Innovating Telecom: the early bird catches the worm
Introduction
In recent years, the telecommunications (telecom) industry has seen an unparalleled surge in demand for its products and services. The recent COVID-19 outbreak, as well as the widespread transformation of traditional services to the digital world, has increased the need for a consumer-friendly and reliable network. This year, an estimated 22.8 million Australians used the internet, compared to only 22.3 million in 2020, a 2.3% rise since the pandemic began (Kemp, 2021). As the number of users grows, so will the impact on customer satisfaction and network maintenance expenses. In this article, we will explore the current pain points and opportunities for data science to address these challenges.
Improving the customer experience is crucial for telecom operators who already experience a high churn rate, due to poor image among customers, other market competitors and high operational costs. According to a PwC survey, an average 24% of dissatisfied consumers will change to another competitor if they have just one bad experience (Puthiyamadam & Reyes, 2018).
Similarly, telecom operators’ infrastructure maintenance is critical to their operations, as failing would result in revenue loss, customer churn, and brand reputational damage. The cost of maintenance is projected to rise as more 5G tower cells are required to offer the same level of coverage as existing cellular towers (Condon, 2020). Since consumers expect the network infrastructure to be accessible 24 hours a day, minimising the chances of equipment failure is essential. The future of telecom would be in the hands of the company who can quickly grasp these innovations and successfully implement them.
Predictive analytics to improve customer experience
Current outlook
Measuring customer satisfaction and decreasing churn are constant challenges in any firm. Telecom is unlike other industries, such as airlines, in that it already holds a negative sentiment among customers for providing a poor customer experience. According to the Australian Communications Consumer Action Network’s (ACCAN) survey, customers spent an average of 13 minutes (through social media) to 65 minutes (by phone call) looking for a solution to their problem. Only approximately 40% of the total participants reported were able to address their issue on the first try, with the average instances of contact being 2.3 times (ACCAN, 2020). To understand the magnitude of this issue, the researchers subsequently transformed this data into an economic cost, estimated per minute. The money lost while waiting for a resolution when the customer could be doing something else, is referred to as the economic cost.
Table 1. Shows the average economic cost of waiting by method of contact, for a resolution.
Solution: what if we had a crystal ball?
Data science innovations like predictive analytics can be used to address this challenge, using customer demographics, purchase, and product usage history. The ability to predict customer concerns and determine who is more likely to buy allows telecom operators to sell their services more effectively to the right audience. According to a literature review by Ahmad et al. (2019), increases in acquisition, customer loyalty, and retention can all be aided by the proper implementation of predictive modelling. Furthermore, the likelihood of a customer cancelling their subscription can be calculated and allow telecom operators to intervene in advance, like using personalised special offers.
Currently, telecom operators use a customer lifetime value and segmentation model to identify groups of consumers and their likelihood of purchasing a new product or service. This data is then crafted into marketing materials, to deliver personalised offers and shopping experiences. However, this reactive strategy is unable to forecast changing customer sentiments and behaviour. For example, during the recent COVID-19 pandemic, telecom operators observed changes in customer expectations as people lost their income and reduced spending.
Data to the rescue
Telecom operators can leverage massive quantities of information available at their disposal, to create a 360-degree profile of their customers to gain further insights (Stojanovic et al., 2019). They will be able to be more proactive in resolving concerns, decreasing complaints, and attrition, if they can predict a customer’s need before they even realise it. If predictions can assist telecom providers in understanding the reasons why customers are cancelling, they will be able to take more personalised pre-emptive action. As a result, expectations are met, and customers feel heard, resulting in higher revenue and satisfaction (Petrovi, 2019).
Machine learning techniques like clustering can aid customer segmentation, which group customers based on their demographics and purchasing behaviour. Moreover, customers expect a better experience and an understanding of their needs as already know that their telecom operators have access to their information (Stojanovic et al., 2019).
Implementation and ethics: it’s not all sunshine and rainbows
Despite the claims of being able to predict customer requirements and understand behaviour, customers’ privacy must be respected and protected. According to Mühlhoff’s (2021) research, predictive analytics has the inherent potential to exacerbate unfair treatment and discrimination due to data misuse. Decisions made because of these models may unjustly target demographic groups, based on aggregate data from other unsuspecting customers.
