What is descriptive analytics?
Descriptive analytics is the use of historical data to describe what has happened. It answers questions such as how many units shipped last month, what average handle time was by queue, or how revenue by region compared with last year. The methods are counting, summing, averaging, segmenting, and plotting over time, the staples of business intelligence reporting.
It is the first of four analytics stages often listed together. Descriptive analytics says what happened, diagnostic analytics asks why, predictive analytics estimates what comes next, and prescriptive analytics recommends what to do. Each stage depends on the one before it, and most companies get more value than expected from getting the first one right.
Descriptive analytics in practice
A community bank wants to understand its small-business lending. A descriptive analysis breaks loan originations down by branch, officer, industry, and size over three years and shows that two branches account for most of the growth while a third is flat. Nothing in it predicts or prescribes anything. It gives the chief lending officer a factual picture previously spread across five spreadsheets, and it points at the question for diagnostic work to take up next.
Descriptive analytics depends more on data quality than on tools. If the loan system records industry codes inconsistently, the breakdown by industry is wrong however well it is charted. Natural language query has made this stage more accessible, since a manager can type a question and have a model write the query against governed tables, but those tables have to exist first.