People's information needs vary widely. Peter Morville and Lou Rosenfeld describe four common categories: known-item seeking, where a person wants one specific, correct answer to a question; exploratory seeking, where a person wants information that adequately satisfies an immediate need without there being a single right answer; exhaustive research, where a person is conducting in-depth, comprehensive research on a topic; and refinding, where a person needs to locate the same or similar information they've encountered before.
These aren't fixed labels a person announces up front. A need can start out ambiguous and shift as the person learns more about a domain. Recognizing which need is in play, or which needs coexist across your users, shapes what your IA has to support: a single correct answer favors strong, precise search; exhaustive research favors flexible browsing and rich cross-linking between related content.
A single information space usually serves all four needs at once, for different users on different days, which is exactly why IA can't optimize for just one of them.
1: Known-item seeking. Maria wants a single correct answer to a specific question, comparable to Morville and Rosenfeld's example of asking which EU country has the largest population.
2: Exploratory seeking. Devon has an immediate need but no fixed target, any sufficiently good option would satisfy him, matching the definition of exploratory seeking.
3: Exhaustive research. Priya needs comprehensive coverage of a topic across many sources, not a single answer, the defining trait of exhaustive research.
4: Refinding. Tomas already found this exact item once. His task now is locating it again, which is precisely what refinding describes.
Each need favors different IA support. Known-item needs like Maria's are best served by precise search and clear metadata, she needs to type a query and get one right answer fast. Exploratory needs like Devon's are better served by curated categories, previews, and browsable groupings, since he's not typing a precise query at all. Exhaustive research like Priya's depends on comprehensive indexing, filters, and links between related content so nothing gets missed. Refinding, like Tomas's, is often best supported by browsing history, "recently viewed" features, or bookmarking, not search at all, since he may not remember the right search terms.The broader point: a single search box tuned only for known-item seeking will frustrate Priya and do nothing for Tomas. Good IA anticipates that all four needs are present on the same site, often within the same day.