Testing

Blog

  • What is AutoML and how it is democratizing AI?

    At a time when businesses are looking at adopting Artificial Intelligence (AI) not just for competitive advantage but even for mere survival, it is increasingly challenging to build a successful AI practice with acute skills shortage for data scientists. On the other hand, Machine Learning (ML), which is built for its application involving laborious tasks such as cleaning data, preparing data and training ML algorithms, validation etc. However, there is continuous effort to automate these tasks by built more intelligent ML procedures and algorithms. AutoML , as we call it, can democratize ML by allowing even business users to develop and execute their own data models with little to no training on data science. Other than bridging the skills gap, automation in ML processes can also eliminate data biases, a major concern today, and reduce human errors while improving overall efficiency. Moreover, AutoML would allow domain experts and technical experts like data scientists, ensuring continued focus on business value.

    The need for AutoML – Challenges with traditional ML processes

    The growing interest in AI and ML means that there is a crippling shortage of data scientists. There were over 2.7 million open positions for data science and analytics jobs, according to a report by the Business-Higher Education Forum.As per the US Bureau of Labor Statistics, the number of jobs in the data science field will grow by 26 percent through 2026, adding nearly 11.5 million new jobs.

    However, demand vastly outpaces supply for data scientists given how challenging it had been for several decades to work in this domain. It is impossible to generate hundreds of thousands of new data scientists in an instant, making it tough for organizations to implement their data science plans.Lack of these skillsets is one of the biggest reasons holding back thousands of companies from starting their AI journey. That said, automation is rapidly trying to solve this problem by making data science more accessible to even those without years of data science experience or even a degree in the subject.

    Even so, lack of required skills is not the only challenge that organizations looking at machine learning face today. Even if an organization has the right skills, it may still be highly under-utilized because of the sheer amount of time that it takes just to clean the data. Data scientists spend as much as two-thirds of their time just cleaning the data. Just imagine if this is automated, what kind of fillip it will provide to the domain.

    Further, data scientists often don’t come with domain and business expertise. However, even if bring domain and business understanding they end up focusing most of their time ingesting and processing data in order to make the models relevant. As a result specific business context often go amiss, leading to unsuccessful adoption of AI/ML.

    Traditional ML processes are also highly dependent on human expertise, given the amount of customization that each ML model requires for the specific problem on hand. This makes the entire process inherently time-consuming. To build a new ML model, you still have to through the rigours of data preparation, feature engineering, training the model, evaluation and selection.

    Biases in AI and ML models are also a major subject of debate today. Biases often creep in because of manual interventions and the inability of humans to analyze massive data sets for possible biases. The complexity of ML models currently has turned them into black boxes with very little visibility into what goes inside and what is impacting the final results.It is therefore vital to automate the process of machine learning to get better visibility into the models, eliminate all biases, and improve the overall efficiencies.

    What is AutoML?

    While machine learning continues to evolve, Automated Machine Learning (AutoML) goes beyond automation to accelerate the process of building ML and deep learning models. It automates several aspects of the ML processes, including the identification of the best performing algorithm from the available universe of features, algorithms and hyperparameters.

    How Does AutoML Help?

    By eliminating repetitive tasks, such as data cleaning, AutoML frees up the highly valued human resources to move towards value-adding analysis and more in-depth evaluation of the best-performing models. This allows enterprises to significantly cut down the time-to-market for the products and solutions built on these ML models.It:

    • Eliminates repetitive tasks
    • Allows enterprises to bring down time-to-market
    • Guided analytics capabilities allow to eradicates biases
    • Enables organizations to leverage their existing components
    • Inspires trust by providing transparency on how the model functions
    • Eliminates human error

    However, complete automation also has its own set of challenges. Tesla founder Elon Musk famously said “AI is far more dangerous than nukes.” Apart from Musk, technology leaders like Bill Gates and Steve Wozniak have expressed concern about the dangerous aspect of AI. For instance, anyone with malicious intent can program AI systems to carry out mass destruction. Any powerful technology can be misused and AI is no different. The truth is that as long as AI systems continue to be Black Boxes, it will continue to remain a threat.

    Some new age solutions are changing that equation by bringing in transparency and making it easier for users to interact better with AI systems. HyperSense AI Studio , for example, is built with guided analytics capabilities, which is a combination of automated ML and interactive ML. This allows usersto develop applications with a combination of automation and human interaction at any stage of the data science cycle based on task and business user requirements. The solution also generates alerts and gives recommendations to users as they are creating a pipeline.

    The process eliminates biases that might have crept in and ensures that the system is not seen as a Black Box by providing details of how it functions and arrives at the results.

    Through AutoML, the user can easily automate tasks like data pre-processing, feature engineering and hyper-parameter tuning. Moreover, it allows reusing features instead of rebuilding again from scratch for different models driving AI at scale.

    What’s trending?

    Several Machine Learning processes do not require any human intervention, allowing domain experts to work on building AI models instead of depending solely on the data scientists.

    Data scientists, however, do not have to be a rare commodity anymore. Just how the power of a mobile phone camera made citizen journalism possible, the power of AutoML is now creating citizen data scientists . This new breed of professionals will now be able to build their own AI models without any formal education in Machine Learning or AI. Anyone familiar with the usage of Excel and interest in data analysis can potentially become a citizen data scientist.

    The role of citizen data scientists will be critical in the growth of AI. In order to scale AI, one needs a massive number of data scientists. Moreover, citizen data scientists don’t just fill the skills gap. The biggest mismatch in ML initiatives is that ML projects are often associated with a lack of domain expertise. Data scientists are great at working on data, but they don’t necessarily come with a good understanding of your business or industry. Connecting the roles of domain expertise and data expertise has been a massive challenge for several firms.

