Click for Takeaways
- Phase Zero Is Non-Negotiable: The single strongest predictor of ERP success is whether a team truly understands its processes before the project begins. Not just what’s broken, but why. Roughly 70% of ERP systems implementations fail to meet their objectives, and the root cause is rarely the technology.
- FP&A Has to Advocate for Its Own Reporting: Reporting requirements get acknowledged in ERP projects and then deprioritized until go-live, when it is too late. If FP&A doesn’t create its own workstream and hold the implementation team accountable for specific outputs, it will be left behind.
- You Cannot Automate What You Haven’t Defined: The same process gaps that sink ERP implementations sink AI initiatives. Up to 95% of enterprise AI budgeting tools and generative AI projects fail to deliver measurable ROI, and the root cause is knowledge infrastructure, not model quality. The fix is the same: document the process before you build on it.
- Data Readiness Is Rarer Than Anyone Admits: Research shows only 7% of enterprises describe their data as completely ready for AI, a gap that affects everything from ERP systems to financial forecasting methods. The gap between governance, timeliness, and ownership is where both ERP and AI investments quietly disappear.
- ROI Framing for Transformative Infrastructure Is Wrong: The value of a well-implemented ERP or AI foundation is not measured in hours saved. It is measured in the ability to scale, enter new markets, and avoid the compounding cost of rework. Small fixes today are large, expensive problems tomorrow.
Based on an interview with Cindy Vindasius, Founder and CEO of Vindasius Advisory
Cindy Vindasius has done 12 ERP implementations. She will tell you she loves it. She will also tell you, with the specificity of someone who has seen the same failure patterns dozens of times, exactly why most organizations get it wrong.
As founder and CEO of Vindasius Advisory and a former Corporate Controller, Cindy has spent more than 30 years helping high-growth and enterprise companies build finance digital transformation programs and ERP systems that actually scale. Her work spans NetSuite, SAP, and Oracle implementations, multiple IPOs, M&A transactions, and global compliance programs. In a conversation with FP&A Today host Glenn Hopper, she unpacked what phase zero really means, where the line is between a true ERP and a modern GL with a good UI, why data readiness remains the bottleneck for both ERP and AI, and how finance leaders should think about the ROI of infrastructure that is too important to measure conventionally.
The 70% Problem
The failure rate for ERP implementations has been quoted for years. Cindy doesn’t dispute it.
“I have analyzed it five ways from sideways. Most of the companies I’ve been with are high-tech Silicon Valley companies, underfunded, going public, and need to get on a system. They don’t have tons of dollars, and they don’t have a lot of time. They just need to get it done.”
According to Gartner’s analysis, 70% of ERP systems implementations fail to meet their objectives, with organizations reporting average cost overruns of 189% — one of the most documented failures in finance digital transformation. The financial damage extends beyond initial implementation costs into operational disruptions that can take years to recover from.
The reasons Cindy cites are consistent and predictable: teams aren’t ready, processes aren’t understood, and the internal lift required from finance and operations teams is massively underestimated.
“You’re added to an ERP implementation, the system implementation provider is scheduling meetings, running the project. And then there are all these things they’re not managing that the teams have said will be internally managed: data cleanup, testing, process understanding, redocumentation, and change management. And the finance people are like, ‘Okay, I can participate in your meetings, but I still have a full-time job.’ And then everyone else is like, ‘Yeah, but we’ve got a timeline.'”
What follows is predictable: feature sets drop out, scope narrows, the team drives through the timeline, and then the rework begins because the original goals were never actually met.
Phase Zero: The Foundation Everyone Skips
The fix Cindy recommends isn’t complicated, but it requires patience that underfunded, pre-IPO companies rarely have.
“Phase zero means you understand your processes. You have clear requirements that aren’t just ‘fix this broken piece.’ It’s: why did that happen? Oh, because fundamentally the difference between your CRM and your ERP and the data exchange is because they’re on different platforms. That kind of clarity really helps you become successful because you actually know what you’re doing.”
Phase zero isn’t a checklist. It is a genuine reckoning with the gap between how the business thinks it works and how it actually works. For teams that skip it, the cost shows up later, at a higher price.
Glenn Hopper put it plainly:
“If somebody’s going from QuickBooks to NetSuite, what happens when you try to put that technology on top of broken processes, bad data?”
Cindy’s answer was direct:
“Or the QuickBooks structure. This is 90% of my clients in Silicon Valley. They’re like, ‘I don’t have time, I’m gonna take my QuickBooks picture and shove it to NetSuite,’ – skipping the budgeting and forecasting in Excel cleanup that should happen before any migration. And then there’s this whole rework project because you’re using the wrong native activities of the new system. You’re jamming in bad processes and doing workarounds where you need to understand your core systems’ capabilities. That’s what allows things to scale.”
