ERP for manufacturing: what the production layer covers

ERP for manufacturing is a single layer that adds data about what a product is made of and how it is made to ordinary accounting. On top of accounting, it holds bills of materials, material consumption rates, routings with operations and work centers, material requirements planning, shift assignments for the shop floor, and actual cost calculation for each specific order. An accounting program answers the question “how much did we buy and how much did we ship.” The production layer answers “what is in progress right now, what is it built from, how much went into it, and when will it be done.” Demand for this layer in Russia is holding steady: according to the Korus Consulting Group study “ERP today: priorities and barriers,” inventory and production management is the second implementation priority at 17.5%, right after finance and budgeting (22.7%). Almost half of the sample, 48.72%, came from the manufacturing sector (ComNews, February 24, 2026, by Alexey Mikolenko). The actual depth of shop-floor automation is a different story: only 10% of machine tool industry enterprises use real-time core production management solutions.

Below: what exactly the production layer adds on top of accounting, and what data it cannot launch without. Where the national standard draws the line between ERP, MES and APS, and why it is worth knowing before talking to a vendor. How make-to-order production differs from series production from the system’s point of view. Why a plant project breaks down not on software but on the item master and outdated standards. Which signs show that an accounting system is no longer enough. What to gather before the start: documented production processes, standards, scrap and rework tracking, and a master data owner after launch.

How does ERP for manufacturing differ from an accounting system?

In short: the production layer stores what a product is made of and how it is made, and that lets it plan rather than just record what has been done.

An accounting system knows receipts, issues, balances and documents. It does not know that assembling one pump takes a housing, an impeller, a shaft and two bearings, that the housing goes through four operations at three work centers, or that setup and teardown on the first operation takes forty minutes. Until this data exists, any plan is built in the head of the production manager and in a spreadsheet next to the accounting database.

That is where the practical difference comes from. The production layer lets you break a customer order down into material requirements and shop-floor load, see shortages in advance, issue a shift assignment, and collect what the order actually cost. Everything else, from reports and analytics to integrations, is built on top of these four capabilities.

This is exactly why manufacturing projects start differently from distribution ones. We covered the general stages, timelines and reasons for failure in our article on ERP implementation; at a plant, they come with an extra layer of work on production standards that simply does not exist in distribution.

What does the ERP production layer include?

In short: six blocks of data, each of which is useless without the one before it.

This table should be read top to bottom: each row depends on the one above it. Material requirements planning does not work without bills of materials and consumption rates, and actual order cost cannot be collected without recording output and scrap. Module vendors demo the bottom rows, attractive load charts and cost broken down by order, but those only work once the top rows are filled in.

One line that is missing from the table and gets expensive in projects is work-in-progress tracking. While a semi-finished product moves between shops, it sits somewhere between the warehouse and finished goods, and in most accounting systems this state is described only formally. We looked at similar mechanics on the warehouse side in our article on WMS systems: until a movement is recorded at the moment it happens, the balance in the system and the balance on the shop floor drift apart.

How does ERP differ from MES and APS?

In short: the boundary is set by a national standard, and it follows the planning horizon.

GOST R IEC 62264-1-2014, Russia’s adoption of the international IEC 62264-1 standard, “Enterprise-control system integration. Part 1. Models and terminology” (introduced by Order No. 1874-st of Rosstandart, Russia’s standards agency, on November 26, 2014), describes a hierarchy of levels. Level 4 is business planning and logistics: the master production schedule and inventory levels, on a horizon of “months, weeks, days.” Level 3 is manufacturing operations management: workflow, dispatching and optimization, on a horizon of “days, shifts, hours, minutes, seconds.” Below are levels 2, 1 and 0: monitoring and process control all the way down to physical equipment.

In practice, the mapping works like this. ERP lives at level 4 and answers the question “what are we launching this month, and when.” MES lives at level 3 and answers “what is this machine doing this shift, and what is happening to it right now.” APS solves a separate task within planning: it builds a detailed schedule that accounts for constraints such as available capacity, changeovers, operation sequencing and shipping deadlines. Small plants combine these functions in one system; plants with several production stages separate them and connect them through data exchange.

