CO2 is a dog. Actually heavier than air particles, CO2 gets swirled around the entire room when entering or opening a door.
Plaster can absorb and release CO2.
CO2 sensors drift over time. Many calibrate themselves. Some have to be manually calibrated over time (0 point calibration).
The CO2 measuring cells have fluctuating tolerances depending on temperature. Typically, tolerances at 21° are specified as 10% of the measured value or ~ ±75ppm. Now, of course, it depends on the measuring range. In the worst case, you have 2 sensors with maximum tolerance up and down, which corresponds to a difference of 150ppm.
All this causes measurement deviations or fluctuations.
You see, a CO2 sensor is a better indicator for air quality and, metrically speaking, rather an estimation tool.
Roughly applies:
0...750ppm = very good air quality
751...1250ppm = good to moderate AQ
1251...2000ppm = moderate to poor AQ.
With two people in a closed bedroom, you easily reach between 5000-10,000ppm (depending on room size and sleep duration).
Besides the CO2 sensor, it would be interesting to install a VOC sensor as well. It measures multiple gases, such as adhesives from materials, cleaning agents, cigarette smoke, food odors, and many other gases that can cause headaches and migraines.