There is a persistent myth about data analyst jobs: that most of their time is spent creating beautiful graphs and sophisticated models. The reality is far more mundane. Most of the time is actually spent cleaning messy data before any number can be trusted.
A frequently cited industry survey states that data analysts spend about sixty to eighty percent of their time preparing and tidying up data. The parts considered "exciting," such as analysis and visualization, actually take up a small portion. Understanding this reality changes the way people learn data.
Why raw data is always messy
Real data is never as tidy as examples in textbooks. Dates are written in three different formats in one column. City names are misspelled. Empty rows appear for no reason. Sales figures are recorded twice due to different systems. As long as the sources are human and many applications are not interconnected, this chaos is almost certain to occur. The analyst's first task is not to calculate, but to ensure that what is being calculated is indeed trustworthy.
SQL for retrieval, Python for tidying
These two tools complement each other, rather than replace one another. SQL excels at retrieving large amounts of data directly from databases, filtering, and efficiently joining tables. Python, through libraries like pandas, takes over when the work becomes complicated: filling in missing values, standardizing formats, detecting anomalies, and merging sources with different structures. Many analysts level up precisely when they stop choosing one and start using both in a single workflow.
From clean data to decisions
Once the data is tidy, exploration becomes meaningful. This stage is called exploratory data analysis: looking for patterns, observing distributions, testing initial hypotheses before drawing conclusions. The results can be very practical. Marketing teams know which products are actually selling, operational managers find where costs are ballooning, and decision-makers no longer have to guess. But all of this relies on one thing: clean data from the start. Insights built on dirty data are merely false confidence.
Leveling up as a data analyst is not about memorizing the most complicated formulas. More often, it is about the discipline of tidying up the mess, then letting honest data speak for itself.
References:
- Anaconda – State of Data Science Report (data preparation time) → anaconda.com
- pandas – Documentation: working with missing data → pandas.pydata.org
- Badan Pusat Statistik – Statistik tenaga kerja dan ekonomi digital → bps.go.id