Big data analytics
Big data analytics is the use of statistical and computational methods to find patterns in data whose size, speed or variety makes conventional analysis difficult. Businesses use it to understand behavior, detect problems and improve decisions.
How it works
A business collects data from sources such as transactions, website activity and sensors, then cleans and combines it. Analysts use database queries, statistical models or machine learning to answer specific questions. Distributed systems spread demanding workloads across multiple computers.
In marketing, this might reveal which customer journeys lead to purchases. In accounting, it might flag unusual transactions for review, not prove fraud.
A startup does not need a big-data platform just to track sales. Start with ordinary database tools; add distributed processing when the workload requires it.
The same $10,000 buys 500 customers or 200
A retailer joins high-volume website events with purchase records to compare campaigns. In this hypothetical month, each campaign spends $10,000 and uses the same attribution rule.
| Campaign | New customers | Ad cost per new customer |
|---|---|---|
| A | 500 | $10,000 ÷ 500 = $20 |
| B | 200 | $10,000 ÷ 200 = $50 |
Campaign A acquires 2.5 times as many customers for the same spend. That supports testing a budget shift, but customer retention and purchase margins could change the decision.
A spreadsheet can calculate these ratios. The big-data work is processing and reliably connecting the underlying records. These figures exclude other acquisition costs, such as staff salaries.
Questions people ask
What does a big data analyst do?
A big data analyst queries datasets, checks data quality, builds reports or models, and explains findings to business teams. Data engineers usually build the collection and processing pipelines, although the roles overlap.
What are the four types of big data analytics?
A common classification is descriptive analytics, which explains what happened; diagnostic analytics, why it happened; predictive analytics, what is likely to happen; and prescriptive analytics, what action to take. These are categories of analysis, not four mandatory steps.
