Treat a set of equations as a list of demands, and a solution as a choice that meets every demand at once. "Two numbers add to 4" and "their difference is 2" can both be met: 3 and 1 do it, so the system x+y=4, x−y=2 is consistent. "Two numbers add to 5" and "the same two numbers add to 7" cannot both be met by any pair, so x+y=5, x+y=7 is inconsistent. Consistency asks only whether the demands are compatible, not how to meet them.
Graphs show it directly. Each linear equation in two variables is a line, and a solution is a point on every line. Lines that cross give a consistent system with one solution. Lines that coincide give a consistent system with infinitely many. Only separate parallel lines give an inconsistent one. In algebra the signal is reaching a false statement with no variables left, such as 0=2; reaching 0=0 is harmless and means one equation repeated another.
Consistency is easy to confuse with having exactly one answer. The system x+y+z=6, x+y=4 has fewer equations than unknowns, yet it is consistent: (1,3,2) and (4,0,2) both work. Too few equations to pin down every value is a statement about how many solutions there are; inconsistency means there are none at all.
Think of it like this. Consistency is like a set of puzzle clues: "the number is even", "it is between 10 and 15", "it is divisible by 3". The number 12 meets all three, so the clues are consistent. Add "it is odd" and no number works. Whether one number fits or several, the clues count as consistent.
Core idea. A system is consistent when some set of values satisfies every equation at once.