Business process automation: from manual to smart
How much does it cost you to transcribe, follow up on, and forward messages every day?
Does this sound familiar? two employees spend hours each day transcribing documents, entering orders, or following up on approvals. You know that automation exists, but you’re not sure where to start, and you don’t have an IT project manager to investigate it full-time. However, automating processes is not a multi-year IT project; it’s a step-by-step approach with a working prototype in just a few weeks that quickly shows whether it works
On this page, you’ll learn what automating business processes means for an SME: which processes are suitable, what process automation is, how the process works, and what benefits it offers. We’ll also explain when it’s better to wait. For a broader perspective, see our overview of robotic process automation. Here, we’ll start by asking which of your processes require the most manual labor.
Which processes cost your company the most?
Manual work tends to pile up in areas that you can identify yourself. There are four of them: document processing, approval workflows, reporting, and order and customer communication.
Automating administrative tasks is all about that work: the tasks your employees currently do manually—typing, following up, compiling, and sending documents. If you’re a team leader in operations, finance, or logistics, read through this list with your own team in mind.
1. Document processing. Invoices, orders, quotes, and contracts are processed manually. Consider a wholesale business where orders arrive via email as PDFs and the office staff enters them manually into the ERP system. In our experience, manual data entry leads to errors. Invoices are a prime example of this type of work; to learn how to streamline this process, see “Automating Accounts Payable.”
2. Approval flows. Purchases, leave and quotes are sent via email to multiple individuals. In a manufacturing company, a purchase request goes through the team leader, the buyer, and the financial director. A request that is sent via email to multiple approvers will wait if one of them is absent until they return, and in the email thread, you cannot see where it is stuck.
3. Reporting. A monthly report that is currently compiled manually from multiple systems could be generated automatically. The time is spent on compiling the report, not on reading it.
4. Order and customer communication. Order confirmations, status updates, and invoice follow-ups are sent manually.
You can determine for yourself which of those four areas is causing you trouble.Four signs indicate that a process is a good fit:
– The process is the same every day;
– The input and output documents are clear;
– Errors are costly or visible to the customer;
– The employee finds the work monotonous.
These are guidelines, not requirements, and they do not guarantee that automation will be successful. If you recognize a process in those signs, the question arises as to what exactly process automation does for such a process.
What is process automation?
Process automation means that software performs repetitive actions according to set rules, so that people no longer have to do those tasks manually. That is the definition we use on this page.
What the software does ranges from simple to smart. At the simple end, there is a form that automatically sends an email. At the other end, there is intelligent automation: software that reads a PDF invoice, recognizes the data, and inputs it into your ERP. A human is still needed for anything that deviates and for decisions.
Along the way, you’ll come across some acronyms. RPA, OCR, APIs, and document recognition are tools, not goals. RPA handles the clicking and typing tasks that an employee would normally perform in your existing systems. OCR converts the text in a scan or PDF into text that software can use. An API allows two systems to exchange data directly. Document recognition extracts the relevant data from a document.
And what about the people who are doing that work now? Automation doesn’t replace people; it takes people away from work they don’t enjoy doing and that leads to mistakes.
Once you understand what automation does, the question remains: how does an SME get started without a multi-year IT project?
What does an automation project look like at an SME?
A project to automate business processes runs in four phases and shows early on whether it works. The four phases, in order:
1. AI Discovery Session. A project starts with an AI Discovery Session of half a day to a full day: which processes qualify, how much time they cost now and where the errors are. The outcome is a shortlist on impact and effort. The session runs with the employee who does the process and the process owner.
2. Scope and data. Together we pick one process, scope it and gather the documents and data we build with.
3. Working prototype. Within a few weeks we build a working prototype on a real use case, with your own documents, and the process owner tests it. Then you decide: continue, adjust or stop. That way you know whether the automation works before you invest in the rollout.
4. Rollout and embedding. We make the automation production-ready, integrate it with your ERP, accounting software or CRM, and test it, together with your IT lead. Your team learns to work with it, and monitoring and maintenance are agreed. Embedding rests with your team.
For you as a sponsor, the decision point comes at the end of the third phase. By then you have seen a working prototype on a real use case.
Once the automation is embedded, what does it give you?
What are the benefits of process automation?
