Taxonomy
Manage the labels Vivollo uses to classify conversations — the built-in dimensions, your own custom ones, and the values it discovers.
Everything in Insights rests on classification — Vivollo labeling each conversation across a set of dimensions. Taxonomy is where you manage those labels: the built-in ones, the custom ones you add, and the new values Vivollo discovers as it reads.
You don't need to touch Taxonomy to get value from Insights — the built-in dimensions work out of the box. But as you grow, it's where you make the classification fit your business.
Dimensions and values
Two words to keep straight:
- A dimension is a way of labeling — Sentiment, Topic, Urgency.
- A value is a specific label within a dimension — positive, negative; or for Topic, billing, shipping, sizing.
Dimensions come in a couple of flavors. Some are fixed — they have a set list of values that doesn't change (Sentiment is always positive / neutral / negative). Others are open — Vivollo proposes new values as it discovers fresh themes (Topic grows with your business).
The built-in dimensions
Vivollo ships with a set of dimensions that cover the questions most teams care about, classified automatically from day one:
- Sentiment — how the customer felt: positive, neutral, or negative.
- Request type — what they wanted: a question, a request, a problem, or feedback.
- Urgency — how time-pressured they were: low, medium, or high.
- Customer effort — how much friction they hit getting helped: low, medium, high.
- Churn risk — signals they might leave: low, medium, high.
- Topic — what the conversation was about (open-ended, and multiple per chat).
- Knowledge gap — tagged when the agent couldn't answer well; this is what powers Knowledge Gaps.
These are always available and can't be deleted — they're the backbone of Insights.
Your own dimensions
Beyond the built-ins, you can add custom dimensions for whatever your business tracks — product line, store location, issue category. You define the dimension, choose whether each conversation gets one value or several, and whether the list of values is fixed or open to discovery. Vivollo then classifies against it just like the built-in ones. There are starting presets for common industries to save you setting up from scratch.
Reviewing what Vivollo discovers
For open dimensions like Topic, Vivollo proposes new values as it encounters new themes — and Taxonomy gives you a review queue to keep those tidy:
- Approve a proposed value to make it an official, established label.
- Reject a value that's noise or a one-off, and Vivollo won't suggest it again.
- Merge duplicates when two values mean the same thing ("shipping" and "delivery") into a single clean label — Vivollo even points out likely duplicates for you.
This light-touch curation is what keeps your topics meaningful over time, instead of fragmenting into dozens of near-identical labels.
Spend a few minutes in the review queue every so often, especially early on. Approving the good values and merging the duplicates quickly shapes Topic into a clean, trustworthy picture of what customers talk about — which makes every other Insights page sharper.
Correcting a label
Vivollo's classification is good, but not infallible. When you spot a conversation that's been labeled wrong, you can correct it — and that correction sticks. It's a simple way to keep your data honest and, over time, nudge the picture toward exactly how your team thinks about your conversations.
Why this powers everything else
Taxonomy is quiet, foundational work, but it's what makes the rest of Insights trustworthy. Clean dimensions and well-curated values mean the breakdowns in Explorer are meaningful, the Findings are real, and the Knowledge Gaps point at genuine needs. Tend the taxonomy, and everything built on top of it gets better.