False Resolution: Definition, Causes and Detection

The mechanics are unglamorous. A customer asks about a delayed refund. The bot returns a policy paragraph that is accurate and useless. The customer closes the chat. No escalation button was pressed, no transfer was logged, so the session is stamped contained and the monthly report goes up.

Simplr's data, cited widely across the industry, put roughly one in five contained tickets in this category: drop-offs from an unhelpful bot rather than solved problems. Calabrio has catalogued the failure modes: abandonment without resolution, customers who escalate after a task was marked complete, intent misreads, and confident irrelevant answers.

Add a fifth that shows up constantly in Indian support operations: the customer who was told an email would follow, waited for it, and never received one. Nothing transferred, so nothing registered.

You can detect it without new software. Look at repeat contacts within 24 and 72 hours on the same customer ID and same intent. Look at sessions that end with no confirmation turn from the customer.

Look at short containments. Two or three turns is rarely a solved problem. Then sample fifty of them and read the transcripts.

The uncomfortable version of this: teams compensated on containment have no incentive to find false resolutions. Fix the incentive before you fix the bot.

Often confused with: A genuine no-need escalation, where the bot correctly answered a simple question and the customer left satisfied. The difference shows up in the repeat-contact data, not in the session log.