abas ERP alternative for manufacturers with a fixed-price custom ERP

What abas is, and who it was built for

abas started in Karlsruhe in 1980 and stayed independent for nearly four decades. Forterro acquired it on 2019-06-25, its eighth European ERP acquisition. Partners Group then bought Forterro in 2022 at a valuation around one billion euros. The group now reports 375 million EUR revenue for 2025, more than 25,000 customers and over 21 products. abas is one of them, with roughly 4,000 customers and 120,000 users worldwide.

We state this without judgement, but it changes the question worth asking in a tender. Not whether the product is good today, which it can well be, but who decides in year four what goes into it. Product direction now follows a group-level cloud and AI strategy rather than a decision made in Karlsruhe.

The proprietary database has one advantage that rarely gets mentioned: there is no separate SQL licence underneath it, which Software Advice reviewers list as a cost benefit. The same decision also means customisation requires a language almost nobody outside the abas partner network writes.

What abas costs

Start with the part that matters: no region has a published price list. The four estimates below come from third parties, they partly contradict each other, and that is exactly why all four are worth seeing.

Our own read: the spread is not analyst sloppiness, it follows from price being set inside a sales process. Fifteen users on a basic manufacturing scope land at the bottom of the band. Add the configurator, APS, several sites and cloud hosting and you are at the top. One ratio holds across every source, and it is the useful one to plan around: implementation costs a multiple of the licence.

Five-year TCO for a 30-user manufacturer

A model calculation, not an offer. The abas columns come from the estimates above, the AppForge bands from our public price list. Currency conversion assumes 390 HUF to the euro, which is broadly right for 2026 without being fixed.

How the AppForge five-year column is built: 30-100M HUF project fee plus 15-25 percent annual maintenance over five years. The bottom of the band assumes one site, around 30 users and a basic manufacturing scope; the top assumes multiple plants, MES and BI. Where you land depends on BOM depth and the number of machine interfaces.

The two bands overlap, and that is worth saying out loud. A custom manufacturing ERP is not cheaper because it is less work. It is cheaper because the licence line stops in year six, eight and ten, while an ERP typically keeps running that long.

Production planning: MRP, scheduling, machine data

This section is written for production and operations managers. In an ERP replacement, finance is usually the easy half; the decision tends to fail on planning logic.

What porting MRP logic actually involves

An MRP run does the same four things in every system. It builds gross requirements from sales orders and forecast, subtracts free stock, allocated quantities and open inbound orders to get net requirements, explodes those down the bill of material by low-level code so each part is calculated once, then backward schedules by lead time to say when to order or release.

All of that is off the shelf in a packaged system and two to three weeks of work in a custom one. The real time goes elsewhere: into how your lot-sizing rules, scrap factors, alternate components and supplier minimums are represented, and into what the system is allowed to treat as an allocation. In our experience no two manufacturers answer those the same way, which is also why no ERP demo can tell you in advance how many exceptions you will have.

The interesting divergence starts at scheduling. Classic MRP assumes infinite capacity: it says when an order should start, without checking whether the machine is free. Finite-capacity scheduling is therefore a separate layer, called APS at abas. In a custom system we solve it with a constraint solver, feeding in operations, machines, tooling, shift calendars and sequence-dependent changeover times, and you choose what the model minimises. Minimising lateness produces a different plan from minimising changeover time or work-in-progress value. In a packaged APS that objective is usually parameterised but not replaceable.

Confirmation is the other place ERP projects bleed out. If an operator types the operation close into a terminal, the data is exactly as reliable as the supervisor's nagging. If cycle time and piece count arrive from the machine over OPC UA, the load calculation corrects itself. So we suggest manufacturers never start an ERP replacement without some machine integration, even if the first round covers only three or four critical work centres.

Fairness runs both ways here. If your product is described by a configurator with hundreds of rules and forbidden combinations, the abas variant engine is a real advantage, and rebuilding it from scratch costs months. In that case our own advice is to keep the packaged configurator and automate the processes around it instead.

Leaving abas: a realistic timeline

This plan assumes a single-site manufacturer with 30-80 users. Migration work sits inside the 30-100M HUF project fee rather than on top of it, but the quality of your own data is the largest unknown: if bills of material have not been maintained for years, weeks 4-6 will slip.

1. Week 1: object model inventory

   Read out over ODBC and SQL which object types, fields and in-house extensions are actually live in the abas installation. This is where you find out how many FO/FOP customisations exist, and which of them still sit behind a real business process.

2. Weeks 2-3: master data export

   Item master, business partners, chart of accounts, warehouses, tax codes, work centres. Item master inheritance and configurator rules hide here and have to be unpacked separately, because most exports flatten them.

3. Weeks 4-6: bills of material and routings

   BOM versions, validity dates, alternate components, operation sequences and standard times. This takes half the migration effort, and it is usually where you discover that the shop floor has quietly diverged from the recorded routing.

4. Weeks 7-9: document history

   Three to five years of sales and purchase orders, delivery notes, production orders, confirmations, stock movements and journal entries, loaded into PostgreSQL so historical reporting survives go-live.

5. Weeks 10-20: build, UAT, parallel run

   The new ERP comes together module by module, production planning last. At least four weeks of parallel operation with daily reconciliation before cutover. abas stays readable for 60 days, and only then do you cancel the licence.

Decide one thing up front: how much history you actually need live. Most companies say all of it, then in practice open nothing older than two years except for audits. Saying that out loud at kickoff frees up weeks.