DeskCrew

Signals: turn app errors into proactive support

Most customers who hit a bug never report it. They just leave. Signals turns your app's errors into support tickets, so you can reach out before the complaint.

What Signals does

  1. Capture: errors arrive through two doors:
    • Widget door: the DeskCrew widget on your site catches JavaScript errors on the page automatically.
    • API door: your backend can POST server errors to /api/v1/errors (Sentry-style payloads work) using an API key with the errors:write scope.
  2. Dedupe: identical errors are fingerprinted into one error group. Five hundred users hitting the same bug show as one signal with an affected-user count, not five hundred rows.
  3. Ticket: when an identified user hits an error, DeskCrew opens a ticket for that specific person. One ticket per person per error group. Nobody gets double-contacted for the same bug.
  4. Proactive draft: with AI drafts enabled, DeskCrew prepares a reach-out reply grounded on your knowledge base ("We noticed checkout hit a snag. Here's a workaround…"). The draft lands in your approval queue and never sends itself. A human always approves first.

What "identified" means

Tickets are only opened for users your site has identified through the widget's identify() call, which uses a signed token from your server. That signature is why an attacker can't spawn tickets pretending to be someone else. And why anonymous visitors are counted in the error group but don't get individual tickets.

Turn it on

Signals is off by default for every workspace.

  1. Go to Signals in your dashboard sidebar.
  2. Click Enable Signals (admins only).
  3. That's it for the widget door. Capture starts on your next page loads. For server errors, mint an API key with errors:write under API Keys and POST to /api/v1/errors.

Keeping the noise down

Three controls, all on the Signals page:

  • Ignore list: mute error patterns you don't care about (third-party scripts, known noise).
  • Daily cap: a hard ceiling on captured events per day.
  • Sample rate: capture a percentage of high-volume errors instead of every occurrence.

What it looks like day to day

Your Signals board shows each error group with its title, path, first/last seen, and how many users it affected. Groups link to the tickets they opened. Approve the prepared reply on a ticket and the customer hears from you while they're still making coffee. Support that arrives before the complaint.

Related articles