Manufacturing ERP for executives

A manufacturer with fewer than ten plants, one or two countries, and revenue under about $500M will usually get a better result from Epicor Kinetic, Infor CloudSuite Industrial, IFS Cloud, QAD Adaptive ERP, or Plex than from SAP S/4HANA or Oracle Fusion Cloud ERP. The published program cost for a tier 2 system in that band runs $300,000 to $3M and 6 to 18 months. A single-region tier 1 program runs $500,000 to $5M or more and, in programs we have advised, 36 months from decision to go-live (sysgenpro). The exceptions are real and specific: many legal entities across many countries, a parent that already runs tier 1, or a plan to pass $1B to $2B inside the life of the system. If none of those apply, tier 1 is overkill.

Manufacturing mode cuts the list in half before revenue enters the picture

How you make things matters more to the shortlist than how much you sell. A discrete assembler, a batch chemical plant, a job shop, and a mixed-mode plant need different planning logic, costing, and shop floor screens. Every vendor claims all four; the product pages and the independent lists agree on far fewer.

For discrete make-to-stock and make-to-order work, all eight systems compete. Epicor says Kinetic covers "make-to-order, configure-to-order, engineer-to-order, and make-to-stock" plus mixed-mode (Epicor). Infor names discrete and process manufacturing for CloudSuite Industrial, with engineer-to-order and configure-to-order called out for machinery and equipment (Infor). IFS lists discrete, batch, repetitive, configure-to-order, engineer-to-order, assemble-to-order, make-to-order, and make-to-stock (IFS fact sheet).

For process and batch work the list is shorter. The manufacturingleadgeneration.com list of 50 systems puts only Plex, Sage X3, Infor CloudSuite Industrial, and the dedicated process products (BatchMaster, Deacom, Aptean Food and Beverage) in its process category (manufacturingleadgeneration.com). Argano adds QAD for regulated process and Oracle Fusion and S/4HANA at enterprise scale (Argano).

Epicor Kinetic is the one system in this comparison that no source we opened names for process manufacturing, and IFS's own blog says Epicor is weaker there (IFS blog, vendor). A food, chemical, or pharmaceutical manufacturer should start with Infor, QAD, Plex, or a tier 1 system, and treat Epicor and IFS as candidates only if a reference plant in the same mode can be visited.

Job shops are a different market. None of the eight systems here appears in the job shop category of the list of 50, which is held by ProShop, JobBOSS2, Global Shop Solutions, E2, and similar products. A job shop under about $50M reading this article is likely looking one tier too high. The exception is a job shop inside a larger group that needs one ledger, where Epicor Kinetic's Job Management module is the usual compromise.

Mixed-mode plants, where one site runs repetitive lines next to engineer-to-order cells, are where the tier 2 vendors do best. Argano names Epicor Kinetic as its mixed-mode pick and the list of 50 labels only Epicor for mixed-mode. IFS names mixed-mode in its own comparison page (IFS glossary, vendor). S/4HANA and Oracle Fusion handle mixed-mode too, but through configuration of several production types rather than a single plant model, and that configuration is a large part of why their programs run longer.

Plant count and country count set the ceiling, not revenue alone

Revenue is a poor proxy for ERP complexity in manufacturing. A $400M company with two plants in one country is a simpler program than a $150M company with six plants in four countries. What pushes a company toward tier 1 is the number of legal entities, the number of countries with their own statutory reporting, and the number of plants that must share one planning run.

The published ranges draw the lines in roughly the same place. IFS puts the point where SAP S/4HANA enters the evaluation at about $500M in revenue (IFS blog, vendor). ERP Research gives IFS, QAD, and Plex employee bands that top out at 5,000 or more (ERP Research, IFS vs QAD; ERP Research, IFS vs Plex). At manufacturing revenue per employee of $300,000 to $500,000 (a planning assumption, not a sourced figure), that puts companies of $1.5B to $2.5B inside the tier 2 range.

QAD claims localized tax and statutory packs for more than 100 countries and 25 or more languages, a direct answer to the multi-country objection (ERP Research, QAD). Infor sells a separate CloudSuite Industrial Enterprise edition for "complex multi-site, multi-company, and multi-country operations" (Infor).

The rule we recommend for a first pass has three bands, and the thresholds are our judgment rather than a published standard. Under $500M and under ten plants in one to three countries, tier 2 is the default and tier 1 needs a stated reason. The $500M line is where IFS says SAP evaluations typically begin, and ten plants is roughly where a single planning run across sites starts to strain a mid-market product.

