ewd unison

Management Information & Solutions
MASTERCLASS
EWD
Fun facts to know and tell about the EWD
Brought to you by Windex! Windex Makes It Shine
Master Class EWD
1.
Introduction to the dialler and EWD
2.
Dialler Metrics & You
3.
Sources of Data Part 1: EWD/Unison
4.
Sources of Data Part 2: Xcalibur/Nice
Dialler Metrics & You
1.
Previously in MASTERCLASS
2.
The Collections Pyramid
3.
Metrics UNISON told me
4.
5.
6.
a.
Where the metrics happen
b.
Preview vs Predictive
Metrics EWD told me
a.
How it gets the data back
b.
Multiple actions per event
EWD or UNISON?
a.
What’s the difference?
b.
Why we did it
Key Metric Overview
Previously in
MASTERCLASS
Master Class EWD – Previously in MASTERCLASS
Host
System
Probe
Download
EWD
Buranda
Dialler
Contact
Lists
Burnley
Dialler
Host
System
Host
System
XCalibur
XCalibur
NICE
MQS
NICE
MQS
The Collections Pyramid
Master Class EWD – The Collections Pyramid
Download
Scheduled
Attempt
EWD/
Connect
Dialler
RPC
PTP
Kept%
Navigator
Metrics UNISON Told Me
Master Class EWD – Metrics UNISON Told Me
We established last week that UNISON was once an independent dialling tool, all
reporting prior to EWD was sourced directly from the UNISON dialler.
The introduction of EWD changed the way we worked but it also had an impact on
how we report out information.
Last week I said its easier to understand EWD by exception. We’ll approach the
metrics in the same way.
UNISON is the tool doing all the dialling, its what the agents log into, so all my data
should be there, true, with a few exceptions.
So lets look at UNISON
Master Class EWD – Where the Metrics Happen
Scheduled
When looking at the metrics we are going to cover off the raw metrics and where they can occur at
each stage. We’ll leave the calculated metrics to a little late on. For this section we are going to
take a look at the following:
Attempt
Connect
Attempt: Count of the number of dials made
Abort: Count of the number of times we have hung up on the customer
Abandon: Count of the number of times the customer has hung up on us
Connect: Count of the number of accounts that have gone to an agent
Worked: Unique count of accounts that have been sent to an agent
Account Called: Unique count of the accounts that have been actioned
System
Dispositi
on.
Account sent back for next action
Account
Attempt = 1
Connect = 1
Account Called = 1
Worked = 1
Accoun
t gets
dialled
Does
the
phone
answer?
Yes
Customer hangs up
Account sent back for next action
Call
abandon
ed
Put call
on hold
Dialler hangs Up
Abandon = 1
Yes
Accoun
t Sent
to agent
No
Abort = 1
Is there
an
agent
availabl
e?
Agent becomes
available
Master Class EWD – Where the Metrics Happen
RPC
PTP
The account is now with the agent. So we are now dealing with the following:
RPC: The count of agent dispositions which are flagged as RPCs (Determined through the dialler and reporting)
PTP: The count of the agent dispositions which are flagged as PTPs (Determined through the dialler and reporting)
Talk: The total time (in seconds) that the agent spent talking
Wrap: The total time (in seconds) that the agent spent in wrap
Available: The total time (in seconds) that the agent spent waiting between calls
Agent
Dispositio
n
Account sent back for next
action
Yes
Yes
Account
No
No
Are we
speakin
g to the
Cust
Did we
get a
PTP
PTP
entered
RPC = 1
Available
Talk
PTP = 1
Wrap
Master Class EWD – Preview vs Predictive
As discussed last week the dialler can be run in two states. Either Preview or Predictive. The metrics we have covered off in the previous
few slides are what occurs in a predictive environment. When the dialler is doing all the dialling. In preview the dialler doesn’t do the
dialling it simply sends the accounts through the agents. So lets see how this changes where the metrics now occur.
Account sent back for
next action
Don’t
send
account
Allocate Time: Time spent (in seconds) once the agent has
the account before they make an attempt
No
Account
New Metric:
Agent
Dispositio
n
Is the
account
eligible
?
