The 30-second answer

If your AI closed 100 tickets last month and 30 of those customers came back through email asking the same thing, your recontact rate is 30%. The 30 tickets that came back weren't really solved — they were dismissed.

The number tells you whether tickets were actually closed or just closed-looking.

Why most contact centres don't measure it

Channel silos. A customer who chats at 2pm and emails at 3pm shows up in two reports. The chat report counts a clean containment. The email report counts a brand-new ticket. Neither knows it's the same person, and the dashboard cannot connect them.

The underlying problem is identity stitching — matching the same customer across channels. Most contact-centre stacks weren't built for it. The phone system has a customer ID. The chat system has a different one. The email system has yet another. Without a single customer identity that travels across them, recontact is invisible.

How to measure it honestly

Pick a window — 24, 48 or 72 hours, the three standard choices. Then:

  • Take every ticket closed by AI in a period.
  • For each, check whether the same customer ID contacted you about the same topic through any channel inside the window.
  • The share where they did is your recontact rate.

The matching is the hard part. Same customer is easy if you have a unified profile (most CRMs help here). Same topic is harder — usually a combination of tagging, ticket-content similarity, or human spot-check on a sample.

What the numbers look like

Industry write-ups suggest 25-40% recontact within 72 hours is typical for AI-handled tickets. Maven AGI publishes this consistently; Customer Experience Decoded and Isara report similar ranges. That means: a reported 70% containment rate is more like a 42-49% real resolution rate, once you subtract the people who came back.

What "good" looks like

  • Under 15% — healthy. Most of what the AI closed actually stayed closed.
  • 15-25% — borderline. The AI is helping but the tail of unresolved tickets is large enough to matter.
  • 30%+ — the AI Volume Paradox in action: containment looks high, but tickets aren't really going down because they're coming back through a different door.

Where it appears in vendor dashboards

Rarely as a clean metric. Some platforms surface it under names like "reopened tickets" or "repeat contacts" — but only within one channel. The cross-channel version, which is the one that matters, is almost always something you have to assemble yourself from CRM + helpdesk + voice logs.

The fact that vendors don't surface it isn't a coincidence: a low recontact rate makes their containment number look great; a high one makes it look like noise. Measuring it yourself is how you find out which.