After this long at DBS, I've noticed something: writing code matters, but it takes up a much smaller share of the job than you might think. I'm writing this down so I can come back to it whenever I start a new project.
What the job really needs, most of the time, is context: an understanding of the whole story behind the work.
§1First, understand what problem it solves
When a colleague or a client brings you a requirement, the first thing to understand isn't what code to write. It's the context: what is this requirement for? What problem is it meant to solve?
§2Then, work out how to complete the puzzle
Once you understand that, the real work begins: how will you help them solve the problem?
- Where do you find the data you need?
- Which department should you contact?
- Which department do you get this data from?
- How should the whole thing move forward?
Much of the time, you aren't writing code at all. You're piecing together data and requirements. Someone gives you a goal, and you have to find the pieces scattered around and fit them into one complete puzzle that makes the goal possible.
That's why context matters so much. You need to know the background and what you're doing before you can deliver it well.
§3Also, work out what to prepare ahead
Understanding the context isn't only about knowing where to get the data. It also shows you what you need to prepare in advance.
We ran into exactly this. At first, we thought we only needed our own database to store the data, plus a few APIs from different departments, and that would be the end of it.
But we missed a lot:
- the design might need Kafka, a message broker;
- there might be backend problems to solve;
- we had to request some internal infrastructure services;
- we had to apply for special access to read the logs after deployment;
- and all sorts of other small things.
Test data is the same. Where do the test users come from? If the system has several permission levels, you need a test account for each level, ready in advance.
None of this is writing code, yet all of it has to be in place before the work can be done.
§4Knowing how other departments work is a bonus
If you understand, even roughly, how other departments operate, the work goes much more smoothly, and you'll feel sure of whatever you need to do.
§5Same data, different names
We usually know what the data should look like in the end. But the same field might be called username on our side and profile name on another. In a third department, it might have a name specific to their own system, and in yet another, it might be business name.
This was a huge problem for us. We spent a lot of time understanding and aligning what each name actually maps to on our side.
It matters a great deal, because the data we show has to be 100% correct.
And when something goes wrong in production, we can tell the other side with confidence: "We've checked this. It isn't our problem." That gives our team, and each of us, solid ground to stand on.
§6Only writing code is a fatal weakness
I've noticed that some developers are excellent at writing code but don't want to join meetings or understand the requirements. What they build may be wrong, or may miss the goal.
In meetings, they often can't answer anything. They know what they wrote, but not the story behind it. So when the product owner or the business analyst asks a question, they have no answer.
That is a fatal weakness.
§7Closing thoughts
All in all, most of my time goes into understanding requirements, piecing together data, aligning field names and preparing everything else. That's what it takes to really finish a task. Writing code matters, but it has to rest on an understanding of the context.
§8Before the next project, ask myself
- What is this requirement for? What problem does it solve?
- Where is the data I need? Which department do I contact, and who do I ask?
- How do the departments involved work?
- Besides the database and APIs, what else do I need? Infrastructure services, access, test accounts…
- What is the same field called in each department? Does each name mean the same thing?
- If the product owner or the business analyst asked me right now, could I answer?