    However, by putting the ability to build a data model into the hands of a business user, AI projects can move towards newer dimensions that can only be perceived by a business domain expert.

    What are the benefits of AutoML?

    Other than democratizing machine learning, AutoML also has several other advantages. Automating the machine learning processes, for example, can tremendously accelerate the speed of training multiple models while also improving accuracy. In addition, AutoML eliminates biases in datasets by limiting human intervention and automating most of the processes in the ML pipeline. The reduced human intervention also cuts down on human errors in the process.

    Automation also makes ML more scalable by enabling multiple ML models to be trained simultaneously, and in doing so, it also optimizes the overall ML processes to a great extent.

    HyperSense AI Studio is an excellent example of AutoML platform . The platform enables enterprises to build and operationalize AI successfully using automated machine learning. It increases the efficiency of data scientists allowing them to focus on higher-value tasks. It automates every step of the data science lifecycle including, feature engineering, algorithm selection, and hyper-parameter tuning.

    By leveraging HyperSense AI Studio , data scientists and domain experts can easily build ML models with higher scale, productivity, and efficiency while sustaining the model quality. By automating large part of the ML processes, the platform accelerates the time to get production-ready models with greater ease and efficiency. It also reduces human errors mainly because of manual measures in ML models.

    It also makes data science accessible to all, enabling both trained and non-trained resources to rapidly build accurate and robust models, thus fostering a decentralized process. Further, it enhances collaboration between domain and technical experts which encourages the focus to remain on business value and not on technical part of the implementation. This helps in bringing down silos and promotes collaboration in other areas as well.

    The quality of the machine learning model is not only based on code but also on the features used for running the model. Around 80% of data scientists’ time goes into creating, training, and testing data. HyperSense AI Studio comes built-in with a feature store that allows features to be registered, discovered, and used as a part of an ML pipeline. It allows reusing features instead of rebuilding again from scratch for different models driving AI at scale.

    Key Takeaway

    AI projects for long have been stuck at pilot stages due to several challenges that include lack of data scientists, slow progress in ML processes and even lack of coordination between business and data teams.According to a Gartner study, about 75 percent of organizations will shift from piloting to operationalizing AI by the end of 2024. Also, 50 percent of enterprises will devise AI orchestration platforms to operationalize AI. This, however, wouldn’t be possible without leveraging AutoML .

    AutoML has the potential of democratizing AI and Machine Learning and finally take AI projects from mere pilots to scaled deployments. AutoML platforms like HyperSense AI Studio increases the efficiency of data scientists by allowing them to focus on higher-value tasks. The platform automates every step of the data science lifecycle including, feature engineering, algorithm selection, and hyper-parameter tuning, ensuring enhanced operational efficiency. In addition, it comes built-in with a feature store that allows features to be registered, discovered, and used as a part of an ML pipeline and even allows reusing features instead of rebuilding again from scratch for different models driving AI at scale.

    Get better results from your data with HyperSense AutoML

    Try AI Studio for Free

  • ¡Deje que el Deep Learning domestique todas las facturas que reciba!

    ¡Deje que el Deep Learning domestique todas las facturas que reciba!

    En esta era de transformación digital, todas las empresas se esfuerzan por obtener lo mejor de sus servicios aprovechando todas las capacidades disponibles. Esto no puede convertirse en realidad sin innovaciones en los productos y servicios de telecomunicaciones.

    Para satisfacer las crecientes necesidades dinámicas, digitales e innovadoras de los clientes, las empresas de telecomunicaciones necesitan expandir su cartera más allá de sus capacidades tradicionales. La digitalización está obligando a las empresas de telecomunicaciones a forjar asociaciones con los nativos digitales para innovar su oferta. Esto también significa que las empresas de telecomunicaciones necesitan una estrategia GTM mucho más rápida para sus servicios que nunca. La asociación es la clave del éxito aquí, ya que las empresas de telecomunicaciones pueden aprovechar la fuerza de sus socios. Las asociaciones no son nuevas para las empresas de telecomunicaciones; de hecho, una asociación es como una estrategia conjunta para la mayoría de los productos y servicios de telecomunicaciones.

    Gestión de una asociación rentable

    Las empresas de telecomunicaciones deben tener una estrategia de asociación para agrupar servicios, salir al mercado más rápido, darse cuenta de los beneficios y continuar innovando. Si bien todos se centran en hacer que la asociación funcione y sea exitosa, el combustible de este modelo no es más que la gestión de ingresos y costos. Si bien no es la verdadera intención de ningún socio cobrar / pagar más / menos de lo que debe ser, es necesario validar, comunicar y resolver cualquier error con el acuerdo. La identificación oportuna de las discrepancias y su resolución sin demoras es fundamental para una asociación sana y rentable. La resolución de discrepancias es, naturalmente, un proceso lento y, si no se aborda en el momento y el ritmo adecuados, puede terminar en una acumulación de facturas pendientes que conduzcan a una asociación poco saludable.

    Por lo general, los analistas leen las facturas, verifican los valores y confirman la exactitud. Sin embargo, cuando se trata de cientos o miles de facturas, existen muchas posibilidades de errores. Las empresas de telecomunicaciones deberían aprovechar la tecnología para abordar este problema de las importaciones y reconciliación de facturas de los transportistas. Se sorprenderá al ver que todavía hay trabajos publicados en el mercado para que los analistas verifiquen tarifas, elementos de costo, verifiquen facturas y realicen conciliaciones manualmente. Tales tareas son propensas a errores y agitadas, lo que eventualmente se reflejará en la calidad del resultado.

    Desafíos con las facturas de socios

    Los socios envían facturas en diferentes formatos, tipos y frecuencias a las empresas de telecomunicaciones. Dichas facturas pueden estar en formato PDF, JPG, imágenes escaneadas u hoja de cálculo.