The shop-floor-level market in Russia has already taken shape as a market of its own. According to a study by the TAdviser analytical center and the Skolkovo Foundation, published on July 20, 2026, the Russian MES market will exceed 15 billion rubles in 2026, growing at no less than 8%. Over 60 products were reviewed, 19 domestic vendors made the list of leaders, and the assessment used 270 criteria: 131 for functionality, 99 for technology and 40 for organizational factors. The study concludes: “Emergency import substitution projects are largely complete. Customers now primarily evaluate the potential business impact.”

Demand, meanwhile, lags noticeably behind supply. The APKIT Council for Digital Technologies in Industry, with support from the Ministry of Industry and Trade, surveyed 39 machine tool industry enterprises from November 2025 to February 2026; the data was verified by the Machine Tool Building Industrial Competence Center. According to results published on May 4, 2026, 10% of enterprises use real-time core production management solutions, 56% use elements of PLM occasionally, and 82% have no written IT strategy. 41% of plants rely mainly on manual labor with minimal automation, another 59% run CNC machines with manual loading, and none reported comprehensive automation with robots and industrial process control systems. Pavel Rastopshin, Chairman of the APKIT Council, describes the overall picture this way: automation “is mostly local in nature; individual areas are automated, but not the whole chain.”

Make-to-order or series production: where is the line?

In short: in make-to-order production the unit of accounting is the order, in series production it is the batch and the rhythm, and systems are configured differently for each.

Make-to-order production means the bill of materials is often created together with the order, and changes arrive mid-production. That leads to specific requirements: links to engineering data, versioned bills of materials, tracking of engineering change notices, and cost collection for a specific item down to individual operations. Planned cost here is an estimate, not a standard, and the gap against actuals is analyzed separately for every major order.

Series production works differently. The product composition is stable, and what matters shifts to rhythm and changeovers: how to distribute batches across equipment so as not to lose a shift to retooling, and where actuals diverged from standard. Standard costing with variance analysis works here, and automatic scheduling delivers the gains.

What this means in numbers is shown by Novabev Group’s presentation at CNews Forum Cases 2026. Vladislav Bazhenov, the company’s production director, said that after implementing automatic planning algorithms, the time to build a plan dropped from two full days to 20 minutes, and planning accuracy rose to 98% (CNews, July 6, 2026). Separately, the company analyzes the structure of downtime by equipment type and cause: technical stoppages, scrap, changeovers.

The hardest case is series production with customer-specific variants. The system has to maintain versioned bills of materials, as in make-to-order mode, and plan by rhythm, as in series mode, at the same time. Comparing vendors on a generic feature list is pointless in this situation: you need to test the specific scenario on your own data.

Why do projects break down on the item master and production standards?

In short: production calculations rely directly on master data, and dirty data stops them before any technical limitation does.

Duplicates in the item master. A bill of materials references items, a consumption rate references units of measure, a routing references work centers. One item under four records produces four balances and four purchasing requirements. In a CNews article of February 12, 2026, IBS gives a typical mix from a single database: “pump,” “impeller pump,” “NP-80,” “centrifugal pump 80 cubic m.” The article estimates the effect of cleaning this up at 15–25% savings on purchasing and up to a 20% reduction in inventory. This is an estimate from an MDM platform vendor, but the mechanics behind it are clear: aggregated demand instead of four independent requests.

Outdated standards. A standard written ten years ago for a different machine and a different material is worse than none: it looks like data and takes part in calculations. A practical sign is regular manual adjustment of write-offs “to match actuals.” As long as these exist, neither planned cost nor purchase planning makes sense, because both are calculated from numbers that even production does not trust.

Undocumented shop-floor exceptions. Recoverable waste, substituting a material with an equivalent, rework after quality control, splitting a batch between shifts, reprocessing after scrap. Inside the plant these are not considered processes (“the foreman will sort it out”), and they are the last to make it into the assessment. Then they surface during pilot operation and require changes that are not in the budget.

Data without an owner. Evgeny Antropov, head of the master data department at IBS, puts the condition plainly: data quality management must become a full-fledged business process. At the same CNews conference, independent expert Igor Khlebnikov estimated that about 95% of accumulated data remains “dark,” with no owners and no business context. We covered how to extract data from primary documents without manual entry in our article on recognizing primary documents in 1C, the accounting and ERP platform most Russian companies run on.

The same conclusion comes from another direction. K2Tech surveyed 150 respondents and conducted 15 in-depth interviews in mechanical engineering. According to a publication of July 29, 2026, 60% of companies rated their level as medium or basic, 19% as high and 21% as low. The study names a shortage of quality data and an immature operational layer as the main barriers to AI projects. 37% named ERP and information security as drivers of efficiency (20% and 17% respectively), and 11% named shop-floor automation.

