Implementing Digital Transformation in Your Company, Step by Step
Table of Contents
Most guides to digital transformation in companies aren’t step-by-step instructions. They’re maturity models, strategy slides, and ten-point checklists that tell you where you stand, but not what to do on Monday. “Develop a digital transformation strategy” isn’t a step. It’s a headline.
In this article, “digitalization in business” refers to something specific: transferring a core operational process—such as quoting, order processing, scheduling, or billing—from emails, Excel spreadsheets, and verbal instructions into a system that works exactly like your business. Not the entire company all at once. One process at a time, then the next.
This article outlines the sequence in which a medium-sized company with between 75 and 200 employees actually implements this: six steps, each culminating in a clear decision. Let’s state our position up front, since it shapes the entire sequence: Digital transformation in medium-sized businesses rarely fails because of the technology. It fails because companies start with a strategy instead of a process.
Step 1: Find the most expensive process, not the easiest one
The common advice is to start with a quick win—a small, simple process that’s easy to digitize. That’s the wrong place to start. A simple process is usually simple because it costs little. You invest effort and get a result that nobody notices.
Instead, start by looking at where money is being lost. For one week, track how many hours two or three candidates actually spend on the job, and calculate the full costs:
- Quotation preparation: 12 hours/week × 48 weeks × 55 EUR = 31,680 EUR per year
- Project Coordination: 15 hours/week × 48 weeks × 55 EUR = 39,600 EUR per year
- Auditing: 6 hours/week × 48 weeks × 45 EUR = 12,960 EUR per year
For a freight forwarding company or a technical B2B service provider, the most expensive process is almost always order or quote coordination, not accounting. The most expensive process is your first one. It bears the brunt of the transition effort because it unlocks the largest amount of savings, and it wins over the team because the relief is immediately noticeable.
Decision at the end of Step 1: A single case, with a year listed after it.
Step 2: Understand the process before digitizing it
The most common mistake in Step 2 is to digitize the process exactly as it appears on paper, rather than as it actually operates. On paper, the quote creation process has five steps. In practice, it has five steps and seventeen exceptions: the key customer with special terms, the rush order on Friday afternoon, the partial delivery that’s billed differently.
These exceptions aren’t just noise. They are the process. A system that only maps the normal case pushes precisely those cases back into Excel and email—the ones that take the most time. You’ve gained nothing. Document the current state with the people who carry it out every day, including special cases. If you digitize a broken process, you’ll end up with a process that breaks even faster.
Here’s an example from a freight forwarding company with 90 employees: On paper, the quotation calculation follows a fixed formula based on distance, weight, and rates. In reality, however, the two most experienced dispatchers calculate toll charges differently for regular customers, grant discounts on return loads that aren’t documented anywhere, and estimate wait times based on experience. It is precisely this unwritten knowledge that determines the margin, and it is precisely this that causes a system that relies solely on the formula to fail.
Decision at the end of Step 2: A process map that includes the most important exceptions, as approved by the people who carry out the process.
Step 3: Choose the approach: buy, configure, or build. How much does AI cost for small and medium-sized businesses?
Only now, with a clear understanding of the process, does the question of tools make sense. For a single process, there are three approaches, and none of them is inherently correct:
Buy. For standard tasks that don’t offer a competitive advantage—such as accounting, time tracking, or a traditional CRM—off-the-shelf software is the right solution. You’re buying licenses, not custom development. If your process is the same as what thousands of other companies do, don’t build it yourself.
Configure. Some platforms can be customized to the point where they can map a specific process without the need for custom development. This works as long as your special cases fit within the platform’s framework. As soon as you start pushing the limits of the configuration, this approach becomes more expensive than the next one.
Build. If the process is at the heart of your business knowledge—such as the cost calculation logic of a freight forwarding company or the scheduling rules of a service provider—and that is precisely where time or profit margins are being lost, then custom development is warranted. For a single operational function, a realistic starting point is around 50,000 EUR; for an end-to-end core operational system, the total cost ranges between 75,000 and 150,000 EUR. We’ve included the complete cost breakdown with calculation methods in the article What Digitalization Really Costs Small and Medium-Sized Businesses written separately so that this one stays in the correct order.
In practice, the answer is often a mix: off-the-shelf accounting software, a customized CRM platform, and custom development focused solely on the one process that gives you a competitive advantage. The trick isn't to choose one camp or the other, but to place each function at the right level.