Any predictive algorithm would necessitate access to massive volumes of customer which must be kept safe and used solely for the stated purpose, which is to improve the customer’s experience. Price discrimination against specific consumer segments to ‘personalise’ and maximise the prices each customer is willing to pay is unethical (Banche et al., 2016). For example, like withholding discounts from high income earners, because they can afford it. In addition, many telecom operators offer data-driven opt-in membership programs that seek customers for permission to utilise their data in exchange for loyalty points. These programs are based on customer trust to improve their customer experience, but usually are unaware of how this information is being used (Davenport & Harris, 2014).
The implementation process is straightforward, using customer data to train a machine learning model using a variety of variables. But it must have the proper protections and data regulations in place to ensure transparency about how data about them is used to drive these prediction models.
Predictive maintenance to save money
Current outlook
Telecom operators have a diverse assortment of hardware that serves millions of clients at the same time, depending on the network demand. Customers have a right to expect their services to be always available, because they are paying for them (Condon, 2020). Internet downtimes and disruptions negatively impact consumer experiences, expectations, and company reputations. Unfortunately, as more consumers utilise their services, the likelihood of network equipment failures also rises, leading to predictive maintenance can help reduce the cost of infrastructure upkeep (Abiad et al., 2018). Rather than waiting, telecom operators can take proactive measures to resolve issues before they become a problem, as the demand for a reliable and powerful network grows when more people and businesses connect to the internet. A four-hour Telstra network outage in 2019, for example, was estimated to cost the economy up to AUD$100 million due to the disruption of banking services (Cooke, 2019).
Solution: What if we knew when the equipment will fail?
Predictive maintenance uses machine learning algorithms to detect anomalies and hardware breakdowns before they occur. The frequency of maintenance and inspections is reduced by up to 50% with this proactive strategy (Karpowicz, 2018). Internet of Things (IoT) sensors deployed on various portions of the infrastructure collects historical and real-time data, allowing synchronised monitoring to detect faults before they become problems (Fiix, 2021). With machine learning (ML), IoT sensors, and AI-based drone inspections, telecom operators can decrease network maintenance costs and avoid dissatisfied customers and revenue loss. Some telecom operators still use reactive maintenance, which means they react to issues and outages after they happen. Alternatively, a run-to-failure approach could be employed, in which the equipment is used until it fails. The combination of learning from historical patterns, analysing data from real-time reporting and autonomous drone inspections can innovate how telecom operators work and promote a more proactive approach (Abiad et al., 2018).
Impact on the future of maintenance
Telecom operators can reduce expenses while simultaneously protecting the safety of their workers by combining machine learning algorithms, real-time IoT data reports, and AI-based drone inspections. Traditionally, field technicians climbed cellular towers to undertake routine inspections and maintenance. They were regarded as the most dangerous employment in a 2012 American survey with a 10 times greater risk of fatality than construction workers (Knutson & Day, 2012). These workers were frequently exposed to the environment, electrical hazards, and the heightened risk of falls from considerable heights and structural collapse. Alternatively, by using advanced cameras and sensors, autonomous drones can perform a safer and more comprehensive assessment in less time than humans. In contrast to hiring experienced tower technicians, there is usually no professional training or risk involved when deploying a drone (Karpowicz, 2018). The data collected can also be analysed in real-time, giving specialists the necessary information, they need to make decisions like urgent repairs.
Predictive maintenance will save telecom operators’ money while extending the equipment’s lifespan using data-driven decisions. Customers will be notified in advance about scheduled maintenance in their area, with work able to be completed outside of business hours to minimise disruption. Problems will be dealt with and resolved ahead of time, resulting in happy customers and increased value for the telecom operator.
Implementation: guessing exactly when the equipment will fail is hard
Telecom operators can foresee and predict the equipment’s lifetime expectancy allowing them to make informed decisions about when to upgrade or perform maintenance (Abiad et al., 2018). To implement this strategy, telecom must:
· collect large amounts of historical and real-time data to establish a comprehensive view of the infrastructure
· develop an advanced machine learning engine that can differentiate between numerous variables
· apply the predictive model’s findings to better understand and avoid network outages and customer complaints
Effective implementation, on the other hand, necessitates professional technicians who comprehend the data, a considerable upfront investment to purchase highly autonomous drones, build the ML system, and employ skilled personnel. As a result, predictive maintenance is out of reach for smaller operators.
