What is your micro-vertical, and how much customization will your ERP require?
Automotive, aerospace, food and beverage, and industrial manufacturing each carry specific process requirements that generic ERP platforms do not accommodate natively. The gap between what a horizontal platform does out of the box and what a manufacturer actually needs gets filled with customization. That customization drives up implementation cost, extends timelines, and creates a maintenance burden that compounds through every upgrade cycle.
The honest version of this question is: how much of our ERP will we need to build ourselves? The more the answer trends toward “a lot,” the more carefully the total cost of ownership calculation needs to account for what happens after go-live, not just during implementation.
What is a realistic time to value given the timelines in front of you?
A three-year strategic roadmap and a divestiture deadline six months out require completely different approaches. A company with time can pursue a more comprehensive migration. A company under deadline cannot afford to.
This question forces clarity about what the business actually needs the ERP to do and by when. In many cases, a plant-level deployment or a two-tier architecture that separates manufacturing operations from corporate finance can deliver meaningful value on a much shorter timeline than a full enterprise migration. The answer depends on honest assessment of the timeline, not on what any vendor’s standard sales cycle looks like.
How hard is change management in your organization?
Change management complexity tends to be underestimated at the start of ERP programs and fully appreciated somewhere in the middle, when it begins driving schedule and budget.
Organizations with M&A complexity, strong divisional power centers, or operations spread across regions and acquired businesses face a structural challenge when trying to align on a single ERP platform. The coordination required before a single plant can go live can extend timelines by years. A two-tier strategy that lets individual plants modernize without waiting for full corporate alignment often reduces the change management burden significantly, and makes the difference between a program that delivers and one that stalls.
Is your data ready for AI, or does it live in spreadsheets and disconnected systems?
When Tenneco’s braking division needed to separate from a shared legacy ERP instance, the starting state told the real story. Despite SAP being listed as the ERP of record, the division was running MRP on manually updated spreadsheets with unverified formulas. Shipping required six people working through multi-step approvals. Production planning lived in files shared by email.
That picture is more common than most organizations want to admit. And it matters because agentic AI, the kind now embedded in modern ERP platforms, requires trusted, connected data to function. AI cannot act on a process that lives in a spreadsheet. Application rationalization, bringing disconnected workflows into the ERP where they can generate clean, structured data, is a prerequisite for AI readiness. It is not something that can be addressed after the fact.
Where are your competitors on AI, and what does inaction cost per quarter?
AI in manufacturing is no longer a future capability. It is a current competitive variable. Manufacturers with AI embedded in their ERP workflows are surfacing supply chain risks before they become shortages, flagging MRP exceptions before they become production delays, and making operational decisions on real-time data rather than reports that arrive after the window to intervene has closed.
The manufacturers who do not have this capability are not standing still. They are falling behind at a pace that compounds. Every quarter on aging architecture is a quarter during which the gap widens. The cost of inaction is not zero, and it does not stay constant.
Does your implementation partner have proven go-live experience in live manufacturing environments?
ERP platforms do not fail in production because the software does not work. They fail because the implementation was not designed for the realities of a live manufacturing environment: tight sequencing between processes, high transaction volumes, operational decisions that cannot wait, and no tolerance for “we’ll patch it later.”
The difference between a go-live that disrupts production and one that does not comes down to how the implementation was managed before the system was ever switched on. Data migration treated as a checklist item, rather than a risk surface to be deliberately managed, has ended more programs than any software limitation. Partner selection is a technical decision only in part; it is primarily an operational risk decision.
What Honest Answers Tend to Reveal
When manufacturing leaders work through these questions seriously, the picture usually becomes clearer than it looked when the ERP conversation first got put on the agenda. In some cases, the default direction holds up. In others, the answers point toward a different architecture, a different timeline, or a different platform than the one the organization had been drifting toward.
The Tenneco deployment is instructive here. When the braking division answered these questions honestly, a standard S/4HANA migration was evaluated and ruled out. It would have cost approximately double the budget and could not meet the divestiture timeline. QAD Adaptive was deployed across eight plants on three continents in eighteen months. MRP moved fully into the ERP. The integrated WMS went live including a China plant that launched on a newly released warehousing module less than two months after the module’s release. Shipping dropped from six people through multi-step approvals to one person following a two-step process.
That outcome was not the result of choosing the right software. It was the result of asking the right questions before choosing.
Go Deeper on the Decision
The eBook Big ERP Isn’t the Only Option works through each of these questions in full, alongside a detailed treatment of what manufacturing-first ERP architecture actually delivers, the complete Tenneco case study, and a six-criteria evaluation framework for leaders who are in or approaching an active ERP decision.
If your organization is already in an active ERP conversation and you want a direct assessment of whether your current direction fits your business, a Tech Clarity Consult with Arista Consulting will give you that clarity. Book a conversation with our team.
Frequently Asked Questions
What should manufacturers look for when evaluating ERP systems?
The most important factors are micro-vertical fit (how well the platform supports your specific manufacturing type without heavy customization), realistic time to value, change management complexity across your organization, AI and data readiness, and the proven go-live track record of your implementation partner. Platform features matter, but organizational fit and execution discipline determine whether an ERP investment delivers.
What is a two-tier ERP strategy for manufacturers?
A two-tier ERP strategy separates the corporate financial layer from the plant operations layer. Corporate finance runs on a platform optimized for consolidation and reporting, while individual manufacturing sites run on a purpose-built ERP aligned to their specific industry. This approach allows plants to modernize on their own timeline without waiting for full corporate ERP alignment, and significantly reduces change management complexity in multi-site or M&A-heavy organizations.
How do you know if your manufacturing data is ready for AI?
The clearest signal is where your operational processes actually live. If MRP, production planning, shipping approvals, or inventory decisions rely on spreadsheets or files shared by email, your data is not AI-ready. Agentic AI requires trusted, connected data inside a modern ERP to surface recommendations and automate decisions. Application rationalization, bringing those disconnected workflows into the ERP, is the prerequisite step before AI capabilities can deliver real value.
