Mais conteúdo relacionado Semelhante a Voice Analytics in Action - AHT (20) Voice Analytics in Action - AHT2. An iconic marketing agency was
unable to establish an average
handle time for successful calls in
outbound campaigns.
The company approached Call
Journey team with a request to
analyse its recordings and find the
reasons behind the large
discrepancy in the lengths of
successful calls.
Context of the study.
Background
Industry
Digital Marketing
Challenge
Outbound call analysis
and improvement
Scope
Call Centre Conversations
Audio Sample
5,000 minutes
3. listen better
© 2016 – TMA COMMS PTY LTD
Approach
Identify • Identify Successful Calls
Profile
• Profile and identify key segments
in a successful call
Create
Model
• Create model using key
words and criteria
Deploy
Model
• Deploy model across all calls to
assess close time
Solutions • Recommend actions
A process devised to segment calls and identify close times.
EVS tools
• CallTranscription
• Keyword Spotting
• Categorisation
4. listen better
© 2016 – TMA COMMS PTY LTD
We identified 11 successful calls.
Call Classification:
1707 calls categorised
11 identified as successful
• Sales – Product 1 (7)
• Sales – Product 2 (4)
Close rate of 0.7% (tagged calls)
Successful calls are the longest
EVS in Action: Identify
20.5
10.0
7.6
2.5
2.1
2.0
1.9
1.1
1.0
0.6
0.6
0.5
0.2
0.0
Sales - Product 1
Sales - Product 2
Cat3gory 3
Category 4
Category 5
Category 6
Category 7
Category 8
Category 9
Category 10
Category 11
Category 12
Category 13
Category 14
Average Call Duration (minutes)
7
4
3
12
249
1
8
1
8
143
1
148
1
Number of Calls
1120
5. listen better
© 2016 – TMA COMMS PTY LTD
Key Call Metrics
Business Result Increase Product
Duration (minutes) 9.8
Word Count 1,867
Agent Word % 87.1%
Customer Word % 12.9%
Hi Kate… my
name is Michael
from
0.16 min
Four major segments in a successful call
Good. Thank you.
Michael and yourself
0.21 min
5.0 minutes 1.4 minutes 3.4 minutesCall 11
Customer
Agent
… total cost
of $264 over
the 12
month…
5.05 min
I’ll give
it a go
5.12 min
… explain to
you where we’re…
3.06 min
… please state your
full name and confirm
you give me
permission…
6.19 min
… put you through
to the new owner…
0.47 min
Robert… I give
you permission…
6.25 min
Confirm contact
details
The Pitch The CloseBilling
… got to let you know…
if not happy… cancel
for the next year.
7.52 min
… thanks
mate.
9.44 min
Thank you
for your
time bye
9.45 min
EVS in Action: Profile
6. listen better
© 2016 – TMA COMMS PTY LTD
A model was created to identify and dissect successful calls.
Analyse
Known
Data
Input
Unknown
Data
Predicted
Results
CREATE
MODEL
Deploy
Model
Feedback Loop
DEPLOY
MODEL
Create
Model
Test &
Evaluate
Model
Deploy
Model
Analyse
Results
Continuous
Improvement
EVS in Action: Create Model
7. listen better
© 2016 – TMA COMMS PTY LTD
KEY WORD / PHRASE
authorise
confirm
email
cost
bill
month
you cancel
UTTERANCE
Confirm you are authorised
…authorising person
Confirmation email
Send you a confirmation
Confirmation of the listings
email to let you know…
products are ready
unless you cancel
you cancel your listing early
UTTERANCE
minimum contract…12
months…total cost
basic cost… additional cost
total cost
in terms of billing
everything is at one place…
your bill
monthly instalment
$[xx] per month
Step 1: We analysed successful calls and established a list of 7 key words.
