, corroborated by recent industry analysis, are sobering. Research from Deloitte indicates that nearly 40% of B2B buyers are already utilizing agentic AI to make purchasing decisions—deploying autonomous software to research, shortlist, and select suppliers. Conversely, only 24% of suppliers have deployed AI agents on the sell side to meet them.
This asymmetry creates what industry insiders are calling the "invisible shelf." In traditional ecommerce, poor product data might result in a lower search ranking. In agent-assisted purchasing, the penalty is absolute. If a manufacturer's catalog lacks machine-readable specifications, dimensional data, or compliance certifications, the AI buyer agent does not rank the product lower. It simply returns nothing. The supplier is entirely excluded from the consideration set before a human buyer ever sees a shortlist.
The Structural Impossibility of Manual Management
The root of this crisis lies in the sheer volume of data required by modern commerce channels. Distribution Strategy Group reports that industry-wide product data completeness currently sits at a dismal 51.3%. For organizations with less than 10% digital revenue penetration, that number often plummets below 40%, plagued by missing images, incomplete specifications, and generic text.
"Every manufacturer at every AI conference this year is talking about the product data problem,” said Mandy Earle, Head of Industrial Markets at Salsify and co-author of the new report. “This paper is the answer to the question they're all circling: what does solving it actually look like, at the scale modern distribution demands?"
For years, manufacturers have treated incomplete catalogs as a localized productivity issue, throwing offshore headcount or temporary contractors at the problem. However, as Jeriad Zoghby, Joan Smith, and Roald van Wyk note in their book Competing at Machine Speed, this approach is fundamentally flawed. "The industry has been calling this a productivity problem. It's not. It's a structural impossibility," the authors write. "The surface area of modern commerce, at the scale that major brands operate, is larger than any human workforce can maintain at the cadence the channels now require."
This structural impossibility is precisely what Salsify's proposed architecture attempts to solve. The Boston-based software firm advocates for an intelligent Product Information Management (PIM) framework divided into four distinct layers. It begins with a System of Record for governed, foundational data. This feeds into a System of Action, where AI handles the heavy lifting of enriching product attributes at scale. A System of Delivery then pushes this optimized data across various channels to interface with buyer agents, while a System of Intelligence provides continuous learning and optimization based on market feedback.
The underlying philosophy of this architecture is simple but transformative: let artificial intelligence handle the crushing volume of data ingestion and formatting, while human employees handle nuanced judgment and strategic oversight.
Machine-Speed PIM in the Wild
The theoretical benefits of automated data governance are compelling, but the bottom-line impact is where the strategy proves its worth. Salsify’s report highlights the operational transformation of Wastequip, a manufacturer of waste handling equipment.
Managing a sprawling catalog of 60,000 SKUs, Wastequip faced the classic industrial dilemma: an immense product matrix managed by a lean ecommerce team of just three people. By implementing an automated, AI-driven PIM architecture, the manufacturer fundamentally altered its cost-to-serve and speed-to-market. Daily product naming conventions and data reviews that previously required an hour of manual effort were reduced to a 10-minute oversight task.
More importantly, this foundational data cleanup translated directly into revenue. In its first year operating on this new architecture, Wastequip reported a 40% growth in ecommerce revenue and an 83% surge in organic traffic. By transforming their catalog from a static, incomplete spreadsheet into a dynamic, machine-readable asset, the company ensured its products were visible to both human search engines and autonomous procurement bots.
This shift represents a crucial evolution in how industrial brands must view their digital assets. Data is no longer just marketing collateral; it is the product itself in the digital realm. As Jason Hein, Vice President at the B2B Ecommerce Association, observed in the report: "Product data has never been more foundational than it is in a world of AI because it represents reality."
The 2028 Negotiation Cliff
The window for manufacturers to modernize their data infrastructure is rapidly closing. While the current threat involves being excluded from AI-generated shortlists, the next phase of agentic commerce will be far more disruptive.
Gartner projects that by 2028, AI agents will handle 30% of routine sales negotiations. This forecast pushes the timeline for digital transformation from a long-term strategic goal to an immediate operational mandate. Software and service providers are already redesigning workflows to accommodate autonomous decision-making, where buyer bots will negotiate pricing, terms, and delivery schedules directly with supplier bots.
If a manufacturer's product data is too unstructured to pass the initial screening phase today, they have no hope of participating in the autonomous negotiations of tomorrow. The asymmetry noted by Deloitte—where 72% of suppliers falsely believe their sales operations are highly automated, while only 47% of buyers agree—is costing suppliers real money. Deloitte estimates this perception gap contributes to a 13% loss in sales bids due to friction-filled buyer experiences.
To combat this, Salsify’s research paper closes with a "90-Day Mandate," outlining five immediate actions manufacturing executives must take to put AI-ready product data on the C-suite agenda. These steps range from conducting mandatory, unvarnished audits of current product data quality to establishing cross-functional AI governance structures that bridge the gap between IT, sales, and marketing.
The transition to agentic commerce is not waiting for legacy manufacturers to catch up. In an environment where procurement software evaluates thousands of SKUs in milliseconds, the strength of a brand's legacy matters far less than the completeness of its digital footprint. For the industrial sector, the mandate is clear: digitize and structure catalog data at machine speed, or risk fading entirely from the digital shelf.
Topics & Related
Manufacturing & Industrial
Agentic AI
📝 This article is still being updated
Are you a relevant expert who could contribute your opinion or insights to this article? We'd love to hear from you. We will give you full credit for your contribution.
Contribute Your Expertise →