Two weeks of sensing nothing but data

Every guide on the internet will tell you a moisture percentage to water at. None of them know your soil, your pot size, your light, or your sensor's particular calibration. Two weeks of logging tells you all four.

Your sensor does not measure moisture

A capacitive soil sensor measures capacitance, which varies with water content and also with soil composition, compaction, temperature, and how deep you pushed it in. The number it gives you is not a percentage of anything; it is a reading you must learn to interpret for your specific setup.

This is why the logging fortnight is not optional. You are not collecting data about plants — you are calibrating an instrument against your own conditions, exactly as you would for any other measurement.

Log every fifteen minutes for two weeks

Soil moisture, air temperature, humidity, and light if you have it. Timestamped, to an SD card or a local file. Nothing actuates.

Water by hand as you normally would, and log when you did it. Those manual events are your labels — they show what the sensor reads at the moment an experienced human decided to act.

The natural range of your soil reading

Highest reading minus lowest, across the fortnight. If it is small, your sensor placement is not seeing the variation the plants experience, and the loop you build on it will be blind.

The reading at which you watered by hand

The average sensor value at the moments you decided to water. This is your candidate setpoint, derived from your own judgement rather than from a forum post.

The range came out tiny

After two weeks your soil reading barely moves, even right after watering. What is the most likely cause?