You type "business process automation in a manufacturing company" into Google and the results bury you under photos of robotic arms, cobots on assembly lines and clips from automation trade shows. You buy one such machine for a hefty sum, and on Monday morning someone still spends half a day on invoices, someone else calls a supplier to check whether an order even went out, and the warehouse worker counts pieces on a scrap of paper because the system shows last week's stock. The robot sped up one operation on the shop floor, but the company as a whole runs at the same pace, because the bottleneck never sat inside the machine. Before you ask an automation vendor for the price of a robot, it is worth checking ten administrative processes that can be automated more cheaply, faster and with less risk, in one system that your company probably already partly uses.
Why the internet keeps showing you robots instead of a spreadsheet
Robots and cobots sell well in content because they look good in a photo and are easy to write about: one machine, one clear function, one number in the vendor's price list. Automating business processes in the office and at the seam between office and shop floor has no such photo, so it lands on the first page of search results less often. That does not mean it pays off less. It only means it is less spectacular.
A robot or cobot automates one physical operation: welding, palletising, moving a part from one machine to another. The entry threshold is high, integration with the existing line takes time, and the return is limited to that single station. Automating administrative processes covers the whole company at once: sales, warehouse, purchasing, paper production (not physical), finance. It draws on data the company already has, only scattered across a few spreadsheets, mailboxes and individual people's heads.
Where to start automating an SME before you think about the shop floor
A good opening question is not "which robot" but "where in the company does someone retype the same data a second or a third time today". The answer almost always leads to administrative processes: invoicing, ordering, quoting, reporting. These are what eat up the hours of people who could be doing something else, and these are what most often generate errors that cost more than the manual work itself, because they then have to be fixed.
There is also a hard reason to start right now. The obligation to issue invoices in the National e-Invoice System (KSeF) for practically all B2B turnover came into force on 1 April 2026, and the obligation to receive invoices in KSeF applies to all taxpayers already from 1 February 2026. The smallest companies, with gross sales up to 10,000 PLN per month, have until 1 January 2027, but that still means invoicing in your company has already changed or will change in the coming months. Since this process has to be rebuilt anyway, it makes more sense to build it once, well, in one system, than to glue yet another patch onto the old way of working.
10 processes to automate before you buy a robot
The list below deliberately contains no made-up savings percentages. The scale of the benefit depends on how much the mess in your company costs today, and that can only be measured realistically on your own data. Below we describe qualitatively what changes when each of these processes stops living in a separate spreadsheet.
1. Issuing and sending invoices, including KSeF compliance
An invoice created automatically from an order has the complete counterparty data, correct units of measure and the order number right away, before it reaches the National e-Invoice System. Manual retyping from one document into another disappears, and with it disappears the most common cause of rejected invoices and last-minute corrections.
2. Updating stock levels
If goods receipts and issues are recorded in the system the moment they physically happen, the warehouse stops being guesswork. Nobody has to run an ad hoc stocktake before every larger purchasing decision, because the system's stock matches what actually sits on the shelf.
3. Purchase orders to suppliers and material requirements
Instead of a planner who reviews a few tabs every week and calls the warehouse, the system itself calculates what is missing based on the production plan and current stock. Emergency orders, more expensive than standard ones, stop being the norm and become the exception.
4. Quoting offers for customers
A quote based on the product structure and current purchase prices calculates itself, instead of coming from the salesperson's memory or a price list several months old. The margin is visible right away, at the quoting stage, and not only after the job is settled.
5. Turning a customer order into a production order
When the customer accepts the quote, the production order should be created with a single click, with material reservation, not from a fresh email to production and retyping the same items again. This eliminates the typical gap between what the salesperson promised and what production knows.
6. Reporting production progress and working time
Work cards filled in by hand on paper or in a separate spreadsheet reach the system with a delay, sometimes after several days. Recording job progress on an ongoing basis, even on a tablet on the shop floor, gives the production manager a picture of the workload in real time, not from last week.
7. Register of nonconformities and complaints
A complaint reported by email, recorded somewhere other than the job it concerns, is easily lost or repeats without any lessons drawn. A register linked directly to the job and the material batch lets you see whether a given quality problem keeps coming back with the same supplier or on the same operation.
