Can you trust the numbers in EvaluAgent?
EvaluAgent is quality assurance software for contact centres that scores and analyses every customer interaction, whether a person or an AI agent handled it. Its own help pages give 3 reasons its numbers can look different from what you expect. The first to rule out: A quality score can look thinner than the reviews your team has actually done, because, as EvaluAgent's help pages say, evaluations that are not yet published are left out of it.
What does EvaluAgent keep track of?
EvaluAgent keeps performance figures for each person who handles customer conversations, so an owner can see how each one is doing.
In EvaluAgent's own words:
The Insights report gives you performance metrics and analysis of individual agents.
How To Use The Insights Report, read 2026-11-12.EvaluAgent keeps quality figures such as pass rates and average scores, from the whole business down to one agent.
In EvaluAgent's own words:
The KPI report gives you detail on a range of data gathered within evaluagent.
How To Use The KPI Report, read 2026-11-12.
What can make EvaluAgent's numbers look wrong?
A quality score can look thinner than the reviews your team has actually done, because, as EvaluAgent's help pages say, evaluations that are not yet published are left out of it.
In EvaluAgent's own words:
Only published evaluations count toward scores
How To Use The CX Performance Report, read 2026-11-12.Two people can open the same report and see different totals, since, as EvaluAgent's help pages explain, each person only sees the agents and teams their role allows.
In EvaluAgent's own words:
The agents and teams you can view depend on your position within the hierarchy and your specific permissions.
How To Use The KPI Report, read 2026-11-12.A count of evaluations pulled through EvaluAgent's API can be lower than the count in its CSV download, because, as its developer docs say, deleted evaluations are left out by default while the download includes them.
In EvaluAgent's own words:
Deleted evaluations are excluded by default.
EvaluAgent API, read 2026-11-12.
How do I check EvaluAgent's numbers myself?
An owner can download the evaluations behind a score as a spreadsheet and count them line by line.
In EvaluAgent's own words:
The Exporter tool lets you export quality assurance data from evaluagent into CSV files for external analysis, reporting, or record-keeping.
How To Use The Data Exporter, read 2026-11-12.If a score changed and nobody can say why, an admin can look up who did what, and when, in EvaluAgent's audit log.
In EvaluAgent's own words:
The Audit Log in evaluagent lets administrators view and track specific user-triggered events across the platform.
How To Use The Audit Log, read 2026-11-12.
How does EvaluAgent work with Feder?
Connect EvaluAgent to Feder and it joins your Experience apps. Feder only reads. It never changes anything in EvaluAgent.
When you connect EvaluAgent, Feder checks the connection straight away and shows Connected only when that check passes. If a reading cannot be taken later, Feder shows it as not read, never as zero.
Other apps that feed Experience
Questions owners ask about EvaluAgent
- Why are cases missing from my EvaluAgent customer experience scores?Cases count by their case date, so a case dated outside the chosen range is left out even if it was evaluated recently.
- Why does an EvaluAgent Insights score differ from other reports?Insights groups results by quarter unless another interval is chosen, so it can differ from a report covering a week or a month.
- Why did my EvaluAgent KPI report change for the same period?Evaluations still under review can be counted or left out, so the same period can show different figures.
Where do these answers come from?
Everything this page says about EvaluAgent comes from EvaluAgent's own pages, quoted above with the date we read them. Last reviewed 2026-11-12. We read them again by 2027-11-12. If one changes, this page changes or comes down.
Know the day a number breaks, not the month after.
Some numbers are too quiet to judge daily, such as a form filled in a few times a week. Until there is enough of your own history to judge a change, Feder says it is still learning and scores nothing, rather than guessing.