The Cost of Standing Still: Why 94% of Companies Never See AI Returns
The Cost of Standing Still
Second in a four-part series on AI, ERP data, and what practical intelligence looks like for mid-market manufacturing, distribution and construction operations.
Read Part 1: The AI Conversation No One Is Having With Manufacturers.
There’s a kind of waiting that feels responsible.
You’ve watched the AI conversation get louder every quarter for two years. You’ve sat through vendor pitches that promised transformation and delivered confusion. You’ve read enough headlines about companies betting hundreds of billions on AI to figure the smart move is to wait until the noise settles.
For a lot of what passes as AI in 2026, that’s probably right.
But there’s a specific kind of waiting that isn’t safe. And it costs more than most people think, because it never shows up in one big moment. It adds up quietly.
Want to see what practical AI actually looks like on a real ERP, not a slide? Join us live for the webinar AI ERP Insights: Cut Through the Noise with InsightsAI — July 29 at 11:00 AM ET.
The Gap Isn’t Closing
88% of companies now use AI in some form, but only 6% see real financial returns. The 6% aren’t using different tools. They’re using the same ones with a different approach. They started earlier, built visibility into how their people decide, and every step now is easier because the groundwork is done.
The distance between that group and everyone else is widening.
For a mid-market manufacturer or distributor, that gap doesn’t look like a competitor running robots or autonomous agents across the supply chain. It’s much smaller than that, which is exactly why it’s easy to miss.
It looks like a finance leader at a competitor who gets an alert Tuesday morning that a product line’s margin has slipped for six weeks, and fixes it before it hits the monthly P&L.
It looks like an operations manager who asks which jobs are over labor estimate this month and has the answer in seconds, not two days later after a report request.
It looks like a CEO who walks into the board meeting already knowing the quarter’s numbers instead of learning them in the room.
None of those decisions is dramatically different on its own. But the speed, the confidence, and the consistency across months and years add up to a gap that’s very hard to close once it’s there.
What Your Peers Are Saying
We hear versions of this from our own customers constantly. When we surveyed 200+ operations leaders earlier this year, the readiness picture was stark.
are planning AI investments this year. Their average readiness score: 4 out of 10.
put themselves at 3 out of 10 or lower on AI readiness.
Those two numbers side by side should stop you. A quarter of the executives we surveyed are budgeting for AI they don’t feel ready to run. That isn’t skepticism. It’s real willingness colliding with real uncertainty about where to begin.
When we asked what they actually need, this came back:
“Needed BI tools to make a visual picture of where parts are at. Material planning window only shows one part and there is no quick visuals to see when stock may be consumed.”
Manufacturing Executive · WM Synergy 2026 Voice of the Customer Report
Sit with that for a second. This person isn’t asking for AI. They want to see where their parts are. It’s a basic operational question their ERP already has the data to answer, and they can’t get to it without building something or waiting on someone.
“Better reporting fields, easier ability to make custom reports without the need of gurus.”
Vice President, Operations · WM Synergy 2026 Voice of the Customer Report
Gurus, again. The data is there. Reaching it takes a specialist, and the specialist is the bottleneck. Every decision that leans on that data slows down. Every week it isn’t fixed, the cost grows.
What Keeps People From Moving
When we ask operations leaders what’s holding them back, three answers come up again and again.
They don’t know where to start. The market’s AI conversation is pitched at companies with data scientists, cloud infrastructure, and engineers. It isn’t aimed at a mid-market manufacturer with 200 employees in Ohio or a regional distributor in the Pacific Northwest.
They aren’t sure their data is ready. There’s a common belief that you need a long data cleanup project before analytics can do anything useful. That’s half true. You don’t need perfect data to start seeing value. But you do need to know where your data is wrong, and most teams have never had a way to find out. The fastest way to learn what needs cleaning is to start looking at the data you already have.
They’ve been burned. A BI project that went nowhere. A tool that needed more IT than anyone expected. A vendor who sold them something and was gone six months later. That history shapes every decision that comes after.
Every one of those is real. None of them is a reason to stand still forever.
Bad Data Isn’t a Reporting Problem. It’s a P&L Problem.
There’s a second cost, and it hides inside the ERP. When the data under the reports is wrong, the reports are wrong, and so are the decisions built on them.
One customer carried more than $1 million in inventory variance on the books before anyone reconciled it. Real money on the balance sheet that wasn’t actually there. Finance teams routinely burn 15 hours a week pulling and reconciling ERP data by hand in spreadsheets, doing work the system should do on its own. Audit findings stack up: negative cost layers, unposted GL entries, WIP that won’t roll forward. And the quietest cost of all — leaders stop trusting the numbers and start going with their gut.
Pointing AI at data nobody’s checked doesn’t fix the data, and used carelessly it speeds up the wrong answer. That’s why a visibility layer has to do two things at once: give you the answer, and show you how much to trust it. AI that just answers, without ever flagging what it doesn’t trust, gives you confident, polished, wrong answers faster than ever. AI that surfaces its own gaps is exactly how you find out what’s wrong before it costs you.
What Waiting Actually Costs
The cost of not seeing your ERP data in real time is rarely one dramatic failure. It’s a hundred small moments that each look manageable on their own.
It’s the margin slide on last month’s report that actually started six months ago, and nobody caught it because the reporting was too slow.
It’s the delivery problem that turned into a customer complaint before anyone saw it in the data.
It’s the weekly leadership meeting where you talk about the numbers but not the causes, because getting to the causes means a report request that takes two days, and the meeting was yesterday.
Over a year or two, those moments add up to real cost. Time, margin, leadership confidence, and competitive ground that erodes quietly while the companies on the other side of the gap make better decisions faster.
The Question Worth Asking Now
Do I still need an AI strategy?
Yes. You still need an AI strategy. What you don’t need is to have it all figured out before you take the first step.
Your ERP is already your single source of truth. It just hasn’t had a way to talk back. Put a practical AI layer like InsightsAI on top of that data and two things happen at once. You get answers to the questions your team is asking right now, and you get a clear picture of exactly where the gaps and weak spots are, the inventory that doesn’t reconcile, the reports nobody trusts, the numbers that take two days to pull.
Frequently Asked Questions
The gap isn’t the tools, it’s the foundation. The 6% seeing real returns built data visibility into how their teams make decisions before layering AI on top. Companies that skip that step end up pointing AI at data they can’t trust, which produces fast answers to the wrong questions.
It rarely shows up as one dramatic failure. It’s margin slides that go uncaught for months, delivery problems that become customer complaints before anyone sees them in the data, and leadership meetings spent discussing numbers instead of causes. Those moments compound over a year or two into real, hard-to-close competitive distance.
Not entirely. You can start getting value from the data you already have. But most ERPs carry years of drift — duplicate records, inconsistent codes, unclosed orders — and the fastest way to find out what needs fixing is to start looking at the data through an analytics layer, not to wait for a perfect dataset first.
Not better tools. Earlier starts, and decisions built on trusted visibility instead of gut instinct. Every step after that first move gets easier because the foundation is already in place.
Sources: WM Synergy 2026 Voice of the Customer Report · NVIDIA State of AI Report 2026 · top10erp.org, “AI in ERP: The Next Wave” (2026)
By Paul Tedford
By Jessica Gadbois
By WM Synergy