Conclusion
In this paper, we have highlighted two important uses of data in the telecommunication industry: predictive customer analytics and maintenance. Benefits can be seen for both consumers and telecom carriers because of advances in machine learning and predictive modelling. By personalising consumer interactions and exploiting previous data, predictive analytics can enable better customer experiences, but data must be appropriately handled to avoid unethical practices. Similarly, predictive maintenance can decrease customer churn while simultaneously enabling telecom operators to make data-driven decisions about how to extend the life expectancy of their equipment. Ultimately, these innovations will revolutionise the telecommunication industry.
References
Abiad, M., Kadry, S., & Ionescu, S. (2018). Preventive & Predictive Maintenance of Telecommunication Equipment — A Review. 2018 4th International Conference on Applied and Theoretical Computing and Communication Technology (ICATccT). Published. https://doi.org/10.1109/iCATccT44854.2018.9001972
Ahmad, A. K., Jafar, A., & Aljoumaa, K. (2019). Customer churn prediction in telecom using machine learning in big data platform. Journal of Big Data, 6(1). https://doi.org/10.1186/s40537-019-0191-6
Australian Communications Consumer Action Network. (2020, December 16). Still Waiting . . . the cost of customer service. https://accan.org.au/media-centre/media-releases/hot-issues/1825-still-waiting-the-cost-of-customer-service
Banche, B., Luisada, F., Oundjian, A., Schicht, R., & Wilms, M. (2016, July 21). How Telecom Can Put Their Money Where Their Customers Are. BCG. https://www.bcg.com/en-au/publications/2016/cost-efficiency-telecom-put-money-where-customers-are
Condon, J. (2020, October 1). Investing in cell towers: 5G is expected to take them to new heights. Institutional Real Estate, Inc. https://irei.com/publications/article/investing-cell-towers-5g-expected-take-new-heights/
Cooke, G. N. C. (2019, July 12). Telstra “sorry” for outage which retailers say will cost them $100m. The Sydney Morning Herald. https://www.smh.com.au/business/banking-and-finance/telstra-sorry-for-outage-which-retailers-say-will-cost-them-100m-20190712-p526jy.html
Davenport, T. H. D., & Harris, J. H. (2014, August 1). The Dark Side of Customer Analytics. Harvard Business Review. https://hbr.org/2007/05/the-dark-side-of-customer-analytics
Fiix. (2021, May 13). What is Predictive Maintenance? [Benefits & Examples]. https://www.fiixsoftware.com/maintenance-strategies/predictive-maintenance/
Hubert, M. H., & Ribeiro, T. R. (2017, November 7). How predictive maintenance can innovate the Telecom sector. Innovation Matrix. https://www.innovationmatrix.com/homepage/how-predictive-maintenance-can-promote-innovation-in-the-telecoms-sector
Karpowicz, J. K. (2018, March 29). Reducing Cell Tower Inspection Costs by Up to 50% with Drones. Commercial UAV News. https://www.commercialuavnews.com/energy/drones-used-help-reduce-cell-tower-inspections-50
Kemp, S. (2021, February 10). Digital in Australia: All the Statistics You Need in 2021. DataReportal — Global Digital Insights. https://datareportal.com/reports/digital-2021-australia
Knutson, R. K., & Day, L. D. (2012, May 22). In Race For Better Cell Service, Men Who Climb Towers Pay With Their Lives. ProPublica. https://www.propublica.org/article/cell-tower-fatalities
Mahajan, V., Misra, R., & Mahajan, R. (2017). Review on factors affecting customer churn in telecom sector. International Journal of Data Analysis Techniques and Strategies, 9(2), 122. https://doi.org/10.1504/ijdats.2017.085898
Mühlhoff, R. (2021). Predictive privacy: towards an applied ethics of data analytics. Ethics and Information Technology. Published. https://doi.org/10.1007/s10676-021-09606-x
Petrović, N. (2019). Churn Prediction in Telco Industry Leveraging Call Center Data. In Proceedings of the 6th International Conference on Electrical, Electronic and Computing Engineering-IcETRAN (pp. 845–850).
Puthiyamadam, T. P., & Reyes, J. R. (2018). Experience is everything: Here’s how to get it right. PwC. https://www.pwc.com/us/en/zz-test/assets/pwc-consumer-intelligence-series-customer-experience.pdf
Stojanovic, M., Bogdanovic, Z., Barac, D., Radenkovic, M., & Mihajlovic-Milicevic, J. (2019). The Role of AI in the Transformation of Mobile Operators. 2019 International Conference on Artificial Intelligence: Applications and Innovations (IC-AIAI). Published. https://doi.org/10.1109/ic-aiai48757.2019.00026
