Andy and Brian are joined by eCommerce expert Howard Blumenthal to talk about how companies should think about the value of their AI initiatives. Howard lays out a useful set of questions for preparing for new AI projects:
- Are we looking at the right workflows for AI?
- Are we measuring the right things?
- Is our data ready?
- Do we have a governance plan?
Plus, learn how B2B eCommerce leaders are approaching the 4 hottest AI categories.
Q: What is the current failure rate for enterprise AI projects? A: The numbers are sobering. According to data cited in the episode, AI projects have an 80% failure rate and are two times more likely to fail than traditional IT projects (per RAND Corporation). Only 5% of enterprise AI pilots currently deliver measurable impact, and just 1 in 10 proof of concepts ever reach full-scale production. Gartner predicts 40% of agentic AI projects will be cancelled by 2027.
Q: Why are so many AI projects failing in B2B? A: Three major factors were identified. First, data unreadiness — Gartner predicts 60% of projects fail because data is too noisy or siloed. Second, the “black box trust gap” — 51% of organizations have stalled deployments due to hallucinations and inaccurate outputs. Third, the “hype trap” — 30% of generative AI projects are abandoned post-proof-of-concept due to escalating costs. As Howard Blumenthal put it, “E-commerce uncovers all the sins of a company. AI does that on steroids.”
Q: What are the four key questions companies should ask before starting an AI project? A: Howard Blumenthal’s framework centers on four questions: (1) Are we looking at the right workflows for AI? AI works best on high-volume, repetitive tasks. (2) Are we measuring the right things? Focus on outcomes, not outputs like how many emails or graphics were produced. (3) Is our data ready? You don’t need to clean everything — just ensure the data within your specific use case is solid. (4) Do we have a governance plan? The most successful model combines top-down structure with bottom-up flexibility for departments to experiment.
Q: What is “AI creep” and why is it a problem? A: AI creep is when employees start using AI to venture outside their core competencies — a content writer starts doing graphic design, a graphic designer starts writing code. While AI makes this feel possible, MIT research shows that the time people spend on these tangential tasks can consume all the efficiency gains from their core work. The solution is ensuring the right people are doing the right tasks and maintaining human-in-the-loop oversight across departments.
Q: What are the four “hot categories” where B2B companies are seeing real AI wins? A: Blumenthal identified four areas delivering results: (1) Search — using semantic and vector search to improve product discovery beyond basic keyword matching. (2) Personalization — but tailored for B2B needs like reorders, cross-sells, and quotes rather than following B2C behavioral patterns. (3) Pricing — adding AI layers to existing BI tools, with companies seeing 2-5% margin improvements. (4) Operations — supply chain forecasting, demand planning, and order exception handling, which benefit from AI’s strength with high-frequency, repetitive tasks.
Q: Should B2B companies focus on ROI or experimentation with AI right now? A: According to the Master B2B community poll, 56% said both broad experimentation and driving ROI are important. But among those who chose one, experimentation outpaced ROI by 3-to-1 (33% vs. 11%). Blumenthal agreed, comparing it to the early days of e-commerce: “People need to get a grasp of it. They need to start using it so they can actually understand it. But you can’t do that forever — you have to start to pick and choose the items that will give you ROI.”
Q: How is the AI moment similar to the early days of e-commerce? A: Multiple parallels were drawn. Like early e-commerce, companies are measuring the wrong things (hits vs. outcomes). Like e-commerce, pushing too hard for immediate ROI on infrastructure investments is premature. And like e-commerce, the real breakthrough won’t come from copying existing processes more efficiently — it will come from rethinking how the business operates entirely. Andy Hoar compared it to the steam engine-to-electric motor transition: “It took 30 years for companies to figure out they got to change how they approach things. And when they did, we got the assembly line.”
Q: What was the breaking news about Genuine Parts Company? A: Genuine Parts announced it will separate Motion (industrial distribution) and NAPA (automotive parts) into two independent public companies. The combined entities represent approximately $24 billion in revenue. The move is driven by activist investors who believe the parts are worth more than the whole — a growing trend as software and AI reduce the back-office synergies that previously justified combining companies.
.
