Have you ever found yourself spending hours looking for the right tool for the job? Or found yourself with a tool that doesn't do exactly what you need? Often you should just force your problem to fit the tools available. But sometimes it's time to bite the bullet and build exactly the tool you need. That's what I ended up doing when I needed to analyse the shape of BigPanda's traffic on our Kafka topics. I ended up with exactly the right tool for the job, and found myself reusing it again and again. In this talk I'll talk about what I learned. When is it time build a tool instead of buying one? What are the hidden benefits? And when is it just a bad case of 'Not Invented Here'?