    Las empresas de telecomunicaciones deben invertir en una solución robusta capaz de manejar todas las variantes de facturas, analizarlas, procesarlas y entregarlas para su conciliación. Invertir en una solución basada en elementos prediseñados puede ayudar hasta cierto punto. Esto funcionará si las facturas que desea procesar encajan en una de sus predefinidas. De lo contrario, es necesario diseñar, desarrollar o capacitar el nuevo diseño de la factura.

    Incluso dentro de las construcciones previas, no es muy fácil encontrar el modelo que se adapte a las necesidades. Los socios pueden seguir cambiando sus formatos de factura como parte de sus mejoras continuas. Las posiciones de campo pueden cambiar o cambiar de lo que fue entrenado. Uno puede construir reglas para discutirlas. Pero la verdadera solución a este problema debería provenir de las técnicas de aprendizaje profundo de AI / ML.

    Deep Learning (aprendizaje profundo): la necesidad del momento

    Las soluciones creadas en base a métodos de detección de objetos convolucionales basados ​​en el Deep learning pueden extraer datos de imágenes escaneadas (OCR), interpretar cualquier tipo de formato de factura, diseños y admitir la traducción de facturas en diferentes idiomas. El modelo profundo de respuesta a preguntas basado en BERT permitirá analizar los valores de los atributos de las facturas con una mayor tasa de éxito. Esto, combinado con NLP, mejora la precisión de la clasificación de los datos de las facturas. Además, busque soluciones que se puedan ampliar según las necesidades; por lo tanto, se puede lograr el paralelismo para mejorar el rendimiento.

    Subex AI Labs es útil para abordar los desafíos de OCR con facturas. Cuatro características clave del módulo de procesamiento de facturas Subex son:

    • Modelo de OCR basado en Deep learning para la extracción de datos de archivos PDF escaneados.
    • Modelo avanzado de PNL basado en BERT para clasificar con precisión los datos de las facturas, resumiendo los datos de las facturas para obtener una mejor representación.
    • Implementación distribuida para admitir el procesamiento de varias facturas simultáneamente en unos pocos minutos.
    • Método convolucional automático para la identificación y verificación del diseño de facturas para clientes conocidos.

    Nuestro módulo de extracción de facturas puede identificar automáticamente los diseños de las facturas y extraer toda la información relevante utilizando las técnicas creadas por expertos mencionadas anteriormente. Impulsamos eficiencias en sus negocios a través de la automatización de procesos para obtener información operativa para respaldar las actividades críticas de toma de decisiones y permitirle lograr una ventaja competitiva. La solución Subex Partner Settlement también le permite introducir servicios innovadores, paquetes de ofertas y productos y manejar la facturación de servicios tradicionales y digitales, abriendo así nuevas corrientes comerciales para modelos complejos de precios variables.

    ¡Interesado en saber más!

    Hablemos ahora

  • Critical Revenue reporting Measures in Telecom Business Assurance

    Critical Revenue reporting Measures in Telecom Business Assurance

    In the highly competitive and rapidly evolving landscape of the telecom industry, telecom operators are expected to expand significantly to bring in optimised profits. In this context, revenue assurance has greatly evolved in its role as a business controller to a business enabler.

    Revenue reporting is a critical area of CSP business. Here, operators face severe challenges when it comes to computing revenue across earned and unearned revenue, compliance, recording of financial data. The operators also find it hard to identify leakages within order to cash processes, voucher generation (electronic and physical), starter kits and handset bundles, pre-provisioning processes (shipment of stock and invoicing of channels), etc., leading to miscalculation, siloed process, inaccuracies in customer collection data or overstatement/understatement of revenue.

    Accounting validations to identify revenue leakage

    The existing sales process across all channels for prepaid and post-paid services comprise crucial configurations and validations. When clearly documented, these validations provide important insights into potential areas of revenue leakages within an operator’s accounting activities. The ‘need to follow’ configurations and validations for prepaid and post services is described below:

     Prepaid validations

    •  End-to-end process validation: Involves computation and validation of the complete process from demand forecasting to voucher generation, voucher transfer, starter kits to warehouse etc. Every single discrepancy observed in this process gets highlighted with the result of the end-to-end support performed in the validation.
    • Stock management review: It helps understand and measure the entire stock management process at the warehouse for recharges as well as starter kits including voucher or roaming recharge packs. If gaps are identified, a mitigation plan can be recommended.
    • Distribution process review: This involves a fulfilment process review for all recharges and starter kit sales. It is majorly recommended to identify issues around distribution
    • Invoicing process validation: This validates the invoicing process for sale of recharges and starter kits. It can highlight issues around invoice quantity, denominations, tax, etc.
    • Account receivables review: This helps understand the account receivables for all vendors and compare this against the bank guarantees to derive the net receivables. Through this, operators can identify those vendors with net receivables that increase bad debt
    • Independent method analysis: This involves independent calculations of liability based on billing methods and compares these with financials to identify discrepancies. It helps operators analyse the product constructs and their impact on deferment of revenue
    • Report and transaction analysis: This includes an independent review of revenue by analysing IT reports, usage transactions, subscriptions, other adjustments, expired vouchers/balances, and deferment logic for monthly rentals. It gives operators a holistic overview of revenue flow and liability.

    Postpaid validations

    • Configuration validation: This validates the configuration of charge code to GL code between Billing system and ERP for accurate integration of automation revenue flow
    • Billable versus billed validation: This facilitates reconciliation to ensure that all subscribers have participated in the invoicing process.
    • Invoice to ERP: Validation of invoice amount billed in the system against the account receivables created in ERP system
    • Unbilled revenue: This validation compares usage and other components against reported numbers. It considers analysis of all the components to bring the result in a form for a detailed comparison.
    • Financial reporting: This provides an analysis of postpaid revenue streams across all billing cycles. It helps to understand and highlight the concerns if they occur.

