An assist is not as universal a statistic as a goal

A goal has a direct place in football's Laws: the ball must wholly cross the goal line between the posts and under the crossbar, with no prior offence by the scoring team. An assist is different. It is an analytical attribution attached to the action before a goal, and the detailed decision rules are supplied by competitions or data providers rather than one universal passage in the Laws of the Game.

That distinction explains why two reputable websites can publish different assist totals for the same player. The disagreement may concern a deflection, a rebound, an own goal, a penalty won, or how much control a scorer had after receiving the ball. Neither total should be copied into a database without provenance. A number becomes auditable only when readers know whose definition produced it.

Opta's published event definitions illustrate the judgment involved. Its assist categories distinguish intentional final passes from other contributions and describe how deflections and defensive touches affect attribution. The important lesson is broader than one provider's rule: an assist is the result of a classification system. Consistency within that system is valuable, but it does not transform the classification into a universal law.

Provider changes can affect comparisons across time. If a database combines seasons collected under different definitions, the career total may contain a hidden methodological break. Historical corrections are normal and useful, but they should update provenance and timestamps. Silently mixing totals from separate sources risks presenting an exact-looking figure that cannot be reproduced.

Assists record only the final classified action before a goal. A line-breaking pass earlier in the move, a dribble that draws two defenders, or movement that creates space can be tactically decisive without earning one. Conversely, a routine pass can receive an assist when the scorer produces an exceptional finish. This is not a defect in the statistic; it limits the question it answers.

Expected assists describe chance creation rather than realised teammate finishing. In Opta's definition, xA estimates the likelihood that a completed pass becomes a goal assist using characteristics of the pass. It adds context but remains model-dependent: different inputs and training data can produce different values. Actual assists and expected assists should be labelled separately, never merged.

FootIndex retains the source definition, avoids filling absent assists with invented zeros, and compares players within a consistent scope. Readers should treat assist totals as evidence about recorded final-ball production—not a complete ranking of creativity. Interpretation should mention coverage, provider and the contributions the metric cannot see.