Every week, another enterprise announces an AI initiative. Another vendor promises transformation. Another board meeting ends with pressure to show progress. And yet, 42% of companies scrapped most of their AI projects in 2025 before they ever reached production — more than double the rate from the year before. The money is moving. The results aren't.
The organizations seeing real returns aren't doing anything revolutionary. They're doing the things that have always worked, in the right order.
At Zaelab, we work with B2B manufacturers and distributors every day on exactly this problem: CPQ implementations that stall, CRM rollouts that nobody uses, AI-powered commerce initiatives that never make it past the proof of concept. The failure pattern is almost always the same. It's not the technology. It's the sequence. This piece breaks down why over 80% of AI initiatives fail, what the organizations actually generating returns are doing differently, and the three questions every leader needs to answer before a dollar gets approved.
The Cost of Moving Fast in the Wrong Direction
Imagine you’re standing at a starting line with two cars, a fully loaded Formula 1 Ferrari and a Ford Focus, which are about to race through the winding streets of Monte Carlo. You’re now told to bet on which car you think will win the race. You may not be a car expert, but you know enough that you think the Ferrari would be the clear choice. Just as you’re about to make your wager and hand over your money, you’re told there’s a catch. The Ferrari will not be told where the finish line is, whereas the Ford Focus will be given the destination’s coordinates as well as a map of the city with the most efficient route highlighted.
Metaphorically, this is exactly what is happening with AI initiatives. Some companies are choosing to step through the tried and true process of identifying the business use case and value up front, re-defining their new processes, and then choosing a tool before implementing. Whereas others are choosing to hop in a Ferrari and go for a very expensive joy ride. This theory is substantiated by the data.
In 2025, global enterprises invested an estimated $684 billion in AI initiatives. Over 80% failed to deliver the intended business value. Meanwhile, 92% of companies plan to increase AI investment over the next three years.
We are collectively doubling down on a strategy that is failing at an extraordinary rate, and leaders are doing it under enormous pressure to show the board something. Anything.
What Happens When You Skip the Foundation
When leaders consume the promise of AI and skip the foundational work beneath it, the people expected to actually use it get starved of clarity, context, and support.
The results are predictable: passive resistance, shadow workarounds, eroded trust, and quietly abandoned initiatives. S&P Global Market Intelligence's 2025 survey of over 1,000 enterprises found that 42% of companies abandoned most of their AI initiatives, a dramatic spike from just 17% the year before. RAND Corporation's research is more pointed: more than 80% of AI projects fail, twice the rate of non-AI technology projects, and the root causes trace consistently back to leadership decisions, not infrastructure or model quality.
This isn't just a failed project. It's a cultural cost. When employees lose trust in leadership's ability to guide transformation, the consequences extend well beyond the technology budget.
For B2B manufacturers in particular, the stakes are compounded. Broken CPQ workflows, disjointed CRM data, and disconnected digital commerce experiences don't just hurt internal productivity. They show up directly in the buying experience your customers have with you — in how fast you can quote, how accurately you can price complex configurations, and how easy it is to do business with your company at every touchpoint.
What the 5% DoDifferently
BCG's 2025 research found that only 5% of organizations are generating substantial AI value at scale. Sixty percent are generating none, despite meaningful spend.
The 5% aren't smarter or better resourced. They're more disciplined. McKinsey's 2025 State of AI survey found that of all organizational changes linked to AI success, workflow redesign had the single strongest correlation with EBIT impact. Yet only 21% of organizations using AI have redesigned at least some workflows. The vast majority are layering AI on top of existing processes without rethinking how work actually flows.
This mirrors what we see across our own client work. When Panduit needed to break into a new market — Electrical Power Systems — the challenge wasn't finding an AI tool. It was that customers were unfamiliar with a complex, growing product catalog, and sales teams couldn't provide accurate quotes or tailored recommendations at speed. We helped Panduit implement AI-driven guided selling through Logik.ai, giving sales reps and distributors the ability to deliver precise recommendations and accurate quotes in real time. The result: 1 in 4 orders now includes a cross-sell. That outcome came from redesigning the selling workflow first, then deploying the AI to support it.
AI didn't change the order of operations. Define the outcome. Fix the process. Bring the people. Then deploy the tool. That sequence has always been the job. The tool just changed.
The Fix: Start Small, Show Something Real
This is where leaders find practical relief from the pressure to show progress.
The answer is not a massive, enterprise-wide rollout. It is a deliberate, contained use case where you do all the right things in the right order, but faster because the scope is manageable.
Pick something high-frequency, low-risk, and process-contained. Something with a clear before and after that any stakeholder can read in a dashboard. Do the foundational work upfront: define the measurable outcome, streamline the process, clean the data, and engage the people who will use it. Then deploy.
This is how DTFO approached their digital transformation. Rather than overhauling everything at once, Zaelab helped them migrate to a platform purpose-built for their B2C and B2B needs, redesign the customer experience around how their buyers actually shop, and modernize their commerce infrastructure to support long-term growth. The outcome: 205% revenue growth. Not from a better AI model — from a better process backed by the right technology.
One use case done properly does three things at once. It produces a measurable return. It opens a margin, financially and from a people-capacity perspective. And it builds the organizational muscle to do the next one better and faster.
That is the flywheel. Proof funds the next initiative. Margin creates the runway. A team that has successfully navigated the process once moves faster the second time because the process isn't new anymore. Only the use case is.
This is how you answer the board's pressure to show progress without gambling your budget on an enterprise deployment that skipped the foundational steps.
Three Questions Before Any AI Conversation
Before a dollar is approved, three questions need clear owners and clear answers:
What specific, measurable outcome does this solve? Not "improve efficiency." A baseline, a target, and a timeline. In B2B manufacturing contexts, that might mean: reduce average quote cycle time from five days to same-day, or increase cross-sell attach rate from 8% to 25%. Name the number.
Has the process this AI will support been mapped and streamlined? Is the data clean? Automating a broken process over bad data produces a faster broken process that spews even more bad outcomes. This applies directly to CRM data quality, CPQ configuration logic, and B2B commerce catalog accuracy — areas where we see the most expensive AI failures in manufacturing and distribution.
Who owns the people change, not the technology change? IT deploys systems. Someone else has to redesign how work gets done and bring the humans along. If that role doesn't exist in your plan, your plan isn't finished.
If those three questions can't be answered before procurement, the initiative isn't ready.
The Bottom Line of the B2B Companies Who Win with AI
The organizations compounding real returns from AI didn't bet on a single enterprise rollout. They started with one contained use case, built in the right sequence, and used the proof and the margin it generated to fund the next initiative.
For B2B manufacturers and distributors, that typically means starting where the business pain is sharpest, and the process is most defined: quoting accuracy in CPQ, pipeline visibility in CRM, or buyer self-service in digital commerce. These are the workflows where AI delivers measurable, verifiable outcomes quickly — and where a well-implemented solution builds the internal credibility to take the next step.
The organizations we've seen fail don't lack ambition or budget. They lack the discipline to do foundational work before they deploy. They pick the tool before they've defined the outcome. They go live on a process they haven't fixed. They launch without bringing the people who need to actually use the system.
Fast and reckless is still reckless. But deliberate doesn't have to mean slow or invisible.
If you're a B2B manufacturer or distributor trying to figure out where AI actually fits in your revenue operations, start with the three questions above. If you'd like a second opinion on whether your current initiative has the right foundation, we're happy to benchmark where you stand.
Sources: S&P Global Market Intelligence (2025); RAND Corporation (2024); BCG "Where's the Value in AI?" (October 2024); McKinsey State of AI (2025); IDC (2024); Gartner (2025)
