See how your skills and knowledge are performingUpdated 8 hours ago

AI Agent gets better when you can see what's working and what isn't. Each skill and each piece of knowledge carries its own performance signals, so you're never guessing whether a change helped: you can watch how a specific skill or piece of knowledge is doing and act on what you find.

This is the close-up view of AI Agent's performance, one skill or one piece of knowledge at a time. For the bigger picture across all of AI Agent (automation rate, overall CSAT, ticket coverage), see review your AI Agent performance. For how performance fits into the broader practice of improving AI Agent, see how AI Agent improves over time.

In this article:

• What the metrics mean
• Review skill performance
• Review knowledge performance
• Drill into the tickets behind a metric
• Act on what you find

What the metrics mean

Skills and knowledge share the same core signals, measured per item over the last 28 days:

• Ticket volume — how many tickets used this skill or piece of knowledge. Shows how much of your workload it's handling.
• Handover — how often conversations using it were handed over to your team. A high handover rate signals that the skill or knowledge isn't fully resolving the conversations it's involved in. In a skill's performance view, the same idea also appears as a success rate: the share of conversations resolved without a handover.
• Average CSAT — the average satisfaction rating shoppers gave on conversations where the skill or piece of knowledge was used.
• Intent — the type of request Gorgias detected in conversations where the skill or piece of knowledge was used.

Since each skill and piece of knowledge has its own set of metrics, a rise or fall in any value can be directly tied to a recent update — so you can see whether your change had the effect you intended.

Review skill performance

Skill performance is available at two levels of detail: the Skills page for a high-level view across all your skills, and a per-skill performance panel when you want to go deeper on a specific skill.

Skills page overview

From the main menu, go to  AI Agent > Skills. Each skill's row shows its ticket volume, handovers, and average CSAT next to its name, linked intents, and status. Metrics cover the last 28 days.

Sort by any column to surface the skills that need attention. High ticket volume with a high handover rate is usually the clearest signal of where to focus.

Skill performance panel

Select a skill, then select  Performance. Summary cards show success rate (the share of conversations the skill resolved without a handover), tickets, handovers, and average CSAT, each compared to the previous period, alongside the most recent tickets the skill was used in.

Review knowledge performance

Your knowledge has its own performance signals — ticket volume, handover, and average CSAT — tracked one piece of content at a time, across all knowledge types: guidance, help center articles, website content, and documents.

1. Go to  AI Agent > Knowledge. Each piece of knowledge shows its ticket volume, handover tickets, and average CSAT, for the last 28 days. Sort by any column, or filter by intent to focus on a specific topic.
2. Open a piece of knowledge to see its full metrics in the side panel, including the intents it's used for and a preview of the most recent tickets it was used in.

These tell you which content AI Agent uses most and which topics generate the most handovers — the clearest signal that a piece of content may need updating.

Each metric reflects the last 28 days of data. For longer trends, use the AI Agent reports.

Open the ticket drill-down

On both the Skills page and your knowledge, click a metric to open the tickets behind it and review the conversations.

• Click a tickets value to see every ticket the skill or piece of knowledge was used in.
• Click a handovers value to see only the tickets that were handed over to your team. These are the conversations to read when you want to understand why AI Agent couldn't resolve something on its own.
• Click an average CSAT value to see the tickets that received a satisfaction score.

From the ticket list, open any ticket to review the full conversation in your inbox, or click Export to download the list. Two export options are available depending on how much detail you need.

Act on what you find

Per-item performance for skills and knowledge tells you where to focus your attention for improvement. A few patterns to look for:

• Prioritize the items AI Agent uses most. Sort by ticket volume to find them — skills and knowledge at the top of your list are what AI Agent relies on most, so improvements here have the widest effect.
• Investigate high handover. When a skill or piece of knowledge is involved in a lot of handovers, drill into those tickets to see what AI Agent couldn't handle, then close the gap.
• Fix low CSAT on high-volume items. When a frequently used skill or piece of knowledge has a low CSAT, read the related tickets to understand why — it might be a policy, an instruction, or knowledge that needs updating.

To act on what you find:

• For a skill, update its instructions, intents, or use of knowledge. See manage your skills and create a new skill for AI Agent.
• For knowledge, update what you have or add new knowledge. See add your brand's content for AI Agent and create guidance to give AI Agent knowledge about your business.

For the full improvement cycle, see how AI Agent improves over time.

Next steps

• How AI Agent improves over time — the cycle of training, testing, deploying, and analyzing
• Manage your skills — edit, enable or disable, and restore versions of your skills
• Review your AI Agent performance — the aggregate view across all of AI Agent