In the last couple of
blogs, we discussed about the three key questions an organization needs to
answer to bring in clarity in its CX improvement program as well as assessing
and enhancing benefits from the same. These are,
- Should Customer experience be measured as stand-alone value or it should be measured in terms of rate of change in the CX measure?
- Should we measure using Direct Customer Feedback or through Indirect (derived) feedback?
- Should we use past data or future (Predicted) data?
We deliberated first
two questions in the earlier blogs and will discuss the third one in this blog.
Refer following links to read through- https://personalandprofessionalexcellence.blogspot.com/2019/11/customer-experience-measurement-stand.html,
This blog talks about
the third question regarding usage of “past” or “future data (predictions)” in
order to bring about business improvement using CX. Before we get into the
discussion around data and its state, let us look at the typical process of CX
measurement and how it helps in business improvements.
Following diagram
depicts the process in simple terms.
Organization assesses
customer journey for a specific aspect of business and identifies areas where
an intervention for improving CX is necessary. It then designs and implements the
intervention based on the findings from journey assessment as well as its understanding
of customer preferences & behavior. Feedback from the customer is collected
through survey / interviews using online / in-person mechanism. If the feedback
is positive i.e. CX is good, then organization continues the journey to reap
benefits of the intervention. However, if the feedback is mixed or not
encouraging or there is no impact at all; organization identifies shortcomings/misses
during design / implementation of the intervention and subsequently defines a tweak
that would be necessary to address the same. This tweaked intervention is again
implemented, and the same cycle follows. If the mistake is such that it cannot
be corrected by smaller tweaks in the intervention, then organization abandons
the initiative as early as possible to contain the damage it might be creating.
The cycle from assessing the feedback till implementation of tweaked intervention easily takes
2-4 months. If the right fix requires more than one tweak, the delay in getting
the desired results increases proportionately.
As we know, CX (as well as
related revenue impact) follows a lifecycle similar to product lifecycle as
shown in the following diagram. Refer following link to understand detailed
discussion around CX lifecycle. https://personalandprofessionalexcellence.blogspot.com/2019/07/cx-life-cycle-management-for-continuous.html Once the CX intervention is correctly
designed and implemented, it starts showing positive results on experience as
well as revenue as shown in the diagram. However, if the intervention is not
appropriately designed and customers do not embrace it as expected, the tweak
cycle kicks in. After one or two tweaks the CX journey settles in. In such
cases the graph for CX shifts to the right (Blue Line in the diagram below) creating
a period of uncertainty from the start of first intervention to the
implementation of intervention after tweak/s. 
During this period, CX may go
down or waiver or may not change at all. The company is busy in finding the
right tweak and can be vulnerable for attacks from competition.
There are chances that it may lose revenue as well as goodwill with customer
base.
There are two
challenges faced by company in this endeavor. First one is feedback cycle
takes significant time delaying the tweak that might be required to be
implemented. Process of getting feedback
from customer has inherent limitations in terms of how frequently you can go to
customer and ask their feedback coupled with no guarantee that customer will
provide feedback when reached through modes like e-mail or online survey. Organization
is pushed into uncertainty zone and there is no easy way to reduce this time. This
limitation could be eliminated to some extent by creating multiple sets of
customer segment with similar characteristics and reaching out to them at
higher frequency; however, the process becomes complicated and does not help
reducing uncertainty period by significant proportion.
Second challenge is
that we really do not know when the CX lifecycle attains maturity. The
trajectory is known but the time period for each zone like growth or maturity
is not known and it changes with the specific intervention and its impact. It
could be a steep curve up and steep down or it could be slow up and steep
downward or it could be steep up and slow down. It is critical for a company to
understand the stage of lifecycle CX is passing through. If the CX is in upward
climb, organization continues to reap in benefits from the intervention whereas
if it is on the downward slope, the organization starts reducing the benefits that
it receives through this intervention and may start losing revenue. Organizations
wish to elongate the growth stage as long as possible and would like to be in
maturity stage forever. However, in reality this does not happen, the impact of
the CX intervention weans away at some point of time and CX graph enters in to
decline phase. It is critical for organization to plan another tweak or new
intervention right in time so that its impact starts showing up by the time the
impact from earlier intervention gets in to decline mode. The organization struggles
to find this point, it cannot estimate the time it is going to take for any
intervention to reach this maturity/decline inflection point (CX fatigue). It
is also difficult to measure the CX level continuously to track its progress
and estimate the next intervention point leaving organization vulnerable.
Can we overcome
these challenges? – YES we can !!!
The problem could be
solved if the organization uses future (predicted) data. Organization can
construct a model using indirect feedback mechanism as described in the earlier
blog. This model will enable organization to obtain the exact stage of CX at
desired frequency like daily. It can also go to a hourly frequency if we have
the right model with right parameters and technological capability to
manipulate the data collected through these points.
If organization
creates a model using various data points from different interventions which
depict the life cycle pattern for an initiative and build a capability to self-learn;
this could create a tremendous intelligence for the organization to predict
what is going to happen in near as well as distant future and when it should be
ready for next intervention. The customer behavior is unpredictable and CX
movement may not be same as what is predicted or there could be a significant
campaign run by competition which is impacting the CX behavior; the
organization can get a sense of these movements from the capture of indirect
feedback without really waiting for physical feedback and can adjust the
projections accordingly. This generates new predictions for organization to re-calibrate
their actions on real time basis. The predictions are dependent on the AI/ML
based self-learning model. The closer it reflects the reality as well as its
ability of mid course correction determines the effectiveness.
In summary
- Mapping the CX value v/s Rate of change of CX can provide the status of CX initiative and direction that an organization should take going forward
- The indirect method to capture CX using parameters available in the organizations business systems wins over the direct data collection mechanisms
- Predicted data can provide organization with inputs which are early and granular than the past data collected from the customers
- The success of the CX backed business improvement depends largely on ability to create the algorithm which could represent the customer behavior using the parameters available for us in the system as well as one for predicting the CX lifecycle. AI/ML can enable organizations to achieve this and can help move ahead of curve in continuously improving CX as well as improving the growth rate for the business.
