I spent time at a water treatment plant years ago doing consulting work during a system upgrade. I was not the expert they needed, but I picked up the work anyway and learned more than I expected. One thing stayed with me: the question of when to filter. It is the same question I face constantly in my business and personal life. Do I filter early, before I process anything, or do I collect everything first and filter later?
Water treatment systems are built around two truths that seem to contradict each other. You cannot apply precision filtering to unfiltered chaos. And you cannot detect certain problems until the system has processed the data enough to reveal them. Both are true. The trick is knowing which applies when.
If you filter too early, you create different problems than if you filter too late. At the plant, if they tried to remove all small particles right at the start, they would clog the filters, waste chemicals, and drive costs through the roof. Worse, they might miss contaminants that need chemical binding before they can even be detected. In business, this looks like over-qualifying leads before you understand what you are looking for. You end up throwing away hidden opportunities because you filtered before the value became visible.
But if you filter too late, you let everything through. Large debris damages equipment. Systems overload. Costs explode in a different direction. In business, this is what happens when you let every inquiry, every request, every idea reach your team without any screening. People burn out. Efficiency drops. Signal drowns in noise.
The plants I worked with followed a consistent pattern. They removed the obvious junk early with coarse screening. Then they transformed the material through coagulation and enrichment. Only after that did they apply precision filtration. Final validation came last, right before output. It was a sequence, not a single gate.
“Don’t try to purify water with a microscope first.”
I think about that when I am tempted to apply high precision thinking to unrefined input. It does not work. The input has not been shaped yet. But I also think about it when I am tempted to let everything pile up without any early sorting. That does not work either. You throw away material before it has a chance to reveal structure.
The answer is not to filter early or late. The answer is to filter at each stage, with the right tool for that stage. Coarse first. Precision later. Transformation in between. I have started asking myself a different question now. Not when do I filter, but what kind of filtering does this stage require? That small shift has changed how I handle incoming work, how I sort ideas, and how I decide what deserves deeper attention. The next time something lands in front of me, I will ask what stage it is in before I decide how to screen it.
