Productivity in SMEs: Simplify First, Then Automate

A process can be slow even when everyone involved works hard. An enquiry arrives by email. Information is copied into a spreadsheet. A quotation is prepared, passed on for review, corrected and later entered again into another system.

In between, questions come back because a detail is missing or an exception was never agreed. When pressure rises, automation sounds like the logical answer.

But technology does not distinguish between a meaningful step and a detour that grew over the years. It executes what has been defined. At best faster. At worst it makes an unnecessarily complicated workflow faster, more rigid and harder to change.

Why bad processes survive for so long

Many workflows in SMEs were never planned. They emerged. An extra check was introduced after a mistake. A spreadsheet appeared because the existing system did not cover something. One person took over an approval because they knew the customers particularly well. Every step once had a reasonable justification.

Over time the context disappears. The step remains. New employees learn what is done, but not always why. Processes grow in layers until nobody can say precisely which parts create value and which only administer the past.

Automation is an amplifier

Good automation reduces repetitive work, improves reliability and makes information available in time. For that, the underlying workflow has to be stable enough. If it is full of unresolved exceptions, missing data and shifting decisions, technology usually just moves the problem.

This explains why some implementations are technically successful and operationally disappointing. Interfaces work, tasks are generated automatically, and processing time stays the same. Perhaps staff now maintain additional fields. Perhaps the old spreadsheet continues in parallel. Perhaps the case still waits at the same approval.

Do not automate a workflow you cannot explain, simplify and take responsibility for.

The five steps before automation

1. 1.

   Make the real workflow visible

   Choose one concrete case and follow it from trigger to result. Work with the people who actually run it. Note handovers, waiting times, follow-up questions, media breaks and exceptions. A realistic picture is worth more than a formally perfect process diagram.

2. 2.

   Clarify the intended result

   What has to be better at the end? Faster response, fewer errors, more transparency, less effort or a more reliable customer experience? Without a clear result, process work easily turns into a debate about personal preferences.

3. 3.

   Remove, combine and decide

   Test every step: does it create value, reduce a relevant risk or is it legally required? If not, it can go. Several checks may be combined. Some follow-up questions disappear when the required information is captured as mandatory at the start. Some approvals become unnecessary once decision limits are defined.

4. 4.

   Separate the normal case from exceptions

   Automation suits repeating, sufficiently clear patterns. Exceptions often need human judgement. A good process does not force every case into the same logic. It lets the normal case run efficiently and routes genuine exceptions deliberately to the right person.

5. 5.

   Only now choose the technology

   Once workflow, result, data and ownership are clear, the technical choice becomes easier. Then you can examine whether existing systems are enough, whether an integration is needed or whether a new solution adds real value. The question is no longer "What can the tool do?" but "Does it reliably support the workflow we want?"

The wastebasket test

Before a team digitalises a process, it should ask together: which steps would we not introduce again today? This wastebasket test gives permission to question historical habits. It works best when every role involved is allowed to explain where it waits, duplicates work or has to add information afterwards.

• Which information is captured more than once?
• Which approval almost always passes without any change?
• Which handover produces the most follow-up questions?
• Which exception has become so frequent that it is now the normal case?
• Which private spreadsheet is actually keeping the official process running?

Choosing a sensible pilot

The best pilot is not necessarily the biggest process. A good candidate repeats often enough, has a clear business effect and involves a manageable group of people. At the same time, the company should be willing to genuinely change the process. Anyone who only wants to transfer the existing way of working untouched into software will learn very little.

A pilot needs a baseline. How long does the case take today? How many follow-up questions or corrections appear? Where do customers or employees wait? After the change, the same points are reviewed again. That is how you see whether the improvement really arrived in daily operations.

Productivity is more than saved time

Time matters, but it is not the only benefit. A clear process can reduce errors, keep customers better informed, get new employees productive faster and lower dependency on individuals. Often the greatest value comes from attention becoming available again for decisions and customer work.

For the leadership team that means an automation initiative should not be judged on technical features alone. It needs a process owner, a relevant result, good data, safe usage and a rhythm for later improvement.

How wis.dom|digital™ approaches it

We start with the business, not with the technology. Together we clarify which workflows and which information have the greatest influence on customers, efficiency and growth. Only then do we decide which digital capabilities, integrations or automations make sense, and how they will be supported over time.

Frequently asked questions

Which processes are suitable for automation first?

Repeating workflows with clear rules, available data and measurable effect. Steps with a lot of manual transfer, repeated checking or unnecessary waiting time are particularly attractive.

Does a process have to be fully standardised?

No. The normal case should be clear. Genuine exceptions may remain, but they have to be recognised and routed to people with the necessary decision authority.

When does a learning solution beat classic automation?

When content has to be interpreted, summarised or classified and sufficient quality control is possible. Deterministic rules remain the better choice for many stable, high-risk or audit-relevant steps.

Sources and editorial references

• KfW Research, KfW SME Digitalisation Report 2024
• Google Search Central, AI features and your website

Hasan H. Hasic

Hasan H. Hasic is an entrepreneur, advisor and Key Person of Influence at wis.dom|bridge™. His focus is on connecting entrepreneurial clarity, durable structures and digital delivery that creates measurable value in daily work.