Farewell, My Little Workflow Engine

A quick hack to get a first shippable product out the door can survive a lot longer than you think. My little story shows you how things can end up if you do the right thing by pure chance.

Back in 2015 I wrote a small workflow automation engine for our second largest application. We were moving away from SharePoint and needed something to automate our minimalistic workflow.

My goal was simple: to be better than the SharePoint workflow system, which produced an error every 10 workflows. I spent about 2 days on this little hack, including all the data processing and PDF creation. The idea was to invest a little time, start production with it and then replace it with something proper when we have more breathing space.

When we ran our first performance tests, we realised that this engine was fast. It was severely limited and could not process anything other than our small workflow, but it did so at high speed. In the time it took us to process 100 workflows on SharePoint and freeze the old application, our little hack processed 10,000 workflows without anyone noticing anything in the new application. So, we decided to take the new engine into production and see how it works under real conditions. Since we had almost no problems, we kept it as it was and never replaced it.

Fast forward to the year 2025. When renewing the application, we decided not only to move a few workflow steps to Hangfire.io, but to migrate the entire workflow engine. Thus, bringing an end to our little hack.

In its lifetime, the little engine has processed 97,000 workflows with 250,000 steps. And all this with less than 30 errors in total or one error for every 3,200 workflows. That was a gamble that paid off immensely.

How was this success possible? The key factor was to understand the problem in its entirety and only solve the actual problem – not anything that could happen sometimes in the future. If the workflow does not change in 10 years, there is no need for flexibility.