How to choose the right process to automate.
The best first automation project is rarely the most advanced one. It is a well-defined workflow with clear rules, usable data, and an effect that can be measured.
In short
- Start with a process that is repeated often and already follows fairly fixed rules.
- Measure the baseline so time savings, quality, and fewer errors can be documented.
- Keep human control over exceptions and high-risk decisions.
- Build the smallest useful version first and expand based on actual use.
What is IT automation?
IT automation means that software performs agreed steps in a workflow that would otherwise require manual action. This can be moving information between systems, validating data, creating documents, sending reminders, or preparing a case for human review.
An automation does not have to be a large platform. Often it consists of a controlled connection between the systems the company already uses. The goal is not technology in itself, but less waiting time, fewer repetitions, and a more consistent process.
Five signs of a good candidate
- The task is repeated often. A small saving has a large effect when the same action is performed many times.
- The rules can be explained. If the employee can describe which information is checked and what the next step is, the logic can often be turned into software.
- Data is moved manually. Copying between email, spreadsheets, CRM, the finance system, and other tools is both slow and prone to errors.
- Errors have a clear cost. Forgotten follow-ups, incorrect data entry, or missing documentation make the benefit easier to measure.
- The process has an owner. Someone must be able to decide the rules, approve changes, and respond if the solution encounters an exception.
A process is less suited as a first project if it changes every week, relies on the tacit knowledge of a single person, or requires complex judgment in almost every case. Here the process should first be simplified and documented.
Assessment model
Six questions before you prioritize
Frequency
How often is the task performed?
Time spent
How many hours does the organization spend in total?
Rules
Can the decisions be described clearly?
Data
Is the necessary information available and consistent?
Errors
Does manual handling create errors or extra checks?
Risk
What happens if the automation gets it wrong?
Give each point a simple rating from 1 to 5. High frequency, high time spent, clear rules, and good data pull the score up. High consequences from errors, on the other hand, require more control and can make the process a poor first project.
Example: From a shared inbox to a prepared case
Imagine a shared inbox where employees receive orders and attachments. Today an employee opens each email, saves the attachment, copies information into a system, and sends a confirmation.
A first automation can monitor the inbox, check the sender and file type, retrieve the necessary information, and create a draft item in the relevant system. The employee approves the draft, while uncertain or incomplete cases are sent for manual handling.
The point is important: The automation does not have to make every decision. It can remove the mechanical steps and prepare the material, while a human keeps control over exceptions.
Fixed rules or AI?
Use fixed rules when the input and decisions are predictable. A customer number can be validated, an amount can be compared, and a status can be updated without AI. It is cheaper, easier to test, and more transparent.
AI can be relevant when the solution needs to understand free text, variable document structure, or other unstructured information. Even here the result should be validated, and there must be clear limits for when an employee takes over.
Build a simple business case
Start with the current time spent: number of cases times minutes per case. Add the cost of errors, waiting time, and rework. Then compare with development, operations, and the time employees will still need to spend on checks.
A good business case also includes quality. Faster response times, more consistent documentation, and fewer interruptions can be more important than pure hours saved.
From idea to operation
A six-step plan
- 1Describe the process as it actually works today.
- 2Measure time spent, waiting time, errors, and the number of repetitions.
- 3Find the rules, exceptions, and the places where people need to make a judgment.
- 4Scope a first version with one clear goal.
- 5Test with real data and get feedback from the people who do the work.
- 6Put monitoring and ownership in place before the solution goes into stable operation.
Typical mistakes that make the project unnecessarily hard
- Automating a bad process. Remove redundant steps before building them into software.
- Starting with the biggest process. A smaller, measurable area gives faster learning and lower risk.
- Overlooking the exceptions. The rare cases often determine how robust the solution needs to be.
- Lacking an owner. An automation must be maintained as systems, rules, and needs change.
- Measuring too late. Without a baseline it is hard to document the effect.
The good next step
Choose three manual workflows and describe each on a single page: who starts the process, which systems are involved, how often it runs, which exceptions exist, and what a good result is. Then use the six criteria to choose the most well-defined candidate with a clear effect.
If the process requires a new application rather than an integration, our page on application development can help scope that solution. If the task is primarily about existing workflows and systems, you can read more about our help with IT automation.