EVS in Action: Create Model
8. listen better
© 2016 – TMA COMMS PTY LTD
Three criteria established to
identify successful calls:
Call duration ≥ 8 minutes
Total Key Words spoken ≥ 10
No. of Key Words spoken ≥ 5
EVS in Action: Derive criteria
KEY WORD / PHRASE
authorise
confirm
email
cost
bill
month
you cancel
No. of Key
Words spoken
≥ 5
Total Key
Words
spoken ≥
10
Call
duration ≥
8 minutes
Step 2: Establish criteria to identify all successful calls.
9. listen better
© 2016 – TMA COMMS PTY LTD
Step 3: Applying model against known data to evaluate outcomes
EVS in Action: Test Model
KEY WORD / PHRASE
authorise
confirm
email
cost
bill
month
you cancel
1,707 Tagged Calls
The 11 successful calls
correctly identified
1 anomaly
1,707
Tagged
Calls
No. of Key
Words spoken
≥ 5
Total Key
Words
spoken ≥
10
Call
duration ≥
8 minutes
10. listen better
© 2016 – TMA COMMS PTY LTD
BUSINESS
RESULT
CALL_ID
DURATION
(MINUTES)
WORD
COUNT
AGENT
WORD %
CUSTOME
R WORD
%
Category 5 Call 12 21.1 2,966 83.7% 16.3%
The one anomaly was due to agent continuing a ‘no sale’ call for 17 minutes.
Should I give you a
call back or are you
gonna send me an
email regarding
want you want?
Agent: 20.2 min
Get back to you
within 24 hours
Agent: 20.4 min
7.1 2.6 11.4
- 5.0 10.0 15.0 20.0
Call 12
1st text criteria met
2nd text criteria
met
I’ll just call
tomorrow…
Cust: 20.9 min
EVS in Action: Evaluate
I’ll have to talk to
the boss before I
spend the money
Cust: 3.8 min
Time criteria
met
Call meets the criteria
but doesn’t qualify
Agent continued call
long after customer was
identified as not
authorized to conclude a
sale
11. listen better
© 2016 – TMA COMMS PTY LTD
13.3
16.2
6.0
8.3
9.5
10.7
12.2
3.6
5.7
3.6
5.0
2.6
0.7
12.7
6.7
3.6
0.5
0.6
3.5
2.1
1.8
1.4
17.4
13.7
2.6
5.9
6.5
4.4
2.5
7.0
5.1
4.6
3.4
- 5.0 10.0 15.0 20.0 25.0 30.0 35.0
Call 1
Call 2
Call 3
Call 4
Call 5
Call 6
Call 7
Call 8
Call 9
Call 10
Call 11
Minutes
Successful Calls: Time to Close
Number of minutes call continued
after Text Based Criteria is met:
Large discrepancy in close times:
• Minimum: 2.5 minutes
• Maximum: 17.4 minutes
Time to meet Text Based Criteria:
• Shortest: 5.4 minutes
• Longest: 16.9 minutes
Model used to analyse successful calls and determine close times.
Close
Period
Over 16 minutes into call before Text
Based Criteria was met and still a
further 13+ to 17+ minutes to close.
Time to
meet
criteria
EVS in Action: Deploy Model
12. listen better
© 2016 – TMA COMMS PTY LTD
Impact
Unnecessary information gathering
Longer calls in favour of more calls
Large discrepancy in “close times”
Close times are longer than necessary
Key driver: script weighs down AHT, incentives promote long calls
Business Insights
Key Issues:
Too much time upfront confirming details
Agents rewarded for long calls
No uniform closing system
Lengthy reading of the T&Cs at the close
13. listen better
© 2016 – TMA COMMS PTY LTD
EVS Action Plan
EVS connects the dots and lets you drive instant change with tailored, data-backed solutions:
• Reduce weight given to long calls
• Introduce Incentives for call volume
• Confirm customer is the authorising person at beginning of call
• Only confirm contact details if a sale is made
• Reduce script by confirming T&Cs will be emailed
Re-design
process
Realign
agent
incentives
Notas do Editor Welcome everyone, and thank you for giving us your valuable time today, we greatly appreciate it.
Today we are going to show you how you can ‘Unlock the Voice of Your Business and Listen Better’ at true enterprise scale