Overview
A compliance rule describes a sample phrase that agents are expected to say during a call — for example, a regulatory disclosure, a consent statement, or a mandatory opening line. For each call, the platform compares the rule's text against the transcript and produces a script adherence score between 0% and 100% per rule. Compliance scoring runs against call transcripts only — chat, email, and other non-call sessions are skipped by the matcher. Very short call transcripts are also skipped.
This article covers the Compliance Rules configuration page, where you create and manage rules. The scores those rules produce show up later inside individual conversations and in the Script adherence filter on the Conversations page.
How It Works
Accessing Compliance Rules
Open Analytics configuration from the main navigation, then select Compliance Rules. The page lists every compliance rule in the workspace.
Who can use Compliance Rules
The Compliance Rules menu entry appears only for users with the analytics-management permission. Those users can create, edit, and delete rules. The same permission is also enforced on every create, update, and delete request on the backend.
Users without that permission do not see the Compliance Rules entry in the navigation, and the backend rejects their list, create, edit, and delete requests for compliance rules. (Opening the page through a direct link still renders the empty form, but no rules can be loaded or saved.) They also do not see the shield menu or matched-segment highlights inside a call — those views render only when the current user can load the rule list. The Script adherence filter on the Conversations page is populated separately and remains available to them as a way to filter conversations.
Creating a rule
Click + Create Rule in the top right. A new editable row appears at the top of the list.
Fill in the row:
Name — short, descriptive name for the rule (for example, GDPR consent). Required.
Threshold — the default minimum match percentage, at least 10% and below 100%, prefilled for this rule in the Script adherence filter on the Conversations page. Defaults to 50%. It is not used by the matcher to decide whether to save a score, and changing it does not change which calls receive shield-menu matches.
Text — the consent or script text the platform should look for. This is the phrase representatives are expected to say. Required.
Save the row to create the rule.
After saving, the rule is added to the list and the platform starts scoring it against conversations.
How the score is computed
For each call, the platform compares the rule's Text against the transcript and produces a score between 0% and 100% per rule. The match is text-level: the rule's text is aligned against the transcript character by character, so missing letters, substituted words, rephrased fragments, and changed punctuation or spacing all lower the score, while extra filler words spoken around the rule's text do not. A score of 100% means the rule's text was found in the transcript essentially as written, but the agent may also have said additional words alongside it — it is not a guarantee of a verbatim recital.
The per-rule Threshold does not affect this score. It is only used as the default value prefilled for the rule in the Script adherence filter on the Conversations page.
Editing a rule
Click the pencil (edit) action in the rule's row to open it for editing. You can change the Name, Threshold, and Text of the rule. Save the row to persist your changes. Saving an edited rule updates the rule itself; previously computed scores on existing conversations are not refreshed automatically. New conversations are scored against the updated rule as they are ingested. To refresh scores for historical conversations against the current rules, an administrator must trigger a re-match explicitly.
Deleting rules
You can delete rules individually from the row actions, or in bulk by selecting multiple rows with the checkboxes and clicking the Delete action that appears in the toolbar.
Deletion is destructive: all compliance scores for the deleted rule are also removed and cannot be recovered. The confirmation dialog calls this out explicitly. If you only want to stop using a rule temporarily, edit it instead — for example, by changing its text — rather than deleting it.
Where compliance scores appear
Once a rule exists, its results show up in a few places across the product:
Inside a call. Open any call and look at the top of the session card. A shield button (Show script adherence matches) lists every rule that produced a match for the call, with its score as a percentage chip. Click a rule in that menu to highlight the matching segments of the transcript; pick Hide matches to clear the highlight. Compliance scores are produced for calls only — chat, email, and other non-call sessions are skipped by the matcher and will not show a shield menu or transcript highlights.
In transcripts. When a rule is selected from the shield menu, the words that matched the rule's text are highlighted inside the transcript so you can verify what the platform found.
In the Conversations filter bar. A Script adherence filter (shield icon) lets you find conversations where one or more selected rules scored above, or below, a chosen percentage. You can combine rules with Any or All logic and optionally set a minimum overall threshold. The filter is only visible when at least one rule exists.
Important notes
Thresholds must be at least 10% and below 100%. Values outside that range cannot be saved.
Creating or editing a rule does not automatically re-score historical conversations. Scores for new conversations are produced against the current rules as they are ingested; to apply the current rule list to past conversations, an administrator must trigger a re-match explicitly.
Deleting a rule deletes its scores. If a rule may be needed later, prefer leaving it in place — there is no archive option for compliance rules.
The Script adherence filter only appears in the filter bar once at least one rule has been created.