Customer Care Analytics · Wisdom notebook
Ask the follow-up.
Agentic conversational BI · from complaint growth to the channel it started in
Which complaint categories are growing fastest this month, and where is resolution failing to keep up?
RKRanked every category by volume growth over four weeks, then put first-call resolution beside it to find where the two diverge.
Growth and resolution do not line up — two categories are growing while getting harder to close:
Complaint volume growth this month, with first-call resolution
Plan change grew 9% and got easier to close — volume growth on its own says nothing about whether there is a problem.
For the billing one, is it the same charge everywhere or different by city?
RKSplit billing complaints by charge type and city, and checked which combinations carry the FCR collapse.
One charge type, two cities — and the resolution failure is confined to exactly those cells:
Roaming charges in Los Angeles and Philadelphia only. Chicago handles the same charge at 68%, so the answer exists — it just is not reaching two contact centres.
Which channel are these arriving through, and what did they try first?
RKSplit the billing contacts by entry channel and traced each back to the self-service attempt that preceded it, so a failed app or IVR journey is separated from a direct call.
71% reach an agent only after a self-service attempt failed — the app and the IVR both stop at the same missing line:
Billing contacts by entry channel · Los Angeles & Philadelphia
The app is where this starts, not the contact centre: 1,870 subscribers opened the bill detail and left without an answer, and most of them then called. The itemisation gap is a self-service defect that becomes a handling cost one screen later.