A universal property characterizes a mathematical object by specifying a unique morphism to or from every other object satisfying a given condition, rather than by internal construction. Objects defined by universal properties are unique up to unique isomorphism, which is often the strongest form of uniqueness available. Examples include free groups, tensor products, polynomial rings, products, and completions—all defined by how morphisms into or out of them behave, not by their internal set-theoretic construction.
Work through the free group on a set S: it is characterized by the property that every function from S to a group G extends to a unique group homomorphism. Verify uniqueness up to isomorphism: if two groups both satisfy this property, construct an isomorphism between them using the universal property of each.
You already know what categories, morphisms, and isomorphisms are. A universal property is a way of pinning down a specific object in a category not by describing its internal structure, but by specifying exactly how it interacts with every other object via morphisms. The classic example is the free group on a set S: instead of building it explicitly, you say "the free group F(S) is whatever group has the property that for every group G and every function f: S → G, there is a unique group homomorphism F(S) → G extending f." Any object satisfying that description is the free group, regardless of how it was constructed.
This external characterization style may feel unfamiliar if you are used to defining things by their elements. The payoff comes from the uniqueness theorem: if F₁ and F₂ both satisfy the same universal property, then the universal property of F₁ gives you a morphism F₁ → F₂ (because F₂ is "one of those other objects"), and vice versa. Composing them gives an endomorphism that must equal the identity by the uniqueness part of the property. So F₁ and F₂ are uniquely isomorphic. This is the sense in which universal properties define things uniquely — not by picking out one construction, but by guaranteeing that all constructions are canonically the same.
The power of this perspective is that it applies far beyond free groups. Products, coproducts, limits, colimits, tensor products, polynomial rings, and many other constructions all have universal property descriptions. Once you know an object is the limit of some diagram, you can derive everything about how morphisms into it behave without knowing anything about its internal construction. Proofs that would require messy element-chasing in one construction become clean one-line arguments using only the universal property.
One common confusion: "unique up to unique isomorphism" does not mean unique up to equality. You should expect many different models — the integers ℤ and the free group on one generator are different constructions, but they satisfy the same universal property and are therefore isomorphic. The categorical viewpoint treats these as essentially identical. This is why mathematicians sometimes speak loosely of "the" product or "the" limit — they mean the unique-up-to-isomorphism object, not a specific construction. Keep this distinction crisp and the rest of category theory will make much more sense.