    Current revenue accounting process have in-built mechanisms to identify discrepancies that may lead to revenue leakages. These mechanisms or validations are based on operator group guidelines, review of amounts credited to subscribers, and revenue realized through usage recorded in charging systems. Careful comparisons between different parameters can alert revenue assurance teams about potential leakages, thereby minimizing variances in the earned and unearned revenue computations across prepaid and billed and unbilled revenue computation for postpaid.

    Learn More on Optimising Revenue Reporting with Subex Business Assurance

    Download Here

  • Redefining Rating Assurance with Analytics

    Redefining Rating Assurance with Analytics

    Significance of Accuracy in Rating Assurance 

    Revenue Assurance has been accepted by the telco world as a critical solution to improve data quality and processes. The solution can help CSPs increase profits, revenues, and cash flows and gain a strong Revenue market share without influencing demand. Globally today, most of the Revenue Assurance controls managed by the RA teams focus on the completeness of data used by business processes focussed towards revenue generation, thereby confirming the undisputable value the function brings to the business.

    However, let’s check some other dimensions relevant to RA. TMForum RA guides show the completeness, validity, accuracy, and timeliness of data quality. To delve deeper into accuracy, a very critical aspect for any RA function; most of the time, accuracy is associated with quantitative measures attached to the charging and rating processes such as: duration accuracy, data volume accuracy, or rating accuracy.

    In this article, we aim to look closer into rating accuracy specifically. According to RAG (Risk Assurance Group), “Data records accurately describe the event or object in the real world that they correspond to.” As a result, the rating accuracy of a tariff applied for a usage event or to a recurring or one-time fee is consistent with the Marketing teams’ intent.

    Limitations of the current approach in Rating Assurance 

    To date, rating accuracy is achieved by using a complex approach that consists of re-performance of rating processes and then comparing the output of the parallel rating with the actual rate. This approach comes with a set of limitations:

    Sample-based method: There is a sample of usage events utilized for a selection of price plans. Though the tried and tested 80/20 rule is a powerful way of cutting through the clutter around a complex decision the outcome remains only as a sub-set of the rating universe tested.

    High efforts required: Configuration of a rate plan, an add-on to a rate plan, maintaining it, etc., are activities which require massive efforts.

    Sampling criteria: It is difficult to identify and arrive at valid sampling criteria, i.e., widely agreed and approved by the company. At the same time, these criteria should be regularly updated to avoid sample bias (providing assurance over the subscribers in the sample).

    Shift from traditional to analytics-based accuracy for rating assurance 

    Due to the above-mentioned limitations, it becomes necessary to identify and execute alternative ways to deal with uncertainties coming from rating accuracy. However, the alternative way has to satisfy several objectives, to be considered effective in validating rating process accuracy. A few of the goals are listed as follows:

    • Ability to reuse existing data:Rated events are already processed, but the usage remains very limited where the central part is only utilizing completeness controls of a RA solution.
    • Ensure complete data usage:The validation is required to be performed for all rated events and for all rate plans. In statistics, this would mean analysing all the population value of measurement, rather than its sample measure only.
    • Perform Statistical measures:Revenue assurance stayed away from statistics for a long while. While sampling is not easy, it shouldn’t be understood that statistics address only decisions under limited data. Measures like mean, median, mode, standard deviation belong to statistics 101 and can be efficiently used to enrich RA’s data.

    Can this analytics approach be accepted as a new Future approach? 

    Definitely, the statistical approach is not new, as it has been used for ages now, to develop quality controls across industries, advance science, and advising on decisions. But fundamentally, the use of statistical controls for validating rating accuracy is proving to be “new.” It may not be a one-size-fits-all approach, and it may not give the same sense of control as that of a re-rating approach however, it shows promising results, particularly for RA teams, which have limited resources to maintain a parallel rating configuration, as required by the standard approach. A few examples of rating accuracy issues can be identified in this way:

    • Different complex implementation ways of tariff and rates from the usual defined
    • Inaccurate international and roaming charges for voice and SMS
    • Mobile data roaming charges applied incorrectly.
    • Unbilled add-ons
    • Free of charge out-of-bundle usage

    Analytics can help solve several sets of the above limitations. Subject matter experts in the area of rating assurance have found that an analytical approach can be very efficient while comparing the limitation of traditional methods.

    The way forward

    In a nutshell, A Revenue Assurance Solution combined with Advanced Analytics, which powers the solution ability and agility to manage and generate insights using large volumes of data, can be a winning approach. This doesn’t mean sunsetting the advanced rating engines from the RA solutions but letting analytics do what they do best: add value from the wealth of data the telco operators are sitting on.

    Download our Point of View to learn more on How AI Can Complement Business Assurance for 5G

    Download Here

  • Customer Protection – More Critical Than Ever

    Customer Protection – More Critical Than Ever

    In August 1992, I bought my first house. The property already had an existing landline, and for convenience, we took the line over and retained the same number. The number associated with the line ended ‘2222’, which at the time we thought was pretty cool until we started receiving calls at 3:00 AM from people trying to order taxis! This was my first real experience of ‘nuisance calls.’ Still, as annoying as getting woken up occasionally by a drunk man or woman ringing the wrong number in the early hours of the morning, at least their intent was quite innocent – they just wanted to get home.

    Through the years, I experienced the ‘waves’ of unsolicited sales and marketing calls from telemarketing companies, which somewhat took the focus off the now occasional ‘taxi call.’ Industry and regulator schemes do try to keep your number private and/or opt-out from these types of calls helped initially, but the perpetrators have modified and evolved their approach, so these types of calls persist.

