Skip to content
ai-agent.review — AI Agent Reviews, home
← All news

Product updates

Ada changes MCP CSAT filtering and adds separate AI and human scores

Vendor-reportedChecked: [release]

Editorial attribution: Gosia Madej

Prepared with AI assistance. Editorial attribution does not imply personal product testing or manual review of every story.

Published: · Event date:

Updated:

Ada’s October 19 release separates AI-agent and human-agent satisfaction metrics from the HANDOFF conversation filter. Assistants that used that filter to select whose score they wanted need a different metric type; unchanged requests can now mean something different.

Vendor-reportedChecked: [release]

What changes for users

The HANDOFF filter and score attribution answer different questions. Check a handoff nobody joined, an email survey and an Influx handoff when validating migrated queries. Keep null distinct from zero, align time zones and compare the same survey population before interpreting a change as improved service.

Editorial assessmentChecked: [release][reference][reports]

Two separate survey trays sit behind a conversation filter, with a small clock marking the date boundary.
AI-assisted editorial illustration of score attribution, filtering and time boundaries; not an Ada report or measured satisfaction data. · AI-assisted original vector illustration for AI Agent Reviews

Ada says csat_rate with HANDOFF now measures overall satisfaction within the matching conversations. Previously, that combination selected an AI or human score without filtering the conversations. The replacement metric types are csat_rate_ai and csat_rate_human; they return null when no survey matches.

Vendor-reportedChecked: [release]

The release also warns that MCP date ranges use UTC while dashboard reports use the viewer’s time zone. Surveys near a range boundary can therefore produce different totals.

Vendor-reportedChecked: [release]

Update, October 21: the MCP reference now documents csat_rate_ai and csat_rate_human. Our earlier source check found them absent from its metric table. The reference also now documents the HANDOFF note and UTC dates; its own publication date is not stated.

Editorial assessmentChecked: [reference][release]

According to the reference, a handoff without a human joining still counts toward AI CSAT, and email surveys always count toward AI. A colleague joining an AI-kept conversation does not count toward human CSAT. Influx is an exception: a handoff message alone assigns the survey to human support.

Vendor-reportedChecked: [reference]

Ada’s reporting guide defines overall CSAT as positive reviews divided by all reviewed conversations. That is a survey measure, rather than a percentage of every support contact or independently verified problem resolutions.

Editorial assessmentChecked: [reports]

Checked against sources

AI-assisted source check

How this was checked

Updated AI-assisted checks compared the October 19 announcement with the current MCP reference and reporting guide, plus the reference captured for our first report. The metric-table mismatch has been resolved. The sources remain vendor documentation; this does not independently verify returned scores.

Editorial assessmentChecked: [release][reference][reports]

No MCP connection, customer data or returned metric was tested. The announcement supports an attributed behavior change, not a measured improvement in satisfaction or proof that a particular account returns the new fields. No genuine source screenshot was obtained.

Editorial assessmentChecked: [release][reference]

Source evidence

Original source excerpt. This shows what the source says; it does not establish product performance.

The report reads dates in the time zone of the viewer.

Read original source [release]

Email surveys are always attributed to the AI Agent.

Read original source [reference]

Sources

Related agents