They do two different things and work together:

A monitor without a scorecard will flag conversations but won't score them. For a complete QA workflow, you need both.

Yes, the following limits apply:

Yes, you can create as many monitors as you need, each targeting a different type of conversation or quality signal. For example, you might run one monitor for a random baseline sample, another for low CX score conversations, and another for conversations flagged as policy risks. Each monitor runs independently. Note that there is a limit of 20 monitors using natural language flag criteria — monitors using predefined filters are unlimited.

A monitor can be partially edited after it has been created. The following settings can be updated at any time from the Monitors page: the monitor name, flag criteria, filters, reviewers, and the attached scorecard. When you save your changes, all conversations matching the updated criteria are automatically re-evaluated — including previously flagged conversations that fall within the updated flag criteria, filters, scorecard, etc.

However, the monitoring mode (continuous, one-time, or scheduled), the start date, and sampling controls (e.g. number of conversations per day or week) are locked once a monitor is saved and set live. These settings control how the monitor runs and how backfills are created, so they cannot be changed after creation.

If you need to change any of these locked settings, the recommended approach is to create a new monitor with the desired configuration. You can use the Duplicate monitor option (available from the three-dot menu on the Monitors page) to copy an existing monitor and make quick adjustments.

Yes, in addition to evaluating Fin AI Agent conversations, Monitors also support human agent QA. You can create a monitor targeting human teammate conversations, attach a scorecard with your QA criteria, and have conversations automatically reviewed on a continuous, one-time, or scheduled basis.

Yes. You can archive a Monitor to hide it from the main Monitors page without deleting it. Archived monitors retain all their historical data and conversation counts. To archive, click the three-dot menu on any Monitor card and select Archive monitor. To view archived monitors, use the filter dropdown on the Monitors page and select Archived monitors. You can restore an archived monitor at any time — see below.

From the Monitors page, use the filter dropdown to switch to Archived monitors. Find the monitor you want to restore, click the three-dot menu on its card, and select Restore monitor. A confirmation dialog will appear. You can optionally check Backfill missed conversations from the last 2 weeks to catch up on conversations that occurred while the monitor was archived. Click Restore to reactivate it.

Yes, click the three-dot menu on any Monitor from the Monitors page and select Duplicate monitor. The same menu also gives you options to Edit, Archive, or Delete the Monitor.

Yes, click the three-dot menu on any Monitor from the Monitors page and select Duplicate monitor. The same menu also gives you options to Edit, Archive, or Delete the Monitor.

No, not at the moment.

No, Monitors only have access to the conversation transcript. They don't have access to the Knowledge base, Guidance, Tasks, or Procedures. For example, a Monitor cannot check whether Fin had access to a specific article, or whether a conversation should have been passed to a human based on a guidance rule.

No, audit history for Monitor edits is not currently available in the UI.

No, the only available date filters are last 7 days, last 8 weeks, and last 6 months.

No, not at the moment.

No, human agent QA Monitors are not available on Fin Standalone.

A conversation must include at least 2 messages from Fin and 2 messages from the customer to be matched into a monitor.

Yes. Each monitor runs as an independent yes/no check, so if a conversation matches the criteria of more than one monitor, it will be flagged in each. Clicking through to a conversation from any monitor shows exactly why it was flagged by that specific monitor.

No. A conversation is evaluated only once per monitor. If the conversation later reopens because the customer sends a new message, it will not be re-matched or re-evaluated under the same monitor version. The original evaluation is the only one recorded.

Yes. When first creating a monitor, you can backfill up to 90 days of historical conversations. From that point, the monitor continues capturing new matching conversations automatically.

No. Fin Voice is not currently supported by Monitors.

No. Monitors evaluate chat and messaging conversations only — phone call conversations aren't currently supported, whether handled by human teammates or Fin Voice.

No. Neither customer tickets nor tracker tickets are matched into monitors — only conversations.

No. Conversations that matched the previous flag criteria stay in the Monitor. Only conversations going forward are matched against the updated configuration.

Conversations should appear instantly. However, the conversation count on the Monitor dashboard can take a few minutes to update when there is a high volume of conversations.

No. It's not currently possible to manually remove a conversation from a Monitor.

No. The Run test option only appears when flag criteria have been configured.

Go to the Monitor, click Edit, and open the Run test section. Under Test flag criteria on sample conversations, enter the conversation ID and click Add. The Monitor result will explain why the conversation matched or didn't match.

If a conversation isn't showing up, check these common causes:

No, but without one your monitor will flag conversations without scoring them. Reviewers will see matched conversations with no criteria to evaluate. For a complete QA workflow, attach a scorecard during monitor setup.