    30 years on the ‘Telesales’ related calls have now been eclipsed by something far more sinister – ‘the scam call.’ This is not a new phenomenon, and many in the UK will recall the first time they got a call from the ‘Windows Support team’! However, for UK and Europe, these types of calls have been relatively low historically, but they are increasing rapidly, and there are strong parallels with the USA experience.

    In the USA, the term ‘robocalling’ was created to cover the different types of unsolicited, largely automated calls impacting end customers. The issue reached such a level that in 2017 the FCC brought in rules allowing phone companies to block unwanted ‘robocalls’ and encouraged carriers to offer customers more advanced call screening options and solutions. However, in 2018 analysis suggested there were still 4 billion ‘robocalls’ being received per month, with the largest percentage of these calls being ‘scam calls.’ Due to the continued customer complaints and negative publicity generated from the victims of these scams, the FCC brought in further regulations/recommendations that provided carriers with the approval to ‘aggressively’ address ‘robocalling.’ We saw the start of the rollout of initiatives such as STIR/SHAKEN to identify the use of CLI Spoofing – to facilitate the scam. Even with these initiatives, the problem persists.

    In Europe, industry groups & regulators have been looking at this issue for some time, but although discussions continue, in most countries, a solution is some way off.

    If I look at my own experience of ‘scam calls’ up until recently as well as my personal landline, I also had a separate business line in the same property. The lines are provided by different UK operators, but over the last couple of years, they have both demonstrated the same behaviors, namely:

    • Reduction in genuine calls to almost zero. I recently got rid of my business line as it had largely been made redundant by a combination of ‘Teams/Skype/Mobile service.
    • A massive increase in ‘Scam calls’ across both lines.

    Although this article references my experience with my fixed-line services, Mobile phone users are equally vulnerable. SMS provides another channel for organized criminals to scam users through various techniques, e.g., Smishing.

    Given that a ‘silver bullet’ solution is not likely to be available in the near future, it is imperative for European operators to act decisively and proactively to ensure they rapidly implement appropriate solutions and controls to protect their customers. If they do not, they take the significant risk of losing ground to competitors who are seeing enhanced consumer protection as an essential service differentiator, handling increased levels of complaints and customer churn, and having to also deal with the associated negative publicity. There is also an inevitability that as the impact of scam calls increases, we will see increased regulatory pressure/intervention on operators to act anyway, so addressing the issue proactively is the sensible approach.

    At Subex, we have over 25 years of experience working with telecoms operators to proactively mitigate fraud and security risks. Our team has created the Consumer Protection Service to support operators in proactively identifying scam calls and other types of unsolicited behaviors by leveraging protocol analysis and AI/ML techniques on traffic.

    Find out more about our service and to see how we can help you deliver enhanced protection to your customers.

    Schedule Demo

  • Re-inventing the Dealer Management for Telecom

    Re-inventing the Dealer Management for Telecom

    Although digital transformation has impacted almost every aspect of the telecom business, few areas still follow the traditional approach like dealer management. It is prevalent in areas where there is an underserved population, and the digital ecosystem is not so robust. The declining revenue and high cost make it difficult for Telcos to reach out to this segment. In this scenario, the dealer ecosystem or other indirect channels contribute massively to telecom operators’ overall sales.  However, reprobate dealers can exploit dealer agreements’ gaps to claim deceitful commissions by inflating the actual sales volumes. Hence, there is an immediate need to understand the reasons responsible for the revenue leakage and how Telcos can create a robust dealer ecosystem.

    Telecom operators use the offline channels of external agencies and third-party resellers to reach out to last-mile customers and smoothen the game of competitive price war, coalition & innovative customer offering, even in the era of online purchase and service delivery. This requires a convergent dealer management solution that can help Telcos create a robust dealer ecosystem.

    Challenges service providers face with a traditional Dealer Management solution:

    Capabilities in a convergent Dealer Management solution

    Huge data in Simple Compensation Out

    Transformation of a complex multitude of data to be used to derive a simplified compensation value. This includes interfacing and translation capabilities of any data from any Northbound system for Settlement processing and reporting.

    From idea to action

    Enable the SP to engage with its customers, thereby reducing the time-to-market of tailored products and services, offers, value-added services etc. It should create a tiered dealer network with immediate market offer availability from premium retail outlets to informal street vendors and rural locations and use DIY Model.

    Simplify dispute handling 

    Opting for a proactive approach for reconciliation and dispute resolution, gathering data across Network elements, Customer Billing Solution, Provisioning system to identify the issue with data mismatch for the number of records, the rate applied to records, timestamp, and other factors, to identify a dispute before it leads to revenue loss.

    Multi-tenant & SaaS platform 

    Deploying a single convergent solution facilitates features like improved collaboration, Pay as You Go, low maintenance cost with in-built capabilities like scalability, security, backup & restore improvements. A multi-tenant solution to view differentiates and still has single ownership of Partner & Incentive Management.

    Advanced Dealer Visibility 

    Equipping dealers with capabilities like dealer self-registration, dealer scores, performance-based dealer recommendations, request for inventory/place order, track order status, view invoices, statements, credit limit management/digital wallet, and payments to replenish that balance, all in one platform. With GDPR mandating a baseline set of standards, a Dealer Management solution should include comprehensive governance functions such as data security with AES and security of transactions with non-repudiation guarantees.

    To reiterate, Dealer Management needs to balance the existing challenges and future roadmap towards a Trusted, High performing, and Intelligent Solution for SP’s Volatile Dealer Network.

    What Subex Dealer Management solution can offer you:

    Strategic Capabilities

    • SaaS Model/OPEX Based/Modular
    • Multi-tenant
    • Dealer Self-Onboarding & Management
    • DIY Model

    Robust Solution

    • Scalable Business
    • Accurate & Immediate Incentive Calculation
    • Proactive Fraud Detection for all LoBs
    • Open Architecture, Open Source

     

    Schedule a meeting with our subject matter experts to know more about the solution.

