How to Choose ERP Software

Choosing an ERP system is one of the highest-stakes software decisions a business makes. ERP touches finance, operations, supply chain, and often every department in between, so a poor fit is expensive to unwind. This guide explains how to choose ERP software methodically: who the real contenders are and how they actually differ, what implementation honestly costs at your size, and a concrete requirements-to-shortlist-to-pilot process you can run.

If you are new to the category, start with what is ERP for a plain-English primer, then come back here to run the process.

Start with goals and pain points

Before you evaluate a single vendor, get specific about why you are replacing or adopting ERP. Vague ambitions like “modernize our systems” lead to vague requirements and bloated shortlists.

Write down concrete symptoms and the outcomes you want:

• Where are teams re-keying data between disconnected systems?
• Which reports take days to assemble because the numbers live in different places?
• What breaks when order volume, headcount, or locations grow?
• Which manual workarounds and spreadsheets have quietly become mission-critical?

Turn each pain point into a measurable goal, “close the monthly books in five days instead of fifteen” or “give operations real-time inventory across all warehouses.” These goals become the yardstick you measure every option against.

Assemble a cross-functional team

ERP is never just an IT project. Pull in finance, operations, sales, and the people who will live in the system daily. Name an executive sponsor who owns the budget and can break ties. The team that defines requirements is also the team that drives adoption, so involving them early prevents resistance later.

Know the vendor landscape before you shortlist

Most ERP selection advice stays vendor-neutral to the point of uselessness. Here is the honest map. The single most important filter is segment: every one of these products is excellent for the buyer it was built for and painful for the buyer it was not.

NetSuite (Oracle) is the default cloud ERP for growing SMB and mid-market companies, especially multi-entity groups, software and services firms, and e-commerce brands. Its differentiator is breadth in one cloud system: financials, inventory, order management, and consolidation under one data model, with strong multi-subsidiary and multi-currency support via OneWorld. Pricing is a quoted annual subscription (base platform plus users plus modules), not published, and the well-known gotcha is renewal uplift after a discounted first term, so negotiate a cap up front. Companies that outgrow QuickBooks land here more than anywhere else.

Microsoft Dynamics 365 is really two products, and knowing which one you are being sold matters. Business Central serves small and lower mid-market companies; Finance and Supply Chain Management serves the upper mid-market and enterprise. The differentiator is modularity and the Microsoft stack: published per-user per-month pricing, a base-and-attach licensing model where additional apps for the same user are discounted, and native fit with Microsoft 365, Power BI, and the Power Platform for low-code extension. Everything is implemented through partners, so the partner you pick shapes the outcome as much as the product.

SAP S/4HANA is the enterprise heavyweight: deep, industry-grade process coverage for complex, often global manufacturers and enterprises. Cloud deals are typically packaged as RISE with SAP (or GROW with SAP for mid-market public cloud) with users counted as full user equivalents, and implementations run through large system integrators. It is the right answer for genuinely complex enterprise operations and the wrong answer for a 100-person company, where its cost and implementation weight are unjustifiable.

Acumatica competes in the mid-market with one genuinely structural differentiator: consumption-based pricing. You pay for computing resources and transaction volume, not per user, so unlimited users can be in the system. That changes the economics for businesses with many light-touch users, think construction field staff, warehouse workers, or retail locations, who would each cost a seat elsewhere. It sells only through channel partners and has strong editions for construction, distribution, and manufacturing.

Epicor (its flagship now called Kinetic) is a manufacturing specialist. Its depth in shop-floor scheduling, MRP, engineering change management, and mixed-mode manufacturing exceeds what horizontal suites offer out of the box. If you run a make-to-order or engineer-to-order shop, Epicor and its close competitor Infor (industry-specific CloudSuites for manufacturing, distribution, food and beverage) belong on the list ahead of generalist systems.

Odoo is the open-source outlier: a free Community edition and a low-priced per-user Enterprise edition covering an unusually wide app range. Software cost is dramatically lower than the names above, and that is real. The trade-off is that depth in complex consolidation, advanced warehousing, and localization often comes from configuration and custom modules, which you then maintain through annual version upgrades. Best for cost-conscious SMBs with straightforward processes or in-house technical capability.

Sage Intacct deserves a mention for finance-led buyers: it is a best-in-class cloud accounting and financial management system rather than a full operational ERP, and it often wins where the pain is purely in finance, consolidation, and reporting, especially in services, nonprofits, and SaaS.