The benefit lies in the actions that are eliminated. This section explains the mechanism but does not promise any specific figures.
Time
The time saved comes from the steps that are eliminated: retyping, following up, and merging. How much time is saved depends on the process.
Document processing eliminates the need for retyping. With approvals, there’s no need to follow up: an approval workflow that automatically forwards requests shows where an application is stuck. And with monthly reporting, there’s no need to compile data manually: a monthly report that used to be compiled by hand from multiple systems can now be generated automatically. The time is spent on consolidation, not on reading the report.
Would you like to put a number on it? You can calculate the annual cost yourself: the time spent each time, multiplied by the number of times per year, multiplied by the hourly rate. That amount represents the value of the hours the craft project currently takes—not a cash savings.
Quality
If you don’t retype, you won’t make mistakes: you’ll have fewer typing errors.
Quality also depends on who understands the process. A workflow that is codified in software is less dependent on any one person who understands it.
Scalability
With automation, the volume of work can increase without requiring your team to grow at the same rate, and seasonal peaks require fewer temporary workers for administrative tasks.
There is a downside, however: automation comes with a cost and a margin of error. That brings us to the question of when automation offers little benefit and it’s better to wait.
When It’s Better Not to Automate Just Yet
There are some processes where automation isn’t worth the effort. A process that plays out differently every time, without fixed inputs and outputs, or where errors have minimal consequences, offers little benefit when automated.
Below are the situations in which automating processes offers little business value, or in which it’s better to do something else first:
– The process does not follow a fixed sequence. That is the case above: there are no fixed inputs or outputs, nor are there errors that incur minimal costs.
– No one can describe the process. A process that no one can describe must first be documented before it can be automated.
– The data is the bottleneck. Consider a service company where a controller compiles figures from multiple systems into Excel for the monthly report, and the report isn’t ready until after the meeting it was intended for. In this case, the bottleneck lies partly in the data, not just in the manual work. If the data is the bottleneck and not the manual work, then you start with the data. You can find more information about this on our page about data & AI.
– The scope is small. With a small scope, a small team working in close coordination can be faster than an automation project.
You can check for yourself whether your case qualifies.
First Steps Without Us
Je kunt morgen beginnen, zonder ons, met dezelfde vragen die we in een AI Discovery Session stellen: welk proces het is, hoeveel tijd het nu kost, waar de fouten zitten en of automatisering de juiste aanpak is. Vier stappen brengen je daar zelf:
1. Pick one process. Go back to the first section of this page. Four signals point to a process that qualifies: it runs the same way every day, the input and output documents are clear, errors are costly or visible to the customer, and the person who does it finds it monotonous.
2. Measure it for one week. Note how much time the process takes, how many documents or requests go through it and where it goes wrong. When we work with you, those questions come up in the AI Discovery Session. Here you answer the questions about the process, the time and the errors yourself.
3. Describe the process on a single page. Who does what, with which document, in which system? If nobody can write that page, the process is one you document first, before you automate it.
4. Put the three figures side by side. Time, volume and a process description side by side form the basis of a business case, with or without a partner.
What you have after those four steps is yours, whether you take it to us, to another partner or to your own IT team. Before you take that first step, you may still have a few questions.
Frequently Asked Questions About Process Automation
These are the questions you still have before choosing your first course.
Is my company big enough to automate processes?
Whether automating pays off depends on the process: how often it recurs, how fixed the input is and what an error costs. The signals above help you judge that. A project starts with an AI Discovery Session of half a day to a full day, in which we look together at which processes qualify. At a low volume, a small team with a clear agreement can be faster than an automation project.
Do I need to replace my systems first?
We work with the systems you already have, such as email, ERP and accounting software. The automation sits as a layer on top of your existing systems, and replacing a system is not a condition for starting.
How soon will I see results?
Within a few weeks you have a working prototype on a real use case. If the automation does not meet the business case, you know it by then, and you decide: continue, adjust or stop. We give no guarantee on the result; the prototype is there to find out whether it works.
What if it doesn’t work?
Then you find out early. Within a few weeks you have a working prototype on a real use case; if the automation does not meet the business case, you stop there. The prototype shows whether the automation works before you invest in the rollout. Before any of that, start with the first steps above: pick one process and measure it for one week.