    Schedule Now!

    Nakul Mathur
    Nakul Mathur is currently working with BSCG team as a Pre-Sales Consultant and looks after FM and PEM portfolios. His responsibilities include customer requirement gathering, solution designing and demonstrations and generating new business opportunities. He is a post-graduate in management from Symbiosis Institute of Digital and Telecom management with marketing as his majors.
  • Network Transformation: Caution – Extreme Conditions

    Network Transformation: Caution – Extreme Conditions

    Photo: Andy Jacobs
    Photo: Andy Jacobs

    Each day, as every skier knows, can present new conditions and a fresh set of challenges. The weather can be sunny, blustery or confront you with a whiteout due to heavy snow and wind. The slopes might be icy and hard-packed (recalling learning to ski in Eastern Pennsylvania), or you may be blessed with “champagne powder” (which we live for here in the Rocky Mountain West). Poor coverage can expose rocks, bare patches, and other obstacles.

    I took the above photo at one of my favorite Colorado resorts. The “EX” at the top of the sign means this is an extreme expert slope.

    Skier beware — know your limits.

    Right about now you’re probably thinking “Andy, I see where this is going… metaphor ahead!”

    I could get a bit heavy-handed and compare the challenging alpine environment to your network—and chair lifts to your need to manage traffic and capacity constraints. But I’ll keep it simple.

    Recognizing and overcoming obstacles is crucial to the success of any network transformation program.

    Photo: Andy Jacobs
    Photo: Andy Jacobs

    I’ve been part of numerous network transformations over a 30+ year career. I recall when ISDN was going to change our world. What I find most interesting now is the confluence of multiple trends and technology shifts.

    We used to have the luxury to handle one major shift at a time – think PDH to SDH/SONET and then to MPLS and IP networks. On the mobile side, we’ve had the stepwise progression of ‘G’s. Of course, it’s not generally this clean. Each generation doesn’t readily give way to the next, rather there is often an awkward, inefficient, and expensive co-existence among technologies.

    Consider the present era, and the transformational forces underway in the telecom industry.

    Here is a sampling.

    • Dedicated network elements to White boxes and virtualized functions
    • Proprietary code to Open source driven by communities of interest
    • SDH/SONET to IP networks (sometimes with CEM as a stop gap)
    • 2G/3G to 4G (many developing markets)
    • Network performance focus (3G/4G) to Enterprise use case focus (5G)
    • Central offices to Web scale data centers
    • On-premise computing to Cloud & edge computing
    • Manual processes to Closed loop automation enabled by AI
    • Internet for cat videos and Instagram influencers to Internet of Things & smart cities

    I’m sure I’ve missed a few. What would you add to this list?

    Here’s the scary news… the pace at which transformational forces will compete for our attention is only going to accelerate.

    It’s inevitable. Wheels are in motion.

    AI and automation are accelerating the pace of innovation. Companies must embrace innovation to compete and, indeed, survive. Historical transformational cycles of 5-15 years (depending on the technology) will soon appear to be almost continuous.

    Ray Kurzweil, noted author and futurist, addresses this pace of change in his book The Singularity is Near. He argues that the rate of paradigm change is doubling every decade. In 2030, the paradigm shift rate will be 2x what it was in 2020. In 2040, the rate will be 4x, and so on.

    This means if you are stuck in a planning cycle for your next network transformation or moving forward with acknowledged (or unknown) blind spots—a bad situation now could become insurmountable in the future.

    So, let’s get back to those obstacles (moguls?) for transformational network change. What might get in your way? How can you ski smoothly over them without turning into a yard sale? (Sorry, a little ski humor.)

    Falling prey to inertia

    Consider legacy TDM networks. Many large, incumbent operators continue to run these networks despite aging equipment (with difficult to find spare parts), high energy and real estate costs, and a dearth of qualified technicians due to a retiring work force.

    These problems will only be exacerbated with time.

    Better to proactively retire old networks, reap the cost savings, and enable the operational benefits of IP networks and SDNs.

    Not knowing your cost structures

    Every significant move made by a telco requires a business case. One might know in principle that a certain project makes sense, but show me the numbers. Network transformation projects are generally not going to be exempted from such fiscal scrutiny.

    I expect you already have access to certain costs—e.g., real estate, power, field force, Capex for network assets, etc. A best practice is to know your cost structure at a product and services level.

    You’re not in business to build a network. You exist to delight customers with products they need and with competitive rate plans.  If you know what it costs to deliver a service now and after a network transformation, your business case becomes easier.

    The trick is to understand all your cost components. Most of your competitors only take educated guesses (trust me).

    If you get this right, you will have a business insights edge that may be more valuable than your technology edge.

    Examples of things you should know:

    • What granular costs (Opex and Capex) should be assigned to each product and service?
    • What are the margins for each product and service (including network costs)?
    • What is the ROI of each of my sites?
    • What is my return on assets?
    • Which assets are underperforming?
    • If I virtualize certain functions (e.g., onboard VNFs) will I trade Opex pain for Capex savings? This is really a separate discussion but thought I would fold it in. There’s no free lunch. While virtualization is certainly a growing trend, make sure you understand the costs of any operational complexities that arise when you replace the convenience of purpose-built OEM platforms.

    Not knowing your network

    You should have the data to know:

    • What assets do I have in the network (what, where, when, and why)?
    • What is each asset doing? What are the numbers? — Capacity, utilization, history/trends, contribution to revenue, etc.
    • What is my network topology?
    • How are services carried on my network and how do they map to my infrastructure?

