=Paper=
{{Paper
|id=Vol-2889/PAPER_07
|storemode=property
|title=Choice of Proper Data Operating Model: A study in telecom industry
|pdfUrl=https://ceur-ws.org/Vol-2889/PAPER_07.pdf
|volume=Vol-2889
|authors=Saumya Choudhury
}}
==Choice of Proper Data Operating Model: A study in telecom industry==
https://ceur-ws.org/Vol-2889/PAPER_07.pdf
Choice of Proper Data Operating Model: A study in telecom industry
Saumya Choudhury
Principal Consultant: Ericsson Global Consulting
Lovely Professional University, Phagwara, Punjab, India
Abstract
In the telecom sector the competition is always very high barring few occasions. Each of the service
providers wants to stay top in the game and the role of data analytics is gaining more and more
momentum to unearth the insights from all the data captured. The events of Covid-19 outbroken in
early 2020 have turned the world completely upside down. Considering the exceptional economic &
health crisis, organizations scrambled to adjust their ways of working to run their daily operations.
They could no longer rely on previous assumptions about their customers, including their buying
patterns e.g.; when the selling curve will go up, what are the seasonal patterns, what product mix
make them buy etc. In no days, brick & mortar stores closed due to panic, e-commerce sales gradient
rising, and customer center interactions exploded. Meanwhile, the new normal defines the new
consumption pattern of media as more people started working from home, spending more time
online and watching TV, and virtual interaction is all time high rather than contacting in person.
Rapid change is obvious in this crisis period, and more than ever, organizations need to make
decisions quickly that are however anchored in data. Yet, even as organizations bury themselves in
data, they are getting an incomplete picture of performance and their customers and almost all the
time this is data related [10]. This creates the classic dichotomy; you rely on data to make the
decision and you are not sure whether data has the proper quality or not. Data is an important factor,
for any strategy the leadership team of any organization is willing to take. In Telecom industry,
managing data effectively and efficiently is one of the toughest challenges. Often, different
functional departments and sub-functional departments create their own version of data and
applications which can help their day to day activities. This kind of fit-for-application and their own
set of data elements create the silos within the organization, duplicate the effort and make it nearly
impossible to manage the data democratically. Different departments having different versions of the
truth leads to plentiful issues including poor operational, predictive & regulatory reporting. In a big
telecom organization, it is common that the same enterprise, network and product data gets
replicated, processed and managed multiple times throughout the company. Transitioning a telecom
organization to a truly data driven organization where data is 'managed' is not only difficult but need
to overcome numerous common challenges. A successful data-operating model across the
organization is the answer to that [1]. A successful data operating model helps to disrupt the
technical silos existing in an organization. It builds upon the business model clearly indicates the
value created out of it with a long-term goal alignment and addresses the way, data is going to be
handled across the newly defined organizational processes; all the way from upstream data
collection, cleansing and enrichment to the referencing and the downstream use of raw or
transformed data [11].
Keywords 1
Data Operating Model, Data Transformation, Centralized/Federated Model, Data Lineage, Data
Quality
WCNC-2021: Workshop on Computer Networks & Communications, May 01, 2021, Chennai, India.
EMAIL: e08saumya@iima.ac.in (Saumya Choudhury)
ORCID: 0000-0001-5711-834X (Saumya Choudhury)
© 2021 Copyright for this paper by its authors.
Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR Workshop Proceedings (CEUR-WS.org)
69
1. Introduction
Figure 1: Example of a common data operating model
Common blockers for the data operating model:
Several data management hubs in a telecom organization like infrastructure, digital business, manages
services are very likely to create unreliable data [12], with succeeding issues including additional effort for
data quality assessment and the probability of poor insight creation and predictive decision-making. Each
group’s rules for data quality assessment and control procedure may diverge and for the same data set from
external data vendors multiple collection requests raised.
To optimize the operational expenditures, it is essential to remove duplication and incompetence in
processes. Operational inefficiencies will hit the organization’s bottom line.