The honest assessment: Does your process represent a competitive advantage or a standard task? In the first case, you build it; in the second, you buy it. Whether in-house development, an agency, or a hybrid model handles the development is determined by this comparison In-House Development vs. Agency.
Decision at the end of Step 3: Buy, configure, or build—depending on whether the process is standard or based on operational knowledge.
Step 4: Build on a small scale, but for actual use, not as an experiment
If Step 3 boils down to “building,” the next fork in the road arises in Step 4: prototype or production system. The instinct to first build a quick, disposable prototype “just to try it out” may seem prudent, but it’s expensive. A prototype without error handling, access control, and clean interfaces will either be discarded or—more commonly—go into production even though it was never built for that purpose. Then your operations will be running on a makeshift solution.
It’s the right approach to start small. “Small” means one process, not ten. “Small” does not mean leaving out the components that turn software into a working tool. Build that one process completely and ready for production, rather than five processes halfway done. The article explains why operational software must be designed for production from the very first version—not for learning. MVP vs. Production System.
Decision at the end of Step 4: A process designed for production, not a disposable prototype.
Step 5: Test with real orders, running in parallel with an end date
You don't test a new system with made-up scenarios. You test it with real tasks from the current week, handled by the people who will be working with it later. Fictitious test cases confirm that the software does what you expected. Real tasks reveal what you didn’t expect. That is exactly the point.
The most dangerous phase is when the old and new processes run in parallel. It feels safe, but it doubles the workload: Every order is entered twice, every decision is made twice. Set an end date for parallel operation right from the start. Without a fixed transition date, it will continue until an exhausted key employee pulls the plug. This article discusses how to manage this phase without burning out your operations team. Change Management in Digital Transformation In detail.
Decision at the end of Step 5: A fixed switchover date, communicated before parallel operation begins.
Step 6: Put one process into production, then move on to the next one
The transition takes place on a process-by-process basis, not as a single, company-wide deadline. A process goes live, stabilizes over the course of a few weeks, becomes the team’s full responsibility, and only then does the next one begin. A single big deadline when everything switches over at once is the surest way to make a transition fail: when three processes are faltering at the same time, it’s impossible to tell which one is causing the problem.
Specifically, “process by process” means this: First, the quote creation process will run entirely through the new system, while scheduling and billing will remain as they are for the time being. Once quote creation is running smoothly, scheduling will follow, then billing. Each step is self-contained and can be reversed if something goes wrong.
This is exactly where a tool designed for a single process gradually evolves into an operational platform that forms the core of the business. Stabilize one process, then move on to the next: This sequence is also the answer to the uncontrolled proliferation of SaaS tools that occurs when companies buy yet another tool for every problem. The article explains why more tools don’t solve operational complexity. SaaS Proliferation and Operating Platform.
From the very beginning, factor in ongoing costs for operation and further development: 15 to 20 percent of the construction cost per year. A system that isn’t upgraded after six months of use becomes a burden rather than a tool.
Decision at the end of Step 6: A live, stabilized process and a planned operating budget before the next process begins.
Why Most Digital Transformation Projects Get Stuck at Step 0
The six steps aren't difficult. The reason companies don't take them is almost never technical. It's because no one in the company is given the time to do so. Digital transformation is assigned "on the side" to someone who's already working at full capacity, and it fizzles out amid day-to-day operations and good intentions.
If you’re serious about following this sequence, it must begin with a decision that isn’t included in any of the six steps: appointing a person in charge who has formally allocated time for the task. Calculate the cost of this time: A project manager whom you release from other duties for six hours a week over six months to implement the project will cost around 10,900 EUR at a full cost of 70 EUR per hour. Compared to a process that ties up 39,600 EUR per year, as in the example above, this isn’t a budget issue—it’s a matter of priority. We’ve outlined the questions you should ask an implementation partner before hiring them in Choosing a Software Development Partner: 10 Questions compiled.
What to do now
Don't start with a strategy. Start with a measurement. This week, pick two or three operational processes and have someone record how many hours they actually take up. Calculate the full costs. The highest number is your Step 1.
The process then proceeds as follows: understand the process, choose a level, build a production-ready version, test it with real orders, and roll it out step by step. No step requires you to overhaul the entire company all at once. Each step ends with a decision you can make.
If you want to know which process is Step 1 for your business and how much it will cost to implement the change, we'll assess it and work through the calculations with you, using your own figures.
Arrange a free, no-obligation consultation with our team.
EN 
DE 
FR 
ES