What is data transformation?
Data transformation is the process of converting data from its raw form into a structure suitable for analysis, cleaning, reshaping, joining, aggregating, standardising formats and applying business logic. It is the 'T' in ETL/ELT and turns messy source data into consistent, usable datasets.
Raw source data is rarely analysis-ready: dates are in different formats, categories are spelled inconsistently, values need converting, tables need joining. Transformation is the work of making it clean and coherent.
Common transformations:
- Cleaning fixing nulls, duplicates and errors.
- Standardising consistent formats, units and codes.
- Reshaping pivoting, splitting or combining columns.
- Joining and aggregating merging sources and summarising.
- Deriving calculating new fields and business metrics.
In traditional stacks, transformation is a heavy, up-front step, you model and clean everything before anyone can query it.
iDBQuery shifts much of this to query time and to natural language. When you ask a question, it performs the joins, aggregations, filtering and derivations needed to answer it, on the fly, against your live data, so you don't have to pre-build transformation pipelines for every question. It also copes with messy inputs: its semantic layer interprets inconsistent or cryptically named fields, and you can standardise, dedupe or reconcile data simply by asking. You still get correct, cited results traced to source rows. Where you do maintain transformation pipelines feeding a warehouse, iDBQuery queries that output too, so it fits alongside your existing transformation work.
Updated 2026-06-22