A no or poor data operating model led to many issues [2]. Some of the common issues [7] in telecom are
listed below:
Collection of same data for different fit-for-applications
Increased operational expenditure to manage the same data set in different source systems
Duplication of data management efforts
Issues facing during reconciliations
Issues regarding reporting using different values of same data elements
Validation of inaccurate data elements
Outdated data elements
No clear accountability of data elements
Trade breaks
Incorrect operational reporting
Incorrect cognitive reporting
70
Incorrect insight creation
Long time risk associated with inconsistent data
Incorrect compliance reporting
Incorrect regulatory reporting
Poor decision due to inefficient market analysis
Improper investment decisions
RFP/RFI – inability to take quick action
Incorrect product performance reporting
Incorrect network performance reporting
Increase of capital expenditure for collection of unnecessary market data
2. Ideas in Brief
Table 1: The Problem, The Argument, and The Probable Solution
The Problem The Argument The Probable Solution
So how the data of an The telecommunication sector is Data Operating Model is a must
organization will be managed. experiencing a major change like as a good data operating model
Will the model be addressing moving from 4G (4th generation) in place can aids the
the centralized model to have to 5G (5th generation) organization to break down the
the structure and un-structured technologies while a plethora of un-related data silos within a
data or the federated model new business functions related to business and can jeopardize the
where each of the sub- newer technologies like cloud, business with wrong
organizations can have their say Internet of Things (IOT), interpretation. Data operating
to keep the data while the Artificial intelligence / Machine model generally builds upon the
overall data architecture & Learning (AL/ML) based service business model as mentioned
policy be in place? operation are taking place. The above and speaks how data is
data operating model starts with being treated across
identifying the data strategy, organizational processes, all the
which data would be required, way from data collection,
how those data can create the cleansing, transformation and
value, data lineage, data quality, enrichment to the sharing and
single source of truth for data use of data, i.e., the whole
etc. Finally, the organization upstream and downstream of
based on the size and vision data. As the data moves around
should identify the model ideal these different phases of the data
for the organization however, operating model and data
should have the scope to move to lifecycle, the business and
some other models if need arises. technical architecture play an
important role as well, yet many
businesses struggle when trying
to transition from the age-old
legacy- systems to newer
technology or complement
existing systems like cloud data
storage. A detailed and
comparative study with the basic
parameter of creating a data
operating model is discussed in
the next section.
71
3. Comparing the different data operating models for telecom organization
As mentioned earlier to start the data operating model journey in an organization where different
department span across different geographies and diverse stakeholders some essential aspects need to be
addressed to achieve a wining operating model.
Figure 2: A Typical Data Operating Model Journey
There is no one size fit-for-all solution. The idea behind defining a data operating model should consider
the aspects of control, type of organization and capabilities in different business units, sales organization and
common enterprise functions [5, 6]. An effective data quality service orchestration highly depends upon the
socio-cultural aspects of the business. The solution needs some of the common understanding in the
organization to achieve the vision & strategy e.g.; shared business accountability, sponsorship from data
domains and line organizations, shared attitude to data governance, knowledge base created & imparted and
inclination to embrace changes. Striking a right balance amongst to the above aspects defines a successful
data operating model that results to the enterprise to reach its data quality goals. A constant feedback process
[3], helps to reduce the gaps and refine the processes continuously and assists the enterprise to its’ goals.
The value driven approach should be the deciding factor whether the governance model needs to be
centralized or federated. If the value created from one of the models can outweigh the others by at least 20%
then the call should be in favor of that model.
Below is the table showing the merits and demerits of both the types which gives us the clear idea how to
choose the right operating model:
72
Table 2: Parameters for Data Operating Models
Value Parameters Centralized Model Federated Model
Easy to manage. Reporting is Managed by the business unit’s
Benefit management easy to the organization data management team if the
leadership team unit is operating independently
The central cross functional team The source system owner and
produces the rules, policies, the business unit produce the
Accountability
controls etc. but for static data rules, policies, controls etc. for
only static as well dynamic data
Here decision is taken by data
Mostly leadership decision i.e.,
Change management domain managers, data stewards
top down approach
etc., i.e., bottom up approach
Time required to turn
High to very high Low, mostly agile
around
Normal skillset with some re-
Required skillset Highly skilled people required
skilling would be required
Process efficiency should be Regular knowledge with some
Process management
high re-skilling will suffice
Stakeholder The number is less and how the
High to very high
management business unit structured
Agile approach can be taken
Data catalog handling Agile approach can be taken once the single truth of data
established
Since handled centrally, can be Since handled locally, reporting
Reporting mechanism easier at enterprise level but at enterprise level might be an
quick change might be an issue issue, but change can be faster
Centralized but continually Siloed and the model can be
Framework
improved by the feedback tweaked having the structure
management
mechanism, with little flexibility intact
Data quality Comparatively lesser effort
Large effort is required
management required
Knowledge A central repository to be Locally driven by the business
management formed units
Stakes are high. A central expert Stakes are low. Localized
Rules management
team is required control is required
The impact is difficult to
Any change impact is easy to measure. Alignment should be
Policy management
measure and align taken case from the very
beginning
Infrastructure Centralized control of the source Distributed source system
management systems control
Most of the time it is
Easy to access, monitor and push
Service management overarching and less dependent
to change
to applications
Based on the type of the organization and assessing the value parameters a decision can be taken.