Yes. Each monitor can have its own scorecard, so you can evaluate different types of conversations against different quality criteria. For example, a monitor for vulnerable customers might use a different scorecard than a monitor for billing conversations.

Yes — criteria titles and descriptions are reusable. Once you have created criteria, you can add it to multiple scorecards. However, previous rating scores cannot be reused and will need to be set from scratch in each scorecard.

Yes. You can choose on a per-criteria basis whether AI or a human reviewer handles scoring. If Auto-review is enabled and any criteria requires manual scoring, those conversations will still appear in the Unreviewed queue.

If critical criteria receives a failing rating, the overall review score drops to 0% — regardless of how other criteria scored. This overrides all weights and the pass threshold. Not scored ratings do not trigger failure. Critical criteria are useful for non-negotiable standards like compliance requirements, safety or policy adherence, and escalation handling.

Weights are relative to each other, 0 to a scale of 100. The total can add up to any number — what matters is the proportion each criteria contributes. For example, criteria with a weight of 25 out of a total of 50 contributes the same as one weighted at 50 out of 100. Weights only apply to criteria that are included in the review score.

Yes. Teammates can manually override any AI score if they spot a discrepancy. Simply click the rating in the conversation's Score tab to change it.

No. Only the name and description carry over. Weightings always need to be configured from scratch.

The criteria is excluded from weighting entirely. The overall score is calculated based on the remaining scored criteria only.

Scores are visible in the Monitor conversation list (showing overall score and individual criteria ratings as columns) and within each individual conversation under the Score tab.

No, not at the moment.

No. Audit history for Scorecard edits is not currently available.

Auto-review is a feature you can enable on a scorecard that automates the entire QA process. When turned on, if AI scores all criteria in a scorecard, the manual review step is skipped entirely. Conversations are marked as reviewed automatically, and only failures or edge cases require human attention.

The reviewer status will show as Auto-reviewed. These conversations bypass the manual Unreviewed queue unless the AI detects a failure or cannot confidently score an attribute.

No. AI doesn't learn from human overrides at the moment.

Yes — and we strongly recommend it. For Monitors using natural language flag criteria, use the Test monitor tool to validate your criteria against real conversations before going live. It shows which conversations would be flagged and highlights mismatches so you can refine and reduce false positives.

No. The Run test option only appears when flag criteria have been configured.

Go to the Monitor, click Edit, and open the Run test section. Under Test flag criteria on sample conversations, enter the conversation ID and click Add. The Monitor result will explain why the conversation matched or didn't match.

No, not at the moment.

We've replaced the conversation list view you see when clicking into a monitor. The new experience is built on a different foundation that's more reliable, more stable, and faster. While it doesn't yet include all the functionality from the previous view, the missing features are temporary and will be reintroduced over the coming weeks.

The following features are temporarily unavailable and will be reintroduced over the coming weeks:

There are two review states:

Go to the Unreviewed conversations queue on the Monitors page. From there, filter by your own name to see only conversations assigned to you across all monitors.

Yes. To avoid disrupting existing workflows, you can still filter conversations using legacy statuses, so previously classified conversations remain accessible. However, new conversations can no longer be assigned to those legacy states — only Unreviewed and Submitted are available going forward.

These statuses were part of an earlier version of Monitors and have been replaced with a simplified two-state system: Unreviewed and Submitted.

You can still filter by the old statuses to find previously classified conversations, but new conversations can only be set to Unreviewed or Submitted. Conversations submitted with a failing score are automatically flagged for follow-up.

The change is intentional — it's the foundation for a more flexible system. We’ll be introducing a new workflow built around Back-office Tickets, where you’ll be able to define and manage your own custom statuses.

Yes. The simplification to Unreviewed and Submitted is intentional and won't be reverted, but it's the foundation for a more flexible system. We'll be introducing a new workflow built around Back-office Tickets, where you'll be able to define and manage your own custom statuses.

Once all scorecard criteria are scored, set the review status to Submitted to mark the review as complete. Conversations submitted with a failing score will be flagged for follow-up. You can still add a note before submitting — for example, to flag a content update needed.

When a review is submitted with a failing score, the conversation is automatically flagged for follow-up. The right next step depends on why it failed:

Note that submitting a review doesn’t directly change how Fin responds — changes to content, guidance, or procedures are what improve Fin’s future performance.

No. Tagging teammates in score notes is not currently supported.

Two permissions are required, both enabled by a workspace admin:

Any teammate with both the Can view Fin and Automation settings and Can create, edit and internally share Reports permissions can create, edit, and delete monitors and scorecards. Teammates with only the first permission can review conversations and adjust criteria values, but cannot create or delete monitors.