Yes
Agent
decides to
call
account
Account sent
to agent
Yes
Agent
decides to
call
account
Are we
speakin
g to the
Cust
Yes
Account Called = 1
Connect = 1
Worked = 1
PTP
Entered
PTP = 1
Available Time
Allocate Time
Yes
RPC = 2
Talk
Did we
get a
PTP?
Wrap
No
No
No
Attempt =1
Master Class EWD – Preview vs Predictive
Considering the slight differences in how preview and predictive capture their information. Can we explain the
following:
1. Can Accounts Called be higher than Attempts?
2. Explain how Accounts Called could possibly be higher than Worked
3. Would I expect to see my Connects higher than accounts called or Worked?
4.What are the key signs I might look for to tell which state a contact list was worked in?
Metrics EWD Told Me
Master Class EWD – How EWD gets the data back
Remember how in last weeks masterclass we said that the UNISON dialler returns a disposition whether it be agent or system
back to EWD. This enables EWD to evaluate its next course of action. More than the disposition it sends back additional
information that we can then use for reporting. Through this information we can identify the following by account and
contact list etc:
Dial Start Time = Wether the account was attempted or not (Attempts)
Connected Time = Wether the account was sent to an agent (Connect)
Response Status = This field holds the individual disposition codes. We can simple count the instances of certain
dispositions to get the following:
Abort = CHU (Customer hung Up)
Abandon = SHU (System Hung Up)
RPC = NPTP (or any other dispositions that we have determined are to be counted as RPC)
PTP = PTP (or any other dispositions that we have determined are to be counted as RPC)
As we get the account level information we can run a quick query over this data to get the following information
Accounts Called (either Connected Time or Attempt time)
Worked (Connected time)
You’ll notice that the metrics I haven’t covered off above are my agent time pieces (Talk, Wrap, Allocate and Idle) this
information only exists in UNISON and so we source it from there.
Whilst we can see all the account level information through EWD and can do a great deal with it, there is actually a shortfall
in the data that we can not measure through EWD
Master Class EWD – Multiple actions per event
One record per event………anyone see a problem if I have multiple attempts in a single event?
An agent wether in preview or predictive can make multiple calls. Unison can see all dialling attempts made,
wether dials it has done or dials by the agent. But EWD only gets one record back.
Take a closer look at what I mean.
1.
Agent working in predictive has the account presented to them when the home number was dialled
(attempt 1). Agent is advised to try and contact the customer on the mobile. Agent then dials the mobile
(attempt 2). Agent speaks to the customer, gets a promise and sends the account back to the dialler.
What UNISON saw is 2 attempts
What UNISON sent back to EWD is 1 attempt
2.
Agent working in preview attempts to call the customer on the business number (attempt 1), but gets no
answer. Agent attempts to dial the home number (attempt 2), but gets no answer. Agent then tries to dial
the mobile number (attempt 3), but gets no answer.
What UNISON saw is 3 attempts
What UNISON sent back to EWD is 1 attempt
As there is only one record per event, there is only 1 field that can record the dial time. So EWD can not
measure and we can therefore not record all of the dials that actually took place through EWD.
So that being the case, why do we use EWD to source our data? We can’t see everything.
EWD or UNISON
Master Class EWD – What’s the Difference
Scheduled
Even though all the action takes place in the dialler we actually source a majority of our data from EWD itself, even though
when it comes to attempts it seems more logical to pull that information from UNISON. So why don’t we use UNISON?
Because I’ve missed a metric. Scheduled accounts.
Technically what we call a scheduled account is a reporting term. We covered last week that through the filter process we
create a large number of contact lists. We class some of these as scheduled (meaning we should be dialling them) and others
as non-scheduled (we can’t dial them). So scheduled accounts is best thought of as volume. How many accounts do we have
in a particular contact list.
The volume in a contact list will not change through out the day. It remains static. Prior to EWD we could measure a metric
like contact list volume and my other unique metric, Accounts Called* through Unison, no problems. Not so much any more.
*the other unique metric is accounts worked. This is a new measure that we were able to introduce through EWD now that
we could see the account level detail.
Master Class EWD – What’s the Difference
Remember this from MasterClass 1. Its how the dialler worked without EWD.