73
4. Conclusion
The big question is which model is suitable for the organization or more specific to the target organization?
The answer may not be simple as many factors are associated with; let say what kind of organization (as we
are talking only about the telecom companies) is this.
Telco organization can be B2B (telecom gear manufacturer or app manufacturer) or B2C (service providers).
The business strategy is the prime, so the decision should be taken in alignment with that. Sometimes size of
the organization is so huge and have varied businesses all related to telecom but in different sectors like
service, infrastructure, network etc. In those scenarios, having one single model or centralized model might
exercise control on different department but on the other hand flexibility will be much lower and
organization will be more bureaucratic. Each change in the system will take more time as compared to an
agile organization. So, for the diversified organization it is advisable to use federated model rather than
centralized one.
A good data operating model is the catalyst to break the business and technical silos. On the other
hand, a poor or non-existence of data operating model can create frustration, unnecessary delay in
reporting, compromised reporting etc. For the same type of telecom application developer companies
can afford to have the centralized model as it can give the leverage to prototype the applications in
framework and can sell & distribute under common standard operating procedure.
Not all the value creation parameter will be applied every time for every telco organization. As
discussed above, in telecom we use many value creation parameters but for different set of
businesses we use different subsets. Let say for service offerings, parameters like touchpoints
reduction, mean time before failure, reduction of response time etc., are considered. If the
organization is a service provider like Reliance Jio or VI or Airtel, they need to have the clear-cut
data strategy in place which acts as a guideline to the whole organization, but the different verticals
should have the liberty to implement the inner structure according to their business.
Not every time a single model would be the answer to the data solution in a telco organization. It can
be centralized, federated or a mix of both like some hybrid model needs to be adopted to find the
right solution.
5. References
[1] https://www.quantzig.com/blog/big-data-challenges-troubling-managers-telecom-industry
[2] Introducing the next generation operating model, Mckinsey on Digital Service by Joao Dias, Somesh
Khanna, Chirstopher Paquette Marta, Rohr Barr Seitz, Alex Singla, Rohit Sood, Jasper van Ouwerkerk
[3] Competitive Strategy, What Sales Teams Should Do to Prepare for the Next Recession
by Mark Kovac and Jamie Cleghorn November 23, 2018, HBR
[4] Guide to Build Better How Mergers Change the Way Your Company Competes
by Benjamin Gomes-Casseres June 12, 2018, HBR
[5] Big Companies Are Embracing Analytics, But Most Still Don’t Have a Data-Driven Culture by Thomas
H. Davenport and Randy Bean February 15, 2018, HBR
[6] Innovation: Discovery-Driven Digital Transformation by Rita Gunther McGrath
and Ryan McManus From the May–June 2020 Issue, HBR
[7] Building an Effective & Extensible Data & Analytics Operating Model, August 2018,
https://www.cognizant.com/whitepapers/building-an-effective-and-extensible-data-and-analytics-operating-
model-codex3579.pdf
[8] Entrepreneurship: The New-Market Conundrum by Rory McDonald and Kathleen M. Eisenhardt From
the May–June 2020 Issue, HBR
74
[9] Business Models: Reinventing the Direct-to-Consumer Business Model by Len Schlesinger, Matt
Higgins and Shaye Roseman March 31, 2020, HBR
[10] https://www2.deloitte.com/us/en/pages/consulting/articles/data-analytics-operating-model.html,
Delloitte US
[11] Racknap Blog: Top 5 challenges & trends in telecommunication industry in 2020 February 16,
2020 by Munesh Jadoun
[12] The Data-Driven Operating Model, Cameron Warren, Nov 15, 2019,
https://towardsdatascience.com/the-data-driven-operating-model-2fa1b72c0f1d
[13] Data-Driven Decisions Start with These 4 Questions by Eric Haller and Greg Satell February 11, 2020,
HBR
[14] Using AI to Make Knowledge Workers More Effective by Paul R. Daugherty and H. James Wilson,
April 19, 2019, HBR.
75