Between $500M and $2B, or ten to twenty-five plants, or four to ten countries, both tiers belong on the shortlist and the decision turns on the mode and MES questions below; the $2B ceiling follows from the 5,000-employee bands above. Above $2B, more than twenty-five plants, or more than ten countries with statutory reporting, tier 1 is the default and a tier 2 vendor needs to prove it with a reference customer of the same shape.

The cost of getting the band wrong runs in both directions. A tier 1 system in a $300M two-plant company carries the 36-month program and the tier 1 subscription for the life of the system without using most of what it bought. A tier 2 system in a $3B twenty-country group tends to run out of road at the intercompany, consolidation, and localization layer, and the fix, in our judgment, is usually a second program five to seven years later. The QAD limits that ERP Research lists (basic CRM and HR, reporting that needs custom work, a smaller partner network) are the shape that running out of road takes (ERP Research, QAD).

Where each of the eight systems fits best, and where it runs out of road

The fit column is where a system is the natural first choice. The limit column is the condition under which it stops being one. Sources are the vendor pages and independent comparisons opened for this draft, with vendor-sourced claims marked.

Your MES situation belongs in the vendor decision, not after it

Shop floor and MES integration is the single largest cost driver in a mid-market manufacturing ERP program, ahead of quality and traceability and ahead of multi-plant rollout, according to ERP Research's cost breakdown (ERP Research). Most selection teams still treat MES as a later question. That ordering is backwards, because the answer changes which vendors belong on the list.

There are three starting positions. The first is a company with no MES, running paper travelers and spreadsheets on the floor. Here a system with MES built in saves a second program.

Plex was built this way; its MES is one module of the same platform, and Rockwell positions it to run either standalone beside a corporate ERP or with Plex's own ERP (ERP Research, Plex MES). Epicor sells Advanced MES alongside Kinetic for schedule changes and plant floor data collection (Epicor). Infor sells Infor MES as a separate add-on to CloudSuite Industrial (Infor).

The second position is a company with a working MES it intends to keep, such as a Rockwell, Siemens, or AVEVA product on the floor. Here the ERP question narrows to finance, planning, and the interface, and tier 1 loses its main disadvantage. S/4HANA's MES connectivity where SAP is already the core system is rated strong by sysgenpro, and Oracle Fusion's hybrid integration is described as "common for MES" (sysgenpro). A company in this position with 15 plants and six countries has a fair case for tier 1 at the top and its existing MES below.

The third position is the hardest: several plants, each with its own MES or none, acquired over time. Sysgenpro's observation that buyers "often overestimate native ERP capabilities" at the plant level applies here more than anywhere (sysgenpro). The practical choice is between standardizing on one MES-included ERP across the plants (Plex or Epicor in discrete, Infor or QAD in regulated process) or picking a corporate ERP and a separate MES standard, which is two programs and two SIs. In our judgment the second route is right above about 20 plants; below that, it is usually more program than the company can carry. The boundary between the two systems and how to test it in a demo is the subject of E22 and S14.

For a $500M five-plant manufacturer, tier 1 costs about twice as much and takes twice as long

Take a $500M discrete manufacturer with five plants in two countries, about 400 named ERP users, and no MES standard. This company sits exactly on the line, and both sets of vendors will pursue it hardest. The table applies the published ratios to both tiers; where a vendor does not publish a price, the row says so.

The ratios are the same in both columns. What differs is the base they multiply. Note that sysgenpro's published Oracle Fusion range, $300,000 to $3M, overlaps the tier 2 range; the tier 1 premium shows up in the implementation and the months of PMO, not in that published band. On the published figures, our planning number for a tier 1 implementation at this company is $2M to $5M, and more once a parent insists on a global template. For tier 2 it is $1M to $3M.

The timeline gap is wider than the cost gap. Thirty-six months against 12 to 18 means two extra years of PMO, two extra years of the old system's maintenance, and two extra budget cycles in which the design has to catch up with the business. For a finance committee, the timeline is the difference it will feel.

Two cautions apply. First, holding custom code under 10 to 15% of requirements saves 20 to 40% of implementation cost in either tier. A tier 2 system in a vertical it was built for needs less custom code than a tier 1 system configured from scratch (ERP Research cost breakdown). Second, a survey reported by ERP Today found 31% of implementations took longer than expected and 12% had significant delays, and that risk is not tier-specific (ERP Today). A late tier 2 program is still cheaper than an on-time tier 1 program for this company, but by less than the headline suggests.

The plant-level view makes the same point from the bottom up. Take a $400M manufacturer with three discrete plants and one process plant, 320 named users, and two legal entities in one country. On a tier 2 subscription of $150 per user per month, the software line is $576,000 a year. At the 1x to 3x ratio, implementation is $576,000 to $1.7M, of which the SI takes 40 to 60%, or $230,000 to $1.04M.

