How Many Categories Is Too Many?¶

There's no universal right number of categories - it depends entirely on the question and the variety of what respondents actually said. But there are real signs your category set has drifted too far in one direction or the other, and both directions cause different problems in your final report.

Signs You Have Too Many Categories¶

Several categories only have a handful of responses each. If you're looking at a long list of categories and most of them barely register in the frequency count, they're likely fragments of what should be one broader theme rather than genuinely distinct ideas.

A reader can't hold the category list in their head. If summarizing your findings means walking through fifteen or twenty categories, the report has stopped communicating and started listing. A reader needs to be able to grasp the shape of the results, not just see every category that technically exists.

Categories overlap in ways that are hard to explain. If you'd struggle to explain to someone else why a specific response landed in one category rather than a very similar neighboring one, those two categories are probably splitting a single theme unnecessarily.

The fix: merge related categories (see Editing and Refining Categories in Text Analytics) until each one represents a genuinely distinct, sizable theme.

Signs You Have Too Few Categories¶

One category is doing too much work. If a single category holds a large share of your responses and, on closer read, actually contains several different underlying complaints or comments lumped together, it's too broad to be useful - the report will report one big number that hides real variation underneath it.

You keep wanting to add caveats when presenting a category. If explaining a category's meaning to someone else always comes with "well, it's really more like two different things" - that's a sign it should be split.

The fix: split the broad category into its real components (see Editing and Refining Categories in Text Analytics).

A Practical Target¶

There's no fixed number to aim for, but as a rough sanity check: if you can't summarize your category list in one sentence per category without repeating yourself, it may be too granular. If any single category needs more than a sentence to actually explain what it covers, it may be too broad.

Let the Research Question Guide You¶

A category set built for a quick internal pulse-check can reasonably be looser than one built for a client-facing deep-dive report. Match the granularity of your categories to how much detail the eventual reader actually needs, not to an abstract idea of thoroughness.

FAQ¶

Is there a hard limit on how many categories I can have?
No hard cap - the practical limit is how useful the category set stays as it grows, not a system restriction.

Should I aim for the same number of categories across every survey?
No - the right number depends entirely on how much genuine variety exists in that specific question's responses.

Does merging or splitting categories require reclassifying?
Yes - any change to your category set means re-running classification for the results to reflect the new categories. See Editing and Refining Categories in Text Analytics.

Can different questions in the same survey have very different numbers of categories?
Yes - category count is specific to each question's own responses, not consistent across a survey by design.

Next: see Editing and Refining Categories in Text Analytics for the actual tools to merge or split.