    There are many operational parameters you should know about your network which tilt more toward planning, optimization, and fault management. I haven’t listed them here since my context is on preparing for transformation but there is certainly room to consider many other types of data.

    Migrating boxes, not services

    There is a tendency to think about network transformations in terms of technology. For example, you will be replacing SDH/SONET network elements such as DACS and ADMs with routers and IP switches. In fact, many fixed line transformations have taken this approach—one box at a time.

    This is inefficient and can show a lack of empathy for the customer.

    Any transformation carries risk of disrupting customer services. Let’s consider the migration of a TDM network to IP. The recommended best practice for a TDM migration is to swing end-end services. This minimizes the risk of outages since each circuit is touched once during the migration, and the customer is taken into consideration during every phase of planning, execution and testing. This approach can also reduce cost and schedule by executing with an end-to-end view of the network.

    As I write this, much of Colorado is settling in for an epic 2-day snow storm. I’m hoping for another ski day in my near future. But that will mean dealing with heavy skier traffic since everyone here will have the same idea. Alas, I-70 (the highway between Denver and ski areas to the West) is one “network connection” in serious need of transformation!

    Are you prepared for your next transformation program?

    Talk to an expert now!

  • From Default to Debt: Stemming the Flow with Bad Debt Analytics

    From Default to Debt: Stemming the Flow with Bad Debt Analytics

    Telecommunications customer service teams have long been tasked with multiple roles – delivering superior experiences, being savvy with multiple tools, working with bots or assisted automatons, and curating omni-channel interactions to name a few. They are also responsible for ensuring that customers complete payments and clear their dues on time.

    While many solutions are available to enhance the operational efficiency of contact centers, this particular role – managing bad debt – requires a different approach, and here’s why. Globally, the incidents of fraud and defaulters are increasing, which put customer service teams on the backfoot as they struggle to make finance-related decisions. like:

    When late payments morph into bad debts

    The words ‘bad debt’ began as a way for banks and credit card companies to refer to customers who defaulted on their loan or credit card payments. Today, the telecommunications industry borrows this term from finance for postpaid and enterprise customers who have not cleared their dues as per their billing cycle. Bad debt is a real challenge within the telecommunications industry and a contributor to annual revenue losses. Bad debts can be as high as 2% of the total revenue for telecom operators.

    Billing to collection lifecycles in telecom are a bit different from the banking and lending industry. Banking and credit firms often deal with chronic defaulters through one-time settlements or loan restructuring. Telecommunication players, on the other hand, have a different play. Since initial debt recovery falls within the realm of customer service teams, the agents execute gradual actions like sending recurring payment reminders and barring outgoing calls. They often have no way of knowing whether a late payment is a genuine one or whether it is nefarious, leaning towards non-payment. Finally, as the default risk becomes severe, the job is outsourced to an external vendor – a third party collection agency – whose job it is to recover the payment or ‘bad debt’.

    The question is: Is there a way to predict the risk of bad debt way ahead of time?

    Here too, telecom can take a leaf out of the lessons and examples demonstrated by finance. Many global banks and lending agencies are investing in ways to capture 360-degree customer data in a bid to create large datasets that feed intelligent risk scoring models. Such models use atypical customer data to forecast the likelihood of credit risk and credit fraud. These predictions are then leveraged to make smarter decisions about lending, like levying higher interest rates, thereby bringing down the rates of default.

    Bad debt analytics

    An analysis of the data collected during each stage of billing to collections lifecycle will yield key insights. The five main stages are:

    Stage 1: Bill generation

    Stage 2: Pay-by-date

    Stage 3: Early delinquency stage (time of default > 30 days)

    Stage 4: Advanced delinquency or bad debt (time of default is between 60-120 days)

    Stage 5: Collection, disconnection or legal action (time of default > 120 days)

    Analytics models can ingest this lifecycle data to reveal patterns and correlations that provide a probability score of defaulting to bad debt. This entire process is known as bad debt analytics. It focuses on, among other things, minimizing how many customers move from Stage 2 to Stage 3. As seen in the image, stage 3 is the critical stage that separates late payments from defaulters.

    Tools to facilitate early detection of bad debt

    Bad debt analytics are a constellation of modern tools and technologies that can greatly assist customer service teams in making the right decision at the right time when late payment instances arise. It can also provide a host of correlations that inform onboarding about the propensity to default. Some of these tools:

    • Automated data capture – Telecom databases are replete with voluminous customer information that contain key insights. But first, this data lying in disparate systems across sales, marketing, finance, and network teams must be integrated into a single repository. Through automated data capture using connectors and APIs, Telcos can unlock data siloes, mechanizing the job of real-time data integration. This integrated data is what feeds bad debt analytics models for relevant output.
    • Machine learning models – These models accelerate the data-to-insight lifecycle by crunching data, understanding trends, and finding patterns that cannot be perceived by humans. An example is how one of our models uncovered that a majority of the defaulters in the handset sales department of a leading Middle East operator were migrant customers. This insight helped the telco adapt its credit processes to promote appropriate repayment behaviors and lower its bad debt risk.
    • Smart dashboards – Output from the models must be displayed in the right manner to the customer service teams so they can take or recommend the right actions at the right time. For instance, high risk of default in a specific location may indicate poor KYC practices by salespeople. Knowing this, the operator can immediately halt new sales transactions in that area and redirect staff for training.

    Learn how our Analytics Centre of Trust can help your business

    Schedule Demo

  • The True Cost of Insurance Fraud

    The True Cost of Insurance Fraud

    “Identity theft.” “Consumer fraud.” “Insider trading.” “Falsifying data.” 

    Many of us may instantly know what each of these words mean thanks, in part, to the rising numbers and types of fraud that plague customers and corporations, making headlines globally. Over the past two years, fraud has become a growing menace, with an average of 6 frauds being reported per company.

