AI-Powered Analysis

Reference

Understanding AI Confidence Scores in Text Analytics

Not every AI-classified response deserves the same level of trust. Confidence scores tell you which classifications are solid and which are worth a second look - here's what they mean and how to use them.

Reference

What Text Analytics Actually Analyzes: Your Full Response Set, Not a Sample

It's a fair question before you trust a category set built by AI: did it actually read everything, or just a sample? Here's what Text Analytics looks at when it generates categories and when it classifies.

Reference

Running Multiple Classification Passes on the Same Question

Generating categories a second time on a question doesn't erase your first attempt - it creates a new one alongside it. That's a genuinely useful safety property, and worth understanding before you regenerate.

Reference

How Many Responses Do You Need Before Category Generation Is Worth Running?

Generate categories too early, with only a handful of responses in, and you'll get a thin set that doesn't hold up as more come in. Here's how to think about timing.

Reference

Classifying a Response Into More Than One Category

Not every open-ended answer fits neatly into one theme. Text Analytics supports assigning a single response to more than one category when the answer genuinely touches on several - here's how it works and when to use it.

Reference

AI vs. Manual Coding: When to Use Text Analytics and When Not To

Text Analytics isn't the right tool for every open-ended question. Here's a practical way to decide when AI-assisted coding earns its keep and when reading the responses yourself is genuinely faster.

Reference

AI Follow-up Questions: What Triggers One and What Doesn't

A follow-up doesn't fire on every open-ended answer - it fires when a condition you set is met. Here's how to configure when AI Follow-up actually kicks in, and how to keep it from firing too often or not at all.