Support at 15 to 25% of implementation adds $86,000 to $432,000 a year after go-live. Spread over four plants, the implementation works out to $144,000 to $432,000 a plant. A phased rollout of one plant a quarter after the pilot puts the last plant live inside 15 to 18 months.

Run the same plant on a tier 1 subscription at a floor of $180 per user per month and the software line is $691,000 a year. That is only 20% more, which is why subscription alone never settles the argument. The difference lands in the implementation. The published single-region tier 1 range of $500,000 to $5M or more starts above the tier 2 floor of $300,000, and its ceiling is nearly twice the tier 2 ceiling of $3M (sysgenpro).

Per plant, that is $125,000 to $1.25M against $144,000 to $432,000. The process plant usually sits at the top of its range in either tier, because batch costing and recipe management are configured, not switched on. Over 36 months the PMO line alone runs $180,000 to $720,000, as much as a whole plant's share of a tier 2 implementation.

Four situations where tier 1 is the right call for a manufacturer

The argument so far is not a case against SAP and Oracle. There are four situations where a manufacturer under $2B should still shortlist tier 1, and a CFO who skips it in any of them is making the opposite mistake.

The first is a parent that already runs tier 1. A $400M subsidiary of a $6B group on S/4HANA will spend more on the interface, the consolidation, and the second set of auditors than it saves on a tier 2 subscription. The two-tier pattern, a tier 1 corporate ledger with tier 2 systems in the plants, is covered in E9. The second is a planned path through $1B to $2B inside the next seven years by acquisition, with each deal adding countries and legal entities. A company that knows it will hit the tier 2 ceiling should not plan to pay for two programs.

The third is a company where finance and consolidation, not production, carry the business case. IFS's own comparison concedes that SAP S/4HANA is strongest in global financial governance and multi-entity reporting and that Oracle Fusion is strong in financial management and consolidation (IFS blog, vendor). A contract manufacturer with thin margins, 15 entities, and a public listing may care more about a three-day close than a scheduling board. The fourth is a company with a working MES standard across its plants that it intends to keep; with the floor already solved, tier 1's finance and planning strengths matter more.

Outside those four, the burden of proof sits with tier 1. Ask the SI proposing S/4HANA or Oracle Fusion to name which of the four applies. If the answer is "scalability" or "future-proofing" without a number attached, the answer is none of them.

The wrong tier shows up around year three, not at go-live

A tier mismatch rarely shows in the selection or even at go-live. Our assumption is that it shows about three years in. The program team disbands in the first weeks after go-live, the first full budget cycle on the new system runs in year two, and by year three the business has added a plant, a country, or a product line the design did not anticipate.

A tier 1 system in a company too small for it shows up as a change request backlog nobody can afford to clear, a release cycle the IT team cannot keep up with, and a subscription line that grows faster than revenue. A tier 2 system in a company too large for it shows up as spreadsheets around the consolidation, a localization gap in the newest country, and a plant that runs its own system because the corporate one "does not do what we do."

Consider the overkill case first. A $350M discrete manufacturer with two plants, one legal entity, and 250 users picks S/4HANA because the board wants a name it knows. The published single-region range is $500,000 to $5M or more, and a program of this shape with a global SI template tends to land near the middle, call it $2.5M. The tier 2 range is $300,000 to $3M, and a two-plant program sits near the bottom, call it $800,000.

The program takes 36 months instead of 12 to 15, so the company pays two extra years of old-system maintenance and $120,000 to $480,000 of extra PMO. In year three the plant asks for a scheduling board and a shop floor data screen, and the answer is a separate MES selection, because S/4HANA's plant depth needs extra modules (IFS blog, vendor). The extra cost sits in three P&L lines: a subscription about 20% above a $150 per user tier 2 rate, an SI bill roughly three times larger, and an internal team that stays at program size for two years longer than planned.

Now consider the opposite. A $600M process manufacturer with six plants in two countries goes live on a tier 2 system in 14 months for $1.8M, a good outcome. Over the next four years it buys three companies and ends up with fourteen plants, nine legal entities, and six countries, two of them with statutory reporting the vendor has never localized.

Consolidation breaks first, in year three, when the ninth entity arrives and the close stretches from five days to twelve. The localization gap follows in year four, and the newest plant keeps its old system. The second program, now tier 1, starts in year five at $5M to $10M, at or above the top of the published single-region range. That is roughly three to six times the first program, and it runs another 36 months. The first program was not a mistake on the day it was signed. The mistake was signing it without writing down the entity and country count at which it would stop fitting.

Both failure modes come from deciding on revenue alone. Ask the mode, plant, country, and MES questions in that order, write the answers down before the first demo, and you will usually land in the right tier and know when to revisit it.