Key concepts and glossary
Plain-language definitions of every core Synthight term — finding, topic, sentiment, churn risk, and more.
A quick reference for the terms you'll see throughout Synthight.
Finding
A specific, evidenced problem or request pulled out of your conversations, always linked back to the tickets it came from. Findings are the main output of Synthight — see Understanding findings.
Topic
A cluster of conversations about the same underlying subject. Where a finding tells you what's wrong, a topic tells you what people are talking about and how that volume is changing over time. Every conversation gets exactly one topic, unlike findings, which a single conversation can produce several of. See Understanding topics and Topics vs. findings.
Sentiment
A per-conversation score capturing how positive or negative the tone of a conversation was, on a scale from -1 (very dissatisfied) to +1 (very satisfied). Rolled up as an average across a topic, person, or company, and used as a filter across Findings, Topics, People and Companies.
Customer traits
A set of AI-inferred characteristics about how a customer communicates and behaves — things like technical proficiency, communication style, patience level, and trust in support — shown as Low/Medium/High or a short descriptive label on the Topic and Person detail pages. Useful for tailoring how your team responds to different kinds of customers, separately from what they're asking about.
Churn risk
A computed Low/Medium/High risk level, used as a filter on the People and Companies pages, built from a weighted combination of signals in a customer's incident history — including sentiment, support outcomes, and engagement patterns. See How churn risk is calculated for more on how it's read.
Resolution status and resolution rate
Every incident (a single ticket tied to a finding) is tracked as solved, not solved, or unclear. Resolution rate is calculated as:
solved / (solved + not solved)— unclear incidents are excluded from the calculation entirely. This is the same definition used everywhere resolution rate appears: the Findings page's stat cards, People and Companies reports, and KPI grids across the product.
Priority
The urgency level assigned to a finding, used to filter and sort the Findings table.
Category
Findings are grouped into a fixed set of categories (Usability, Friction, Issues, Operations, Improvement Opportunity, Positive Feedback, Uncategorized) that describe the kind of problem, separately from its priority.
Incident
A single underlying ticket or conversation instance tied to a finding or topic — the raw evidence everything else is built from, always linkable back to the original ticket in your source tool via View Original Ticket.
Synthight only analyzes a conversation once it's closed or resolved in your source tool — not while it's still open. See How integrations work for why.
Person / Company profile
Customer records Synthight builds automatically from your conversations — no manual CRM-style data entry. See Tracking people and Tracking companies.
Report
An AI-generated document built from a set of findings, topics, people, companies, an interview, or a Simulation — with KPIs, charts and narrative you can view or export. See Report types explained.
Test
A hypothesis you write ("customers who churn cite pricing"), checked against the evidence Synthight already has from your conversations, with a confidence score.
Simulation
A structured set of questions run against a simulated panel of AI personas matching a target audience you define — a fast, directional read before you talk to real customers.
Interview
A structured, branching, AI-moderated conversation you share with real customers or prospects over a link, summarized automatically when they finish.
Frequently asked questions
What's the difference between a finding and a topic?
How is resolution rate calculated?
Are customer traits the same thing as churn risk?
What counts as an "incident" versus a "finding"?
Does Synthight analyze conversations while they're still open?
Was this article helpful?