For setup instructions and an overview of how Signals work, see Setting Up Signals.
My Signals look stuck or haven't updated
The Detected tab only shows Signals that match an Account already imported into Cora. If Cora has been analyzing new activity but the companies involved aren't in your Accounts list, those Signals won't appear.
To fix this:
Broaden your CRM import filter (e.g., include Opportunity or SQL stages, not just Customer) so more companies become Accounts in Cora
Or manually add specific Accounts for companies you want tracked
Can I add an event myself?
Yes. On the Signals page, click Create event to record a lifecycle event manually, or just ask Cora in chat (for example, "log a go-live for Acme on Friday"). This is handy for moments that happened offline or outside your connected tools.
Manually added events always appear on the account's timeline regardless of CRM source matching, and are labeled Added in Cora so you can tell them apart from automatically detected events. By default they are recorded on the timeline but do not start agent workflows. For details, see Adding an Event Manually in Setting Up Signals.
New activity just happened but no Signal appeared yet
Data from connected sources (Gong, Gmail, Slack, etc.) can take anywhere from a few minutes to a few hours to become available, depending on the integration. Cora's detection then runs on a regular cycle after that. In most cases, expect new Signals to appear within a few hours of the underlying activity.
Activity with a known customer didn't generate a Signal
Cora matches activity to Accounts by checking whether the participants (call attendees, email recipients, Slack users, etc.) are linked to a company record in your CRM. If the contacts involved aren't associated with the correct company in HubSpot or Salesforce, the activity won't match — even if the company itself is already an Account in Cora.
To fix this, associate the relevant contacts with the company record in your CRM. Future activity involving those participants will then match automatically.
Why does Cora use contact-based matching instead of domain matching?
We've seen many CRM setups where company domains are incorrect, or participants use personal or mismatched email addresses. Matching via contact-to-company associations in your CRM is the most reliable way to attribute activity to the right customer.
An old record generated a Signal after a minor update (example: Closed Won deal in CRM)
Unrelated field edits can update last modified. If your rule only checks current state (e.g. stage), the record can match again even though the event happened long ago.
Add a temporal filter on the business event date rather than last modified or state alone.
Lots of Signals when we first turned on rule-based detection
This usually means detection conditions match historical records with no date window — for example, in a CRM, all deals currently in Closed Won. State-style conditions (e.g. stage or status) match the record's current value, not when it changed.
Add a date compare on the date that represents when the event happened, and combine it with your other conditions using ALL — for example, in CRM, close date within the last 14 days plus stage. See Detection Conditions in Setting Up Signals for how to configure this. Rule changes apply going forward; they do not remove Signals already created.
Will a date window (e.g. 14 days) retrigger my agent every day?
No. A record may keep matching while it's inside your date window, but Cora deduplicates Signals so the same event is only detected once. You'll only see another Signal when something is genuinely new — for example, a different record or event type. See Detection Conditions in Setting Up Signals for temporal filters and deduplication.
How far back does Cora look when detecting Signals?
The detection pipeline runs every hour, processing new activity from your connected integrations each cycle. How far back data is available to analyze depends on the source:
Gmail — 90 days
Gong — 30 days
Slack — 7 days
Intercom — 30 days
After the initial sync, Cora continuously analyzes new activity as it arrives on the hourly cycle. When classifying an event, the AI also considers up to 90 days of contact engagement history and 48 hours of cross-source context, so it can spot trends even if the triggering activity just happened.