Reading / Synthetic data
Examples made for exploration
A generated example can help examine a digital experience without being a record of something that actually happened.
Make the distinction visible
Synthetic data is artificially generated information. It can take forms such as tables, text, or images. Depending on its purpose, it may follow simple rules or attempt to resemble selected properties of another dataset. Artificial origin and useful realism are separate questions.
In a product discussion, generated examples can make a situation concrete. Their role needs to remain clear: they illustrate conditions to examine, rather than establish facts about real people or prove that a system works in everyday use.
Hypothetical example · A crowded timetable
Imagine a timetable filled with invented session titles. Some are unusually long, some repeat, and others contain punctuation. That collection could help someone inspect how the timetable reads when space is tight. It would not show which sessions visitors want, how many people attend, or whether the timetable reflects an actual event. Every session in this example is fictional.
Look beyond a convincing sample
A realistic-looking dataset can still leave important situations out. Its usefulness depends on what was generated and what was excluded. Synthetic information derived from sensitive source material also needs careful evaluation; calling it synthetic does not itself establish privacy. Useful accompanying notes explain the generation method, intended purpose, and limits so that another reader can interpret the examples responsibly.
Further reading: IBM’s explanation of synthetic data.