    Nearly every industry incurs some amount of financial loss from external threats or malicious insiders who intentionally defraud organizations through process loopholes. But the insurance industry is particularly susceptible due to a wide canvas that includes many channels, products, and processes for fraudsters to manipulate. Insurers report different types of fraud, including internal fraud, external fraud, underwriting fraud, and claims fraud, each having varying degrees of deceit and preparation. Fraudsters are also constantly evolving their techniques, becoming more and more sophisticated with convincing threat actors that exploit system ambiguities, especially in newer digital channels and networks. More recently, banking, financial services, and insurance (BFSI) companies are seeing a spike in what is being termed as ‘cross-channel’ fraud, whereby hackers steal user credentials from one channel to execute fraud on another channel. But setting aside the higher risk posed by digitalization, even in-person and traditional fraud techniques are advancing and remain a serious threat.

    Drivers of insurance fraud

    As in any other industry, the drivers for insurance fraud boil down to three aspects – pressure, opportunity, and rationalization. People who are overwhelmed by financial pressure may deliberately look for easier ways to make money. Some may find opportunities to derive financial gain through weak links that can be exploited. Others may rationalize padding a claim or exaggerating an incident with the view that they have paid their premiums diligently and yet have never claimed anything to date.

    Insurance scams are executed by individuals or corporations engaging in ‘opportunistic’ or ‘professional’ fraud. Opportunistic fraud is more common, and perhaps a part of human nature often found at the nook where opportunity and rationalization meet. Examples here are inflating a medical bill or falsifying the value of stolen/damaged goods. Professional fraud involves a group of individuals that defraud insurers through schemes like arson-for-profit where owners deliberately set fire to their property to claim their policy or staged auto accidents that entrap unsuspecting motorists into collisions.

    While many consider insurance fraud as a ‘victimless crime’ affecting only insurance giants that can easily stomach the losses, its true impact is far larger than imagined.

    The actual losers of fraud

    • Fraud losses cost insurers steeply.A study of global claims fraud showed that 3-4% of all claims filed are fraudulent. The Coalition Against Insurance Fraud puts the global cost of insurance fraud at USD 80 billion. It is important to note that these losses do not include corporate spending on anti-fraud controls, compliance, and employee training. What is often unknown is that when insurers pay out large sums in fake claims, it weakens their financial position, causing grave consequences to other stakeholders.
    • Policyholders suffer escalating premiums.Theoretically and socially, insurance is a boon. It protects society’s wealth from risk and maintains cash flow despite adverse events. For example, in light of COVID-19, some American auto insurers have actually issued rebates of 15-25% on premium payments of policyholders. However, escalating fraud losses compromise an insurer’s ability to refund gains to stakeholders. To make matters worse, underwriters often increase the price of the insurance products and plans to combat these losses, forcing honest policyholders to bear higher or excess premiums for a reasonable risk. In effect, everybody loses.
    • Fake pay-outs drive organized crime globally.Ill-gained proceeds from insurance fraud can fuel terrorism and organized crime across the world. It also acts as a prime channel for money laundering. In one such case, prosecutors in the state of New York uncovered a massive auto-insurance fraud that cost insurers  millions of dollars. The perpetrators included an outfit of Russian gangsters, doctors, and lawyers that set up fake accident scenes and clinics. Soon after the incident, the New York Senate passed three bills to crackdown on auto-fraud through tougher measures.

    One of the biggest obstacles to combating insurance fraud is the fact that most countries across the globe do not consider insurance fraud as a crime. This means that reporting insurance fraud to a policeman is usually ineffective because law enforcement agencies lack the protocols to investigate insurance fraud. It is no wonder then that the global insurance fraud detection market has been seeing steady growth, accounting for nearly USD 4.1 billion in 2018.

    Faced with the rising frequency and sophistication of fraud coupled with a lagging regulatory pace, it is up to insurers to identify modern, faster, and more effective ways to shield themselves and their stakeholders from fraud.

    Learn more on Insurance Fraud: Building a multi-faceted defense in a risky digital world

    Download the Point of View

  • Midiendo la verdadera rentabilidad: un enfoque holístico y granular

    Midiendo la verdadera rentabilidad: un enfoque holístico y granular

    En el mercado actual de rápida evolución pero sobresaturado, la industria de las telecomunicaciones ha cambiado su forma de trabajar. Ha comenzado a fusionarse o cruzarse con otras industrias, como los medios y los servicios financieros, ofrecer más variantes de productos y servicios personalizados a clientes valiosos, y adoptar rápidamente nuevos avances técnicos y adaptarse a las últimas ofertas, como 5G. Como resultado, el proceso comienza a involucrar a varios departamentos en la estrategia central y la toma de decisiones. Pero cada departamento monitorea su propio conjunto de KPI, que están aislados de los demás. Como resultado, los KPI más fáciles de monitorear como los ingresos, la base de suscriptores, el número de instalaciones y las actualizaciones del sitio tienen prioridad, mientras que el concepto de ganancias y márgenes se ve afectado y, por lo tanto, la responsabilidad de múltiples elementos de costo. Para superar este problema se requiere un ajuste en la estrategia general que implica la colaboración entre departamentos y un cambio en los criterios de éxito del crecimiento de la línea superior (basado en los ingresos) al crecimiento de la línea de fondo (basado en las ganancias) con la presencia de todos los interesados ​​en el área de ganancias y ganancias. Vista de pérdida (P&L) de la empresa, lo que resulta en una única versión de la verdad.

    Enfoque

    Un enfoque holístico y granular es un intento de medir la verdadera rentabilidad de los clientes, productos, activos de red y regiones geográficas mediante la adopción de un enfoque más estructurado y metódico. Todo el proceso se puede dividir en cuatro pasos, como se muestra