When is an accounting system no longer enough?

In short: when the plan and actuals live in different files, and the cost of an order shows up after shipment.

The signs form a set that is easy to check against your own situation:

• the shop-floor plan is kept in spreadsheets separate from the accounting database, and one person reconciles them;
• material shortages for an order are discovered on the day the batch is launched, not when the order is accepted;
• nobody can put a number on work in progress as of today;
• order cost is calculated a month after shipment and is not broken down by operation;
• write-offs by standard are regularly adjusted by hand “to match actuals”;
• lead times quoted to customers include a buffer, because actual shop-floor load is not visible.

The overall scale of the problem shows in the Korus Consulting survey: 43% of companies have implemented ERP only partially or believe the system needs optimization (23.08% implemented it partially, 20.51% say it needs optimization), 10.3% use it as their core management system, and 36.9% are preparing to implement it. Georgy Shmonov, deputy director of Infostandart, describes the typical state like this: “Many large companies have what is called island automation, where individual areas are automated but there is no single system.”

The second trigger for a project is set by the vendor’s calendar. In the same sample, 24.1% of companies still run 1C:UPP, the manufacturing configuration of 1C, and the timeline was announced in advance. According to 1C information letter No. 30064 of December 9, 2022, the configuration is supported until the end of 2026, and in Q1 2027 updates are planned only for filing 2026 reports. Legislative changes from January 2027 onward will not be supported, and consultations on the configuration end on April 1, 2027. The deadline is known, but it does not define the scope of work: migrating “as is” will reproduce all the accumulated workarounds in the new system. We covered separately the point at which spreadsheets stop coping and a single system becomes a concrete topic: when a business should move off Excel.

How to choose ERP for manufacturing and warehousing?

ERP for manufacturing and warehousing is chosen based on real data and operating mode: master data, bills of materials, standards, routings, stock balances, reservations and costing. Then integrations, load, security, total cost of ownership and implementation methodology are tested on make-to-order or series production scenarios.

Production data

Check bill of materials versions, consumption rates, routings, work centers, scrap and rework. Without a master data owner, the system will not be able to plan and calculate costs reliably.

Warehouse and production mode

The system must link requirements, purchasing, stock balances and reservations to the production order. Check separately whether you need make-to-order, series or mixed mode.

Integrations and ownership

Compare data exchange with the accounting system, CRM, WMS and electronic document interchange (EDI), legally binding electronic document exchange between companies in Russia, as well as load, security and TCO. Run the demo on your own scenario, not on a generic feature list.

What needs to be prepared before the project starts?

In short: seven things, and most of them are gathered by process engineers and production, not by IT.

1. Documented production processes and routings. Operations, work centers, setup and run time, tooling requirements. Not “how it should be according to the instructions,” but how the product actually moves through the shop, including areas where only one foreman knows the sequence.
2. Current consumption rates. Materials, recoverable waste, acceptable loss percentage. Recalculating standards for current equipment and current materials is a separate job for the process engineer, and it is planned before the start, not during migration week.
3. Versioned bills of materials. Product composition, assembly nesting, the procedure for making changes, and a flag showing which version a specific order was made to. For make-to-order production, this is a precondition without which cost cannot be collected.
4. Tracking of scrap, rework and material substitutions. Who records scrap, at what stage, what happens to a rejected part, how a substitution with an equivalent is documented. These rules become acceptance criteria for the phase, not something agreed along the way.
5. Access to your own data. Exports of stock balances, work in progress, orders and settlements in machine-readable form. If one person compiles this data by hand, migration turns into a separate project within the project.
6. A list of integrations. The accounting system, engineering data and PLM, CNC equipment and monitoring systems, weighing and labeling equipment, the bank, document management. Each connection means separate money, separate risk and a separate owner. How this integration layer works is covered in our article on system integration.
7. A master data owner after launch. A person responsible for maintaining the item master, bills of materials and standards, with the authority to refuse a duplicate entry. Without them, master data drifts apart within the first few months, and the layer reverts to its original state.

The first four items are about production, not IT, and they are the ones most often skipped. Which numbers the finished layer must deliver to management after launch is covered in our article on management reporting; without an answer to that question, there is nothing to measure the project’s impact against.

