Newlin Phone System - Build checklist

0 of 12 done

Ticks are saved in this browser only. They are a working aid, not a record anyone else can see.

Where we start

Measured from call records the day before any change, so the same measurement can be run again afterwards and the difference means something. Nothing in Vonage had been altered when these were taken.

Measure	Before	After
Callers who reach a person	50.5%	not yet
Calls ending in voicemail	33.4%	not yet
Outbound showing a shared number	80.0%	not yet
Outbound showing the main menu line	40.7%	not yet
812-232-9599 calls reaching the team	0 of 98	not yet
812-232-3246 calls reaching the team	0 of 12	not yet

Re-run the same measurement after the changes and fill in the second column:

python phone_fix_snapshot.py --days 14 --compare output/phone_before.json

The seven changes

The two numbers that never reach the team

110 calls so far this month. Not one of them reached the team behind the number.

Repoint 812-232-9599 and 812-232-3246 from their call groups to a queue with a 20 second ring. Check what the CasePulse portal message and the outbound voicemail script tell clients to dial, team by team.

The teams are not the problem. Vonage's own month-to-date report shows Case Status answering 342 calls and Decision Support 231. Those calls arrive by transfer. What fails is the published number: of 110 calls that dialled these two DIDs directly between 1 and 18 August, every single one was missed or blocked, and not one ever reached the group extension. The number rings and dies before the team is offered the call.

The portal message and voicemail script matter for the same reason: they hand clients a number that cannot connect them, and that is already reaching clients in writing.

Verify: calls entering on 812-232-9599 start reaching extension 6526 at all. Today none do, so anything above zero proves the routing changed.

The front door's no-input setting

476 business-hours callers a fortnight reach the greeting and go no further.

Read the no-input action on Virtual Receptionist 400000 before changing anything. If it matches the others, change the action to forward to extension 5000 and widen the window to 5 seconds.

Confirm before quoting. Three other receptionists we inspected are each set to wait 2 seconds, play a message and disconnect. That screen has never been opened on 400000 itself. An earlier draft of this plan quoted the 2 second figure as though it had been.

Verify: the greeting's abandon count falls. One dropdown, and safe regardless of what the setting turns out to be.

Queues, not groups

This is the whole fix. A queue can be seen; a group cannot.

Create the Front Line Queue and staff it. Create or convert Case Status, Medical Records, Hearing and Decision Support to queues. Set ring to 20 seconds, not 5.

A queue has a Live View: callers waiting, agents logged in, average wait. A call group has none, which is why nobody could see that in 56% of unanswered operator calls both operators were on no call at all. The constraint is availability, not capacity.

Verify: the Live View exists and shows a number. That is as much the point as the answer rate is.

Voicemail that does not vanish

56% of voicemails currently get no callback.

Untick "delete voicemail after sending email" everywhere it appears.

A message deleted after emailing cannot be recovered when the email is missed.

Verify: messages remain in the box after the email goes out.

Direct extensions, then caller ID

The highest return per minute of work in the whole plan.

Repoint every direct extension's no-answer destination to its team queue. Then set outbound caller ID to that person's own direct line.

Order matters here. Giving a client a direct number that dead-ends on no-answer is worse than the shared number they get today. Repoint first, then change what we show them.

80.0% of outbound calls hand the client a shared number, and 40.7% hand them the main menu. This converts a callback into a direct connection instead of a fresh trip through the menu.

Verify: the shared-number share falls from 80.0%. This one moves the moment the config changes, so it is measurable the next day rather than in a week.

The new greeting

Last, not first.

Record it and assign keys 1, 2, 3 and 0.

Only after steps 1 to 5. Doing this earlier just routes people into broken groups faster. The destinations have to be real before the menu points at them.

Verify: walk every key yourself from an outside line.

The outbound voicemail script

135 "returning our call" cases a fortnight arrive not knowing who rang.

Change what it tells clients to dial, so returning a call reaches the team rather than the menu.

Verify: repeat-caller volume falls. Slowest of the seven to show, because it depends on clients hearing the new message first.

Afterwards

Re-run the measurement and compare. Anything that did not move is either not done or not the cause. Both are worth knowing, and the comparison tells them apart, which arguing about it cannot.