Fibi AI Agent reporting
Measure Fibi AI Agent deflection, resolution, escalations, handoffs, and customer satisfaction with the built-in Fibi AI Agent report template.
Written By Markus from Featurebase
Last updated About 8 hours ago
Overview
The ‘Fibi AI Agent’ template shows how much of your support Fibi handles on its own and how well it does it:
How often Fibi gets involved
How many conversations it deflects and resolves
Where it hands off to teammates
How customers rate it
Go to the Reports tab and create a new report. Select the ‘Fibi AI Agent’ template to get started quickly.
Note: You can also combine reporting metrics and attributes to create your own custom report.
Headline metrics
The top of the report answers the three questions that matter most:
‘Fibi AI Agent deflection rate’ - the percentage of Fibi-handled conversations that ended without human involvement
‘Fibi AI Agent resolution rate’ - the percentage of Fibi-involved conversations that reached the resolved state
‘Fibi AI Agent CSAT score’ - the percentage of positive ratings out of all Fibi ratings
‘Fibi AI Agent's impact over time’ trends all three on one line chart, so you can see whether changes to Fibi's content and configuration actually move the numbers.
Tip: Hover the info icon next to any chart title to check its metric definition, Timestamp, Aggregation, and applied filters.
Fibi AI Agent involvement
This section measures Fibi's reach. The ‘Fibi AI Agent involvement rate’ KPI shows the share of reportable conversations where Fibi participated, and ‘Fibi AI Agent involved conversations over time’ tracks that volume across the date range.
Two donut charts split the same picture by channel: ‘Conversations where Fibi AI Agent was involved - by channel’ and ‘Conversations where Fibi AI Agent was not involved - by channel’. Use them to spot channels where conversations bypass Fibi entirely.
Fibi AI Agent resolutions
This section breaks down what happens inside Fibi's conversations. ‘Fibi AI Agent resolution rate’ and ‘Fibi AI Agent resolved conversations’ show the share and count of Fibi-involved conversations Fibi resolved, and ‘Fibi AI Agent resolved conversations over time’ trends the count.
Four KPIs then show how Fibi-involved conversations progress:
‘Fibi AI Agent answer rate’ - the share of Fibi-involved conversations where Fibi generated at least one answer
‘Fibi AI Agent deflection rate’ - the share that ended without human involvement
‘Fibi AI Agent escalation rate’ - the share escalated to teammates
‘Fibi AI Agent pending rate’ - the share still waiting on a final resolution state
Resolution and handoff breakdowns
Resolutions split by how Fibi answered: AI Answers Fibi generated from your knowledge sources and Custom Answers you configured, each with its own resolution count and rate. Fibi handoff reason separates handoffs caused by a usage cap, a teammate handoff, or an error. Fibi handoff trigger shows the concrete condition that caused the handoff, such as a customer request, your configuration, default behavior, a Procedure, a usage cap, or an error. Both handoff attributes are available in filters and breakdowns, and their values appear in drill-ins and exports.
Content performance
The ‘Fibi AI Agent content performance’ table ranks the knowledge sources Fibi used in its answers by Fibi AI Agent involved conversations and Fibi AI Agent resolved conversations.
Note: Fibi can use more than one source in a conversation, so the same conversation can count under several sources and the rows can add up to more than the summary total.
Assumed and confirmed resolutions
Two KPIs qualify the resolution number. ‘Fibi AI Agent assumed resolution rate’ shows the percentage of Fibi conversations that became assumed resolved, while ‘Fibi AI Agent confirmed resolution rate’ shows the percentage with explicit customer-confirmed resolution. Together they separate inferred resolutions from customer-confirmed resolutions.
Fibi AI Agent customer satisfaction
The last section covers how customers rate Fibi. For a view across Fibi and your whole team, use Customer Satisfaction reporting. A rating counts toward Fibi when its Rated agent type is Fibi AI Agent, meaning the customer rated Fibi's handling rather than a teammate's.
‘Fibi AI Agent CSAT score’ - the percentage of positive ratings out of all Fibi ratings
‘Fibi AI Agent CSAT survey responses’ - how many ratings came in
‘Fibi AI Agent conversation ratings’ - a donut splitting responses by rating label, from amazing to terrible
‘Fibi AI Agent conversation ratings - by time’ - rating volume over the date range
Customize this report
Click Use this template to turn the template into your own report: add, remove, or reconfigure any chart, change the date range, and filter the whole report, for example by Channel or Team. For your own charts, attributes like Fibi resolution state, Fibi answer type, Fibi handoff reason, Fibi handoff trigger, and Fibi answer source title let you slice Fibi's behavior further - every definition is in Reporting metrics and attributes.
Learn more about customizing reports →
Fibi AI Agent report use cases
Find where Fibi is bypassed
Compare the involved and not-involved channel charts to find channels where conversations regularly bypass Fibi.
Diagnose escalations and handoffs
Use the outcome rates with Fibi handoff reason and Fibi handoff trigger to separate customer requests from configuration, usage-cap, and error-driven handoffs.
Improve Fibi's content
Sort ‘Fibi AI Agent content performance’ by involved conversations to find the sources Fibi relies on most, then compare their resolved conversations:
Keep Fibi's most-used content accurate and up to date as a priority
Review sources with many involved conversations but relatively few resolved conversations
Check underperforming content for missing, inaccurate, or ambiguous information
Review customer satisfaction
Use ‘Fibi AI Agent CSAT score’ together with its survey-response count to judge satisfaction without overlooking the sample size.