Turning Category Results Into Quotes and Examples for a Report¶
A chart showing that 34% of respondents mentioned onboarding friction is a finding. A specific respondent's own words describing exactly where they got stuck is what makes that finding land with a stakeholder. Since Text Analytics keeps every response's original verbatim attached to its category, pulling supporting quotes is always available - the question is doing it well.
Every Classified Response Keeps Its Original Text¶
Classifying a response never rewrites or discards what the respondent actually wrote (see How Text Analytics Works: From Open-Ends to Categories) - you can always trace a category count back to the exact responses behind it, which is what makes quote-pulling possible in the first place.
Finding Quotes Within a Category¶
Open a category's responses and read through them - Understanding AI Confidence Scores in Text Analytics is useful here too, since a high-confidence response is a clean, unambiguous example of that theme, which is usually what you want for a quote that's meant to represent the category clearly.
Avoiding Cherry-Picking¶
The point of a supporting quote is to represent the category, not to find the single most dramatic or articulate thing anyone said. A few practical habits help: read a genuine range of responses within the category before picking one, favor a quote that reflects the tone and substance of the category as a whole rather than an outlier, and if you're using more than one quote for the same category, choose ones that show different angles on the same theme rather than three versions of the same point.
Quoting Responsibly¶
Keep quotes close to verbatim - light edits for readability (removing filler, fixing an obvious typo) are reasonable, but changing the substance of what someone said isn't. If a survey's respondents were told their responses would be kept confidential or reported only in aggregate, check that quoting individual responses (even anonymized) fits what respondents were actually told before you publish one externally.
Bringing Quotes Into a Report¶
Reporting widgets are built around charts and numbers rather than long-form text, so a supporting quote usually lives in the narrative you build around the exported deck (see Exporting Your Report to PowerPoint) rather than inside a chart widget itself - export the category data, then add quotes directly into the PowerPoint slides once you're editing there.
FAQ¶
How many quotes should I include per category?
Enough to give a reader a real sense of the theme without turning the section into a wall of text - one or two well-chosen quotes usually does more than five mediocre ones.
Should I attribute a quote to any respondent detail, like their segment?
If you have that data and it adds useful context (and doing so doesn't compromise anonymity commitments made to respondents), a light attribution like "a respondent on the Enterprise plan" can add credibility without identifying anyone.
Can I search for a specific word or phrase across responses in a category?
Check your Text Analytics results view for search or filter options within a category's response list.
Does pulling quotes require going back to the raw export?
Not necessarily - you can read responses directly within a category in the Text Analytics results view, or work from an export if you prefer. See Exporting Classified Responses with Confidence Scores.
For the bigger picture of interpreting a category chart responsibly before you build the narrative around it, see Reading a Category Frequency Chart Without Overinterpreting It.