What implementation honestly costs

The subscription or license fee is the tip of the iceberg. Implementation services usually run one to three times the first-year software cost, and total cost of ownership over three to five years is the only honest comparison basis. Broad, realistic all-in first-year ranges by segment:

• Small business (under roughly 50 users): from the low tens of thousands to around $150,000 all-in, on products like Business Central, Odoo, or an entry NetSuite edition. Fixed-scope, template-based implementations sit at the bottom of that range; add entities, integrations, or migrations from messy data and you climb fast.
• Mid-market (roughly 50 to 500 users): commonly $150,000 to $750,000 for software plus implementation, on NetSuite, Acumatica, Dynamics 365, Epicor, or Infor. Multi-site manufacturing and multi-entity consolidation projects push toward and past the top of the range.
• Enterprise: from around $1 million upward, and large global S/4HANA or Oracle programs routinely run into the tens of millions over multi-year rollouts, with system-integrator services dominating the bill.

Treat these as orientation, not quotes; your integration count, data quality, and customization appetite move the number more than the logo does. Model TCO over three to five years:

• Licensing or subscription, priced per user, per module, by tier, or (Acumatica) by consumption.
• Implementation: often the largest line item, covering configuration, data migration, and project management.
• Customization and integration development, plus the recurring cost of maintaining both through upgrades.
• Training for administrators and end users.
• Ongoing maintenance and support, upgrades, and internal admin time.

Ask vendors for a realistic total including implementation partner fees, and be skeptical of quotes that seem suspiciously light on services; the money reappears later as change orders.

Map required modules to your processes

ERP suites are modular. The art of ERP selection is matching modules to the business processes that actually drive your operation, rather than buying everything on offer.

Common modules include finance and accounting, procurement, inventory and warehouse management, manufacturing (MRP), order management, project accounting, human resources, and CRM. List your core end-to-end processes, order-to-cash, procure-to-pay, plan-to-produce, and note which modules each one depends on.

Separate must-haves from nice-to-haves. Be strict: a bloated must-have list narrows you to expensive, over-featured suites. Most businesses find that a focused set of genuinely non-negotiable capabilities guides the search far better than a wish list of fifty features.

Choose your deployment model

Deployment shapes cost, control, and IT burden. There is no universally right answer, only the right fit for your constraints.

• Cloud (SaaS) is subscription-based and vendor-hosted. Lower upfront cost, automatic updates, and minimal infrastructure, the default choice for most mid-market buyers, and the only option for NetSuite and Acumatica.
• On-premise means you host and control the software yourself. It suits organizations with strict data-residency rules or heavy customization needs, but carries higher upfront and maintenance costs. SAP, Epicor, Infor, and Odoo still offer this path.
• Hybrid keeps some components on-premise and others in the cloud, useful when regulation or legacy systems prevent a full move.
• Two-tier ERP runs a corporate ERP at headquarters while subsidiaries use a lighter, cheaper system that integrates upward, a common pattern where a group runs S/4HANA or Oracle centrally and NetSuite or Business Central in smaller units.

Industry-specific vs horizontal ERP

Horizontal ERP serves many industries with broad, configurable functionality. Industry-specific (or “vertical”) ERP ships with workflows, terminology, and compliance features tailored to a sector, manufacturing, distribution, construction, professional services, healthcare. Epicor in manufacturing, Infor’s CloudSuites, and Acumatica’s construction edition are the live examples above.

A vertical fit often means less customization, faster implementation, and out-of-the-box compliance. But verify the depth is real and not marketing veneer, and check that the vendor has a viable customer base in your niche. A horizontal system with a strong partner ecosystem can sometimes beat a thin vertical product.

Plan for scale and complexity

Buy for where the business is heading, not just where it is today. Pressure-test each option against your growth trajectory:

• Multi-entity support for consolidating multiple legal entities, subsidiaries, or locations. This is a NetSuite OneWorld strength and a common Odoo customization area; know which side of that line your candidate sits on.
• Multi-currency and multi-language if you operate or plan to operate internationally.
• User and transaction volume headroom so performance holds as you grow, and how the pricing model behaves as you add users (per-seat models scale linearly; Acumatica’s consumption model does not).
• Configurability through settings and low-code tools rather than hard-coded customization that becomes costly to maintain through every upgrade.

