Running Multiple Classification Passes on the Same Question¶

If your first attempt at generating categories for a question didn't land well - too generic, missing an obvious theme, built before you'd collected enough responses to see the real pattern - the instinct is to worry that trying again means losing what you had. It doesn't.

Generating Categories Again Doesn't Overwrite the Old Set¶

Running AI Auto-Generate Categories a second time on the same question creates a new, independent category set rather than deleting or merging into the one you already have. Your original categories, and any classification work already done against them, stay exactly as they were - a fresh attempt sits alongside the old one rather than replacing it.

Why This Is Useful¶

This means regenerating is genuinely low-risk to experiment with. If you're not sure whether a first pass captured the real themes in your data - especially early on, when you might have generated categories before collecting many responses - you can try again later without worrying that you're throwing away earlier work. It also means you can compare two independently generated category sets side by side if you want a second opinion on how the responses break down.

Finding Your Category Sets¶

Since more than one category set can exist for the same question, check your Text Analytics results view for how existing sets are listed and how to select which one you're working with - the interface is built to let you see and manage more than one, rather than assuming there's only ever a single set per question.

What This Doesn't Do¶

This isn't a merge or refinement tool - regenerating doesn't take your prior categories into account or build on them; each generation run analyzes the responses fresh and proposes its own set. If what you actually want is to adjust an existing set rather than start over, editing it directly (see Editing and Refining Categories in Text Analytics) is the more targeted tool - renaming, merging, splitting, adding, or deleting categories within the set you already have, rather than generating a whole new one.

Choosing Between Editing and Regenerating¶

As a rule of thumb: if your existing categories are close but need adjustment - a rename, a split, a missing category - edit them directly. If you think the whole set missed the real structure in your data, or you've collected substantially more responses since you first generated categories, a fresh generation run is worth trying.

FAQ¶

Does classification data from an old category set get deleted when I generate a new one?
No - classification results are tied to the category set that produced them. Generating a new set doesn't touch classification work done against a previous one.

Can I delete an older category set I no longer want?
Check your Text Analytics results view for management options on existing category sets.

Does generating a second category set cost credits the same as the first?
Yes - category generation is a credit-based action each time it runs, regardless of whether it's the first or a later attempt on that question.

If I have more than one category set, which one feeds into my Reporting widgets and exports?
Check your results view for how an active or selected category set is indicated - this determines what data flows into Reporting and exports.

Next: see Editing and Refining Categories in Text Analytics if what you actually need is to adjust your existing categories rather than start fresh.