Complaints are up by a quarter and not one of them is a dropped call
Network complaints across the New York districts reached per thousand subscribers today. One sub-category carries all of it: data speed reached , while the dropped-call rate sits at . This is not a coverage story. The mechanism is underneath. Latency on the 5G layer has risen to and session success has fallen to , and those sessions are falling back to the older layer far more often than they should. On the 4G layer in the same districts, session success is exactly where it always is. The pre-churn score on the same cohort has already reached .
So what? Complaints arrive before disconnects, so this is the cheap end of a problem that gets expensive. Sending it to coverage planning would commit capital against something capital cannot fix. Meanwhile the subscribers affected are the ones paying a premium for the layer that is failing them.
3.7 vs 2.3 per thousand
the whole of the rise
92.2 vs 98.4%
sessions falling back to 4G
85 vs 48 milliseconds
throughput, not reach
0.73 vs 0.72%
rules coverage out
6.4 vs 5.0 per thousand
where the care volume is
58 vs 42
the churn side already answering
Network complaints by sub-category, against each one's own level
One sub-category carries the whole rise.
The line on each bar is that row’s own expected level. The scale runs from 0 to 4.
Network complaints reached per thousand and data speed reached , which is the whole of the rise. Every other sub-category sits on its own level, and the dropped-call rate sits at . This is not a coverage story.
The 5G layer and the 4G layer, in the same districts
One degraded. The other did not.
Data-Session Success · 5G
Data-Session Success · 4G
Latency · 5G, milliseconds
Latency on 5G has risen to and session success has fallen to , with those sessions are falling back to the older layer far more often than they should. On 4G in the same districts, session success is exactly where it always is. One layer degraded and the other did not, in the same places, for the same subscribers.
Complaint Rate and Pre-Churn Score by day
The complaints moved first. The score followed.
The pre-churn score on the same districts has reached , while complaints sit at . Complaint rate leads churn, which is the reason to read this today rather than after the disconnects.
92.2 vs 98.4%
what subscribers are paying for
58 vs 42
what it turns into
6.4 vs 5.0 per thousand
and the care cost meanwhile
It is throughput, not coverage
The dropped-call rate sits at while latency on 5G has risen to . This is not a coverage story.
Why it matters: Coverage is a capital programme. Throughput on a footprint that exists is a performance fix, and the two go to different teams.
One layer, and the subscribers paying for it
Session success on 5G has fallen to and those sessions are falling back to the older layer far more often than they should, while on 4G it is exactly where it always is.
Why it matters: The degraded layer is the one carrying a price premium, which is why the complaints are about speed rather than calls.
The churn side has already started
Complaints reached per thousand and the pre-churn score on the same districts reached .
Why it matters: Complaint rate leads churn. Reading it now is the difference between a network ticket and a retention programme.
Which cells inside the districts carry the latency?
Narrows a district to a cluster the field team can visit.
How many subscribers hit a fallback more than once a day?
Sizes who is actually experiencing it, not who is covered by it.
When did latency first leave its own band?
Gives the fix a start date to work back from.
Do the complaining subscribers overlap the high pre-churn ones?
Decides whether care and retention should work the same list.
Is 5G session success slipping in any other district?
Uses the earliest reading as an early warning.