Skip to main content
Velor

Platform · Insights

Measure what resolves. Find what doesn't.

The dashboard shows resolution rate, conversation volume, and action performance out of the box. Eval-driven testing lets you verify the agent before changes go live. Deeper analytics — conversation monitoring, trend analysis, experiment testing — expand as the platform matures.

Live in your dashboard.

Resolution rate

The primary KPI — percentage of conversations the agent resolved without a human. Tracked weekly with trend direction so you know if changes to your KB or actions are moving the needle.

Conversation volume

Daily and weekly conversation counts by channel. Deflection rate alongside human escalation volume — the ratio your CX team watches most.

Action performance

Per-action metrics: success rate over 7 days, p50/p95 latency, volume, outcome breakdown (success / user-cancelled / failed), and recent failure log with error details.

Eval harness

Upload a dataset of real conversations, run them against your current agent configuration, and see resolution rate on known inputs before you ship a KB or guardrail change.

What's coming.

Roadmap

Conversation monitors

Automated flagging of conversations that need human review — low confidence, repeated escalations, unexpected action failures. Surfaces what you'd miss in a daily spot-check.

Roadmap

Trend analysis

Resolution rate over time with a channel filter and a topic breakdown. Diagnose where the agent is regressing before customers notice.

Roadmap

Live experiments

A/B test the system-prompt addendum, tone, or the sentiment-escalation threshold against live traffic, with statistical significance indicators. Traffic is split per conversation. KB content and retrieval settings are not experiment types.

Why resolution rate, not deflection rate.

Deflection rate measures whether a bot handled the conversation — not whether the customer got their answer. You can deflect 80% of conversations and still have 80% of customers opening follow-up tickets.

Resolution rate measures outcome. The customer asked a question, they got the answer, and they didn't open a second ticket. That's the number that correlates with reduced support volume and improved CSAT — not whether a bot technically processed the request.

Velor's dashboard is built around resolution rate as the primary KPI because deflection without resolution isn't success. It's just moving the problem.

See what your agent is actually doing.

Every workspace starts with a 14-day free trial. Resolution rate, action metrics, and the eval harness are live from day one. No credit card required.