For people starting up with IoT this section contains a rundown of the most commonly available sensors, sensor types and the various brands that can be found, along with their flaws and advantages over other competitors.
Merchants will also typically label things like "buttons" as sensors, when, in fact, it is just a button. Truly, for such cases "things" is a better label because it provides an all-encompassing meaning for sensors and controls.
One of the glories of the wild capitalism in Asian countries, in particular, China is that producers tend to rip off each-other, by stealing each other's idea and then also having the infrastructure to produce the very same product and set it up at the same price or lower than the person coming up with the idea in the first place. Long story short, we wanted to purchase an USD100 sensor by Gravity, only to realize that a competing Chinese producer offered the exact same one for USD20 (the "Gravity" inscription was missing but the PCB and the sensor itself were all identical, along with the components). In other words, be aware that prices can fluctuate wildly and that by searching a little you might come across the exact same product as a knockoff and not all knockoffs are bad, especially if they stick to the original design.
"Meaningful" sensors will always be more expensive than "relative" sensors but be aware that engineering is intertwined with economics such that blasting away money on expensive sensors that might provide "meaningful" values is redundant and a waste of money depending on what the application requires. If you are looking to determine when it is time to water plants, then a "relative" USD7 "soil humidity" sensor that reports some meaningless value, the distinction only being made in terms of "more" or "less', would be perfectly suitable, but obviously if you're looking for a precise measurement, let's say, as in monitoring a plotting soil in general, then you might need to scale up the price. In that sense, "spending more money" is not always better given the constraints of a project and investing too much in one particular area leaves less money for the rest of the project.
"Starter kits", "engineering kits" and all of that are not too great and generally speaking they are way overpriced compared to getting the sensors individually. The main reason for that is that these kits contain a bunch of sensors that are unlikely to be used. The workflow should be to think of the project and then buy the necessary sensors. "IoT kits" are lined with sensors like photoresitor LDEs and MQ-sensors that are less than educational, even for a school-level background. LDEs can be bought in bulk and the circuit is trivial such that the module is overkill and MQ sensors need to be calibrated.
When purchasing sensors, before buying, always look online for:
and assess the consistency, maturity and congruency of the materials that are found. Very often you can find yourself the owner of a new sensor but without any documentation, or it turns out that there are no libraries for it, or that the documentation on how to read the sensor seems conflicted online with different formulas, or that simply is no Arduino code out there to deal with it. While most of the time it is possible to interface with an unknown and undocumented sensor, that would require knowing the specifications of a sensor extremely well, and as things go hand-in-hand, chances are that an unknown sensor does not even have a datasheet.