When does an off-the-shelf ERP not cover manufacturing?

In short: when the volume of customization for your production stages, exceptions and accounting rules starts to exceed the volume of out-of-the-box logic.

An off-the-shelf product sells a ready-made operating model. As long as your procedures, master data and access rights do not diverge from the standard ones, it is a good deal: you get proven logic ready to use and do not have to build it. Divergences begin with exceptions, and manufacturing has noticeably more of them than distribution: its own way of documenting production stages, its own material substitution scheme, its own batch-splitting rules, its own cost calculation by stage, its own roles in quality acceptance. Configuration covers the first few divergences; beyond that, changes are layered onto someone else’s architecture, and every vendor update has to be retested. Igor Khmel, head of ERP at UltimaTek Group, recalls the original purpose of this class of systems: ERP made it possible to identify the costs of each production stage and manage cost more precisely. If the layer does not deliver that, the question of whether it fits remains open regardless of the platform’s rating.

The workable answer is more often a hybrid. Statutory accounting lives on the platform, while the area that determines lead times and money is built separately and connected through data exchange. That is what we did in our wholesale case: order intake, stock reservation and turning shortages into purchase and production requests moved to a separate layer with two-way integration with 1C, while the accounting system stayed in place. Orders began to be processed in minutes instead of hours. We covered the choice between a ready-made solution and a system built around your process separately: custom system or off-the-shelf. How the types of systems themselves are structured and how they differ in cost of ownership is covered in our overview of ERP system types.

We work under contract and keep data in Russia: we build ERP and CRM layers, integrations and AI agents around a company’s specific process, and after handover we take the system on for ongoing support. A conversation about the production layer starts not with choosing a platform but with your actual situation: what shape your bills of materials and standards are in, how scrap and rework are documented, and who will own the master data after launch. If that conversation is relevant for you now, tell us about your task: we will review your processes and estimate the scope of work area by area.

Sources and links

• ComNews, February 24, 2026. “Almost half of companies ‘don’t get along’ with ERP. Finance and production need a reset” (Alexey Mikolenko) — Korus Consulting Group study “ERP today: priorities and barriers”: 43% with partial implementation or implementation needing optimization, implementation priorities (finance 22.7%, inventory and production 17.5%), 48.72% of respondents from the manufacturing sector, 24.1% on 1C:UPP, comments by Georgy Shmonov (Infostandart) and Igor Khmel (UltimaTek)
• APKIT, May 4, 2026. Study of IT in the machine tool industry — survey of 39 enterprises from November 2025 to February 2026 with support from the Ministry of Industry and Trade, verified by the Machine Tool Building Industrial Competence Center: 46% initial and 54% medium maturity level, MES at 10%, occasional PLM use at 56%, 82% without a written IT strategy, 41% mainly manual labor, comment by Pavel Rastopshin
• Skolkovo Foundation, July 20, 2026. TAdviser analytical center study of the MES market — over 15 billion rubles in 2026 with growth of at least 8%, more than 60 products, 19 leading vendors, 270 assessment criteria
• GOST R IEC 62264-1-2014 “Enterprise-control system integration. Part 1. Models and terminology” (introduced by Rosstandart Order No. 1874-st of November 26, 2014) — hierarchy of levels: level 4 (months, weeks, days) and level 3 (days, shifts, hours, minutes, seconds)
• CNews, July 29, 2026. K2Tech study of mechanical engineering — 150 respondents and 15 in-depth interviews: 37% on ERP and information security, 11% on shop-floor automation, 60% at a medium level, shortage of quality data as the main barrier
• CNews, February 12, 2026. “Saving on purchasing and reducing inventory: how to manage master data effectively” — example of item master duplicates, IBS estimates of 15–25% savings on purchasing and up to 20% inventory reduction, comments by Denis Vasiliev and Evgeny Antropov
• CNews, July 6, 2026. “From ‘dark data’ to autonomous plants: the main industrial cases at CNews Forum Cases 2026” — Novabev Group case (Vladislav Bazhenov): a plan in 20 minutes instead of two days, 98% planning accuracy; Igor Khlebnikov’s estimate of the share of “dark data”
• 1C, information letter No. 30064 of December 9, 2022 — 1C:UPP support until the end of 2026, Q1 2027 updates only for 2026 reporting, consultations end on April 1, 2027