8. Payment reminders and receivables collection
An automatic reminder about an approaching or overdue payment date does not require someone to review a list of invoices in a spreadsheet every week. The company reacts to arrears faster, instead of discovering a liquidity problem only at the end of the month.
9. Purchase order approvals
A purchase order that circulates around the company by email for signature waits for someone who happens to be in a meeting or out of the office. A simple approval workflow in the system, with a clear amount threshold, shortens that time from days to hours, without losing control over spending.
10. Management reports and indicators
If the owner or manager spends an evening every month assembling data from several systems into one report, that is exactly the kind of work a system should do, not a person. A dashboard built once, on live data, replaces the repetitive, manual reassembly of the same table from scratch every month.
Administrative automation versus physical automation: what you are really comparing
| Area | Robot or cobot on the shop floor | Administrative process automation |
|---|---|---|
| Scope of impact | One operation, one workstation | Sales, warehouse, purchasing, paper production, finance at once |
| Starting point | New device and integration with the line from scratch | Data and processes the company already has, only scattered |
| Typical implementation cost | Depends on the machine and integration, usually a high entry threshold | Per symfonia.pl 20,000 to 80,000 PLN for a small company, in a broader scope 80,000 to 300,000 PLN with a full package including licences |
| Cost structure | Mainly hardware and mechanical integration | Per humansoft.pl the licence is usually 10 to 50% of the budget, the rest is implementation and training |
| When it makes sense as a first step | When the administrative process is already in order and the bottleneck is a physical operation | Almost always as a first step, because the cost of keeping the mess grows together with the number of orders |
These ten processes rarely live as separate applications, because they all use the same data: the counterparty, the material, the job, the price. They naturally combine into one ERP-class system. We describe what this looks like specifically on the shop floor on our ERP for manufacturing page. You will find other articles about the next steps of digitising a manufacturing company, including implementation costs and KSeF itself, on our blog.
Micro, small and medium companies in industrial processing or manufacturing services, operating for at least five years, may qualify for co-funding of such a project from the Dig.IT programme of the Industrial Development Agency, in line with the programme rules, up to 50% of eligible costs. The current call status must always be checked directly at digit.arp.pl, because the rules of each successive call change over time. Regardless of the funding, the project order should be the same: first you know which process hurts most, then the system design, and finally a possible grant application. We have been implementing Odoo systems since 2011, for clients in three countries: Poland, Germany and Belgium. Our first client saw a working demo the day after the first conversation, and after two weeks was working on a system whose core still runs today, even though the company was later acquired by a German listed group. Administrative process automation really can get moving that fast.
Frequently asked questions
Does administrative process automation really deliver more than buying a robot?
It depends on where the bottleneck sits today. If the sales, warehouse and invoicing processes are already organised in one system, and the problem is a specific, repetitive physical operation, a robot makes sense. In most companies of 10 to 50 people that are only just starting to think about automation, though, the bottleneck sits in administration: in retyping data, in spreadsheets, in emails, not on the shop floor.
Where to start automating business processes in a small manufacturing company?
By measuring where someone retypes the same data a second or a third time today: between a quote and an order, between an order and an invoice, between the warehouse and production. Those places show which of the ten processes from this article is worth putting in order first. Only then does a conversation about a specific tool make sense.
Do you have to implement a whole ERP to automate these 10 processes?
You do not have to start with everything at once, but these processes use the same data about the customer, material and order, so in practice they are hard to automate durably in scattered, unconnected tools. It is usually cheaper and simpler to maintain one system, to which processes are added one by one, by priority, than a patchwork of several separate applications.
Can you get funding for automating these processes?
Companies in industrial processing and manufacturing services, operating for at least five years, may qualify for the Dig.IT grant, in line with the programme rules, up to 50% of eligible costs. The status of the next call must always be checked at digit.arp.pl, though, because the programme does not run continuously and the rules are sometimes updated between calls.
Before you ask about the price of a robot, check how much the lack of automation of these ten processes costs you today. Our Digital Growth Diagnosis measures 39 KPIs across 7 dimensions, including exactly these processes: sales, warehouse, purchasing, production and finance, and shows which of them eats up the most time and money in your company today.