Integrations with your existing systems

An ERP rarely stands alone. Inventory every system it must exchange data with, e-commerce, CRM, payroll, banking, EDI, warehouse and shipping tools, business intelligence.

For each, decide whether you need a prebuilt connector, a well-documented API, or middleware, and establish who builds and maintains the integration and what it costs annually. Prebuilt integrations to the tools you already run save enormous implementation effort. Broken data handoffs recreate the very silos ERP is meant to eliminate, so treat integration capability as a first-class requirement, not an afterthought.

Evaluate vendor viability

You are entering a long relationship, so the vendor matters as much as the software. Assess financial stability, product roadmap, release cadence, and the health of the partner and implementation ecosystem. A stagnant product or a shaky vendor can strand you years into a deployment. For partner-channel products like Dynamics 365, Acumatica, and Epicor, the implementation partner is often as decisive as the software vendor itself: interview the partner’s proposed team, ask for their references separately, and check how many implementations they have done on your exact product and version.

The selection process: requirements to shortlist to pilot

Here is the sequence in practice, with rough timing for a mid-market buyer. The whole run typically takes two to four months.

Weeks 1 to 3: requirements. Run discovery workshops per department, document current processes and pain points, and produce a prioritized requirements list (must-have vs nice-to-have) plus your integration inventory and TCO model skeleton. Output: a requirements document your executive sponsor signs.

Weeks 3 to 5: long list to shortlist. Use the segment map above to build a long list of five to eight plausible systems, then cut to two or three finalists using published materials, intro calls, and a short capability questionnaire. Kill fast on segment mismatch, missing must-haves, weak presence in your industry, or no credible local implementation partner. For complex or regulated purchases, this is where a formal ERP RFP slots in; for simpler ones, skip straight to demos.

Weeks 5 to 9: scripted demos. Never buy on a canned demo. Give each finalist the same script built around your real processes, your order flow, your month-end close, your reporting, and insist they run it with representative data, ideally a sample of your own. Score every demo with the same weighted sheet, filled in independently by each evaluator before any group discussion. Scripted demos expose where a system strains and let you compare vendors on identical terms.

Weeks 8 to 12: pilot or conference-room test, references, and negotiation. For the leading candidate, push past the demo into hands-on validation: a sandbox or conference-room pilot where your own finance and operations staff execute your top ten workflows themselves, with a slice of your real data migrated in. Vendors resist this; serious ones accommodate it for a serious buyer, sometimes as a paid discovery phase, which is money well spent against a six-figure mistake. In parallel, check references (similar size, same industry, live within two years, at least one from the proposed implementation team) and ask what went worse than expected, how support behaved after the deal closed, and whether they would choose the system again. Then negotiate: multi-year price caps, itemized services scope, named implementation staff, and exit and data-ownership terms, all before signing, while you still have leverage.

Your ERP requirements checklist

Before you commit, confirm you have documented:

• Business goals and prioritized pain points
• Must-have vs nice-to-have modules and features
• Deployment preference (cloud, on-premise, hybrid, two-tier)
• Scalability, multi-entity, and multi-currency needs
• Required integrations with existing systems, with ownership and annual cost per integration
• Compliance and data-security requirements
• A three-to-five-year total cost of ownership model, including renewal terms
• Vendor viability and implementation-partner criteria
• A scripted demo plan, pilot plan, and reference-check questions

Common ERP selection mistakes

• Skipping discovery and shopping for products before defining requirements.
• Shortlisting outside your segment, then either overpaying for enterprise weight or outgrowing a light system in two years.
• Underestimating implementation: treating ERP as a purchase rather than a multi-month program, and budgeting from the subscription price instead of the services line.
• Ignoring change management, so adoption stalls after go-live.
• Buying on price alone and discovering the true cost in services later.
• Over-customizing, which raises cost and complicates every future upgrade.
• Rushing the timeline and cutting testing or data migration short.

For the mechanics of a successful rollout once you have chosen, see our ERP implementation guide.

Get to a shortlist faster

ERP selection rewards discipline, but the research is time-consuming. SoftwareSelect builds free, unbiased ERP shortlists from independent advisors, matched to your requirements rather than to who paid the most. You can browse ERP software to see the landscape, or get free advice to have advisors narrow the field for you.

The principles here apply beyond ERP too, for other categories, see how to choose business software.