The rule of thumb she offers for timing: the best moment to fix process and data problems is during a transition. There is no easier time. And rework is always more expensive than implementation work.
What Actually Qualifies as an ERP
The market has blurred the definition. Cindy has a clear one.
“These AI ERPs are GLs with a fragmented, or I would say distributed, system. But they’re awesome.”
Her distinction is not dismissive. She has met with many of the AI-native finance platforms, followed their roadmaps, and talked to their teams. For the right use case, she is genuinely enthusiastic.
“I would highly consider one if you’re a company that has SaaS, never going to have inventory, and your selection team is run by finance, because it’s a finance application for finance. Crypto is another one. Any application with fair market value constantly revaluing assets, these are great.”
The constraint is inventory. Once a business has inventory, procurement, and logistics complexity, the math changes entirely.
“If you’re doing an AI ERP, it’s obviously faster because you have decentralized systems, you’ve got your vendors managed somewhere else, your customers managed somewhere else. But they’re not really working with you on process changes. They’re defining their implementation to get you off your current platform and onto the new one. You’ve gotta take that into effect when you’re thinking about timelines. If you’re choosing your ERP based on the timeline, that might not be the best metric.”
The battleships, as Cindy calls the legacy platforms, were not motivated to change for years. That is shifting now. Oracle, SAP, and NetSuite are all building AI into their core ERP systems, including MCP integrations, embedded language model capabilities, and financial data visualization tools. The conversation about the speed of implementation is real. The conversation about depth of process change is equally real, and still being skipped.
Data Readiness: The Gap Nobody Wants to Own
If phase zero is the foundation for ERP, data readiness is the foundation for everything else, including AI. Cindy’s definition has three components: reliability, governance, and timeliness.
“You’ve got to have reliable and accurate data that’s governed. There are controls and somebody owns it so that if something goes wrong, you’ve got one throat to choke. But the thing I don’t hear often that also impacts reliable data is timeliness. You could have the best data, but if revenue statistics are only available once a month, 20 days after close, that’s not reliable data. The invention of AI is not only extracting information, it’s using AI to eliminate those manual processes and identify where you’ve got timeline lags.”
The current state of enterprise data readiness reflects how far most organizations are from this standard. A joint study by Cloudera and Harvard Business Review Analytic Services found that only 7% of enterprises describe their data as completely ready for AI adoption or meaningful finance digital transformation, with more than a quarter reporting their data is not very ready or not at all ready.
Cindy’s framing connects directly to why: data governance is not sexy. It doesn’t ship a product. It doesn’t close a deal. And it is routinely deferred until a system implementation or AI project forces the reckoning.
“You can’t automate something you haven’t defined. If you’re looking at AI readiness, I want you to look at your pain points. I don’t want you to do a million tools. I want you to land on one or two.”
Why FP&A Has to Advocate for Itself
One of Cindy’s most pointed observations is about a specific failure mode that plays out in almost every ERP implementation: FP&A’s reporting needs get acknowledged, deprioritized, and then dropped.
“Reporting is one of the key drivers for why you do any of these digital transformation changes. But what happens is IT or the system implementers will take that requirement and go, ‘Great, yeah, we have that requirement, but there’s no work stream for it.’ A lot of times, they’ll say, ‘Yeah, we’ll get there,’ but it doesn’t come until the last validation step, once they’ve configured everything and you’ve put in test data. Then the rest of the world says, ‘Well, great, we’re ready to go live.’ So you go live, and you don’t have your reports.”
The solution is not to trust the implementation team to remember.It is to create a separate FP&A workstream focused specifically on financial reporting software outputs and the data that feeds them, to show the implementer exactly how each calculation is done and where the data is coming from, and to hold that thread from the beginning of the project through go-live.
The ROI Problem
Cindy is direct about the ROI question: the value of a well-implemented system is not in labor efficiency. It is in the ability to scale, to enter new channels, to handle new pricing structures, to serve enterprise customers who require capabilities your current system cannot support.
She gives a real case: a company where 95% of orders came in the last week of every quarter, driven by a broken pricing structure. The fix required two quarters of missed public numbers to retrain people and retool the system. How do you calculate ROI that? How do you ROI the alternative, staying on the broken model indefinitely?
“How do you ROI of the inability to scale or the inability to go into new markets? There’s gotta be some intangibles of being able to scale and doing it right the first time.”
The parallel to AI is the one Glenn and Cindy return to repeatedly. The MIT NANDA Initiative has documented that up to 95% of enterprise generative AI and AI budgeting tool projects fail to deliver measurable ROI. The core issue, the research finds, is not model quality. It is a knowledge infrastructure: the layer of data governance, process documentation, and systems integration that sits beneath the AI and has to be right before the AI can deliver anything.
The ERP failure literature says the same thing. The AI failure literature says the same thing. The lesson is the same: the infrastructure investment comes first, the ROI follows, and framing it as a line item on a spreadsheet is the wrong tool for the job.
AI, Audit, and Where the Guidance Is Heading
Compliance is one of the three main objections Cindy hears from finance leaders hesitating on AI adoption. The others are ROI (covered above) and data privacy.
Her read on the audit landscape comes from an unusual vantage point: both of her children are CPAs at Ernst and Young, and the dinner table conversation has been substantive.
“I think we’re gonna see a lot more information in the next six months because the firms are working out their stance and how they’re going to approach it within the audit framework. They are using AI internally to help them with their processes, which my kids think is awesome. But from a client perspective, the auditors are already asking where AI is being used, because if it’s used heavily, that’s going to complicate the control environment.”
The guidance she gives clients is practical:
“You need to document what you’re prompting. You need to be able to show where it’s pulling from and how it’s doing that calculation. You need to prove you’re still using human-in-the-loop or have compensating controls. And you need to know what has changed since you ran it. It’s a new area. I wouldn’t just dive in and add a bunch of AI because you might find yourself in a control problem.”
The broader point she makes is that AI embedded within a major ERP platform carries a different auditability profile than a standalone chatbot doing finance work. When the calculation lives inside NetSuite or SAP, there is a system of record around it. When it lives in a chat window that closes, there isn’t.
What to Do Before Phase Zero
When Glenn asks what finance leaders should be doing right now, before any implementation kicks off, Cindy’s answer is a short list with a clear priority order.
“You want to put an AI or ERP readiness review in place. It is understanding your data readiness infrastructure: your master data, your governance controls, your security access. Where are the gaps and what do you need to fill?”
On process documentation specifically, she offers a test that applies equally to ERP and AI:
“If those projects are not documented well enough that you can hand them off to a third party to do without your assistance, they’re not ready for AI. Then maybe break them down into smaller chunks. If you can’t hand it off, that’s the rule of thumb.”
And for anyone still unconvinced that preparation time is worth it:
“I sat at NetSuite SuiteWorld and heard people saying eighteen months is crazy for a readiness timeline. And I thought, ‘That’s exactly what you need to get your data in a place where you can use the functionality.’ They should immediately say, ‘Okay, everyone, here’s what you do while you’re waiting.’ Well, no one’s gonna do it. It’s not sexy, it’s not exciting. They don’t understand how much the data plays into it.”
Where Datarails Fits In
The challenges Cindy describes throughout this conversation are the operational backdrop against which FP&A teams work every day: ERP data that doesn’t surface the right KPIs, reporting that falls off the implementation timeline, forecasting models built on data that is days or weeks stale, and AI ambitions that stall because the underlying data infrastructure isn’t ready.
Datarails is the AI-powered FP&A platform built for Excel users. It consolidates financial data from ERPs, accounting systems, and spreadsheets into a single source of truth without forcing finance teams to abandon the tools they already know. It bridges the gap Cindy identifies between what ERP systems produce and what FP&A actually needs: real-time variance analysis, dynamic scenario modeling, and narrative generation that turns raw data into insight without waiting for the month-end cycle.
For finance teams in the middle of an ERP systems transition, Datarails preserves continuity through proven financial consolidation tools: the reporting layer keeps working while the system underneath is rebuilt. For teams whose ERP never delivered the reporting they needed, it fills that gap without requiring a new implementation. For organizations building toward finance digital transformation and AI readiness, Datarails provides the governed, structured financial data layer that Cindy identifies as the non-negotiable prerequisite.
The goal Cindy describes, doing it right the first time, is what good FP&A infrastructure makes possible.
To learn more about how Datarails supports FP&A teams at every stage, visit datarails.com.
About Cindy Vindasius
Founder and CEO of Vindasius Advisory and former Corporate Controller, with more than 30 years helping high-growth and enterprise companies build scalable finance systems. She has led more than 12 ERP implementations across NetSuite, SAP, and Oracle, and supported multiple IPOs, M&A transactions, and global compliance programs. Her ERP Mastery Program, a six-module course covering phase zero through cutover, is available at vindasius.com/mastery-program.
Connect on LinkedIn.
FAQs
The pre-implementation stage where a team documents its processes, maps data flows, and defines what the ERP systems or AI tools must do before a single configuration decision is made. Skipping it is the most common reason ERP projects fail. It applies equally to AI initiatives.
AI-native finance platforms are purpose-built general ledgers with modern UIs and fast implementations. They work well for SaaS and inventory-free businesses. Traditional ERPs handle inventory, procurement, and logistics at scale. The distinction matters if your business has operational complexity.
Data should be accurate, governed, and timely, the same standard that applies whether you’re implementing ERP systems, deploying AI budgeting tools, or improving your financial forecasting methods. If key metrics are only available 20 days after the month-end, the data is not reliable for AI purposes. If no single person owns a data domain, there is no control. Start there before evaluating tools.