Download
Dialler
Contact Lists
The process of feeding accounts into the dialler (the download) would occur once at
the beginning of the day. The dialler has its complete set of accounts that throughout
the day that won’t change. You wouldn’t keep adding new accounts throughout the
day. What you start with is what you have to get through.
Download
This is what the process looks like now with EWD.
The accounts aren’t fed into the dialler from the download. They are fed
into EWD and EWD feds the accounts through to the dialler. But the EWD
doesn’t feed the account all at once, it sends accounts in batches all
throughout the day.
EWD
Contact
Lists
Given that we can only access summarised data through
UNISON what do you think happens to its ability to
work out a metric like scheduled accounts and accounts
called?
Buranda
Dialler
Burnley
Dialler
Master Class EWD – What’s the Difference
Unison can no longer measure unique accounts.
Every time EWD sends accounts down to the dialler, the dialler believes that it is receiving new volume. So through its
reporting mechanisms it does something similar to the below.
Lets say my total volume in the contact list C_FEDS is 5000 accounts.
EWD sends 2000 accounts from C_FEDS down to the Buranda dialler
EWD sends 1500 accounts from C_FEDS down to the Burnley dialler.
EWD then sends 1500 accounts from C_FEDS down to the Buranda dialler
EWD then sends 1000 accounts from C_FEDS to the Burnley dialler.
Lets say that each of these accounts sent were dialled. The reporting would like something like this
EWD
UNISON
Scheduled = 5000
Scheduled = 6000
Accounts Called = 5000
Accounts Called = 6000
EWD can send the same account, multiple times and to different diallers. Through EWD we can work out what the true
scheduled volume and accounts called data is. Through UNISON we can not.
This means if anyone is pulling information directly from UNISON they need to be incredibly careful. Some of the
metrics are just wrong. That’s why the business does not have access to UNISON reporting.
Master Class EWD – What’s the Difference
So lets look at our base metrics and where we can and where we do source them from:
Metric
Available
Reporting Sources From?
Note
Scheduled Accounts
(Volume)
EWD only
EWD
Accounts Called
EWD only
EWD
Uniquely count of accounts
actioned by the dialler or
agent
Accounts Worked
EWD Only
EWD
Unique count of accounts
actioned only by an agent
Attempts
Both
EWD
Missing additional attempts
through EWD
Connects
Both
EWD
Count of the number of
accounts sent to an agent
RPC
Both
EWD
PTP
Both
EWD
Talk
UNISON Only
UNISON
Wrap
UNISON Only
UNISON
Allocate
UNISON Only
UNISON
Idle
UNISON Only
UNISON
Master Class EWD – Why we Did It
An argument can be made about where we should/could be taking the data from. We have no choice with
scheduled and accounts called. The data is simply not available any where else.
We came down on the side of EWD for those metrics in both purely because we can provide much more
in depth reporting metrics because we can get to the account level information. We simply can’t get to
that level through EWD.
By having access to the account level information we can split the dialler data by product, dpd risk. All
these additional ways to splits and understand the dialler performance. Without it, we could only report
out by contact list. No lower.
So what’s the exception with the EWD metrics:
• Only agent time data is sourced from Unison
• Additional attempts are not reported on.
Master Class EWD – Collections Pyramid
Download
Scheduled
Attempt
Connect
RPC
PTP
Kept%
Key Metric Overview
Master Class EWD – Key Metric Overview
In this session we’ve mainly focused on the raw metrics themselves. Some of these are important on
their own, others are used in the calculated metrics. I would recommend that every reads the metric
definitions document on the dialler dashboards for the complete list.
Scheduled Accounts : Volume
Saturation = Accounts Called/ Volume
Penetration = Attempts/Volume
Worked/Accounts Called
Agent Hours = Talk Time + Wrap Time + Allocate Time + Available Time
Idle % : Available Time/ Agent Hours
Abandon % = (Abort + Abandon)/Attempts
PTP/RPC
AHT = (Talk Time + Wrap Time + Allocate Time)/Connects
Connects/Attempts
PTP/Scheduled
Master Class EWD – Key Metric Overview
Calculated metric fun.
1. Saturation is at 70% and Penetration is at 50%.
2. Penetration is 230% and saturation is 70%.
3.Saturation is at 120%.
4. Worked/Accounts Called is at 50%.