Connect a resistor to an adjustable power supply and measure the current at a few voltages. A 4 Ω resistor gives 0.5 A at 2 V, 1.0 A at 4 V and 1.5 A at 6 V. Each time the voltage doubles, the current doubles, and the ratio V/I stays at 4 Ω. Plot V against I and the points fall on a straight line through the origin whose slope is the resistance. That proportionality is Ohm's law.
The law ties three numbers to one component: the voltage across it, the current through it and its own resistance. Suppose two resistors in a single loop share a 12 V battery. Neither resistor has 12 V across it, so pairing the battery's 12 V with one resistor's ohms gives a current that is too large. To use V=IR on the whole circuit, R must be the equivalent resistance of everything the current passes through.
Not every part obeys the law. A filament lamp's V-I graph bends over, because the filament heats up and its resistance rises as more current flows: a small lamp might measure about 2 Ω cold and 24 Ω glowing at full brightness. A diode hardly conducts until about 0.6 V and then lets the current shoot up. You can still divide V by I at any single point, but the answer changes from point to point, so one value of R will not predict the current at another voltage.
Think of it like this. Ohm's law behaves like water forced through a pipe packed with sand: double the pressure difference and twice as much water seeps through each second; pack the sand tighter and less gets through at the same pressure. The comparison fails for parts like lamp filaments, whose 'sand' packs tighter as they heat up.
Core idea. For an ohmic component, the current is proportional to the voltage across it, and the constant ratio V/I is its resistance; every use of V=IR takes all three values from the same component.