Basic Aggregate Functions — Use COUNT, SUM, AVG, MIN, MAX correctly

Before relational databases existed, computing summary statistics over large data collections required writing custom procedural code — iterating through files, accumulating totals in variables, and manually tracking extrema. This approach was error-prone, tightly coupled to data layout, and demanded that every analyst reinvent the same logic. The emergence of the relational model and its declarative query language fundamentally changed this paradigm by embedding aggregation directly into the data retrieval layer. Rather than telling the system how to loop through records, analysts could simply declare what summary they needed, and the database engine would determine the most efficient execution strategy.

The central question these aggregate functions address is deceptively simple: how can a declarative language collapse an arbitrary number of rows into a single scalar result while correctly handling NULL values, duplicate entries, and empty sets? Mastering these five functions is the prerequisite for every subsequent SQL aggregation topic, from GROUP BY partitioning to window functions and OLAP extensions.

An aggregate function is a set function that accepts a multiset (bag) of values drawn from a column and returns a single scalar value. Unlike scalar functions that operate on one row at a time, aggregates consume the entire group of rows that survives the WHERE filter (or a partition defined by GROUP BY). The SQL standard defines five fundamental aggregates, each satisfying distinct statistical or counting needs. Understanding their behavior requires internalizing several cross-cutting principles that govern how all five functions interact with NULL values, the DISTINCT keyword, and empty input sets.

The diagram above illustrates the fundamental transformation that every aggregate function performs: a multiset of row-level values enters on the left, and a single scalar exits on the right. The critical detail to observe is the asymmetry between COUNT(*) and COUNT(amount). The former counts all eight rows because it tests for row existence, not column value. The latter counts only the six rows where amount is not NULL. This distinction propagates directly into AVG, which divides SUM by the count of non-NULL values (6, not 8), yielding 258.33 rather than 193.75 — a difference of roughly 33% that can silently corrupt analytics if misunderstood.

Although SQL is declarative and hides implementation details, understanding the mathematical semantics of each aggregate function clarifies edge cases and prevents misuse. Let S denote the multiset of non-NULL values drawn from the target column within a given group, and let N = |S| represent its cardinality.

The interplay between NULL values and the DISTINCT qualifier creates subtle behavioral differences across the five aggregate functions. The table below provides a comprehensive reference that maps every combination of function, qualifier, and edge condition to its expected result. Study it carefully, as exam questions and real-world debugging scenarios frequently exploit these corner cases.

The second diagram reinforces a point that cannot be overstated: AVG never divides by the total row count. It always divides SUM by the number of non-NULL values. If you want the average to treat NULLs as zeros (perhaps because a missing sales figure means no sale occurred), you must explicitly convert them: AVG(COALESCE(col, 0)). This forces the denominator to equal the total row count by substituting zeros for NULLs before the aggregate computes.

Consider an e-commerce database with a table order_items containing the following data. We want to generate a summary report showing the total number of line items, total revenue, average unit price, cheapest item price, and most expensive item price — grouped by product category. We will also need to count how many distinct products appear in each category.

Even experienced developers encounter subtle bugs when using aggregate functions. The following table catalogs the most common pitfalls, contrasting the incorrect assumption with the actual SQL behavior and providing the corrective pattern.

The five basic aggregate functions serve as the foundation upon which SQL's more advanced analytical capabilities are built. Understanding COUNT, SUM, AVG, MIN, and MAX thoroughly is a prerequisite for window functions, OLAP extensions, and user-defined aggregates. The table below maps each basic aggregate concept to its advanced counterpart, illustrating how the same mathematical operations are reused in progressively more powerful contexts.

The conceptual leap from basic aggregates to window functions is arguably the most important transition in intermediate SQL. Window functions reuse the exact same five aggregate operations but alter the scope of computation: instead of collapsing an entire group into one row, they compute the aggregate over a defined window frame and attach the result to every row in the partition. Once you have internalized how SUM, AVG, and COUNT behave with GROUP BY — particularly their NULL handling and DISTINCT semantics — extending that understanding to OVER clauses and frame specifications becomes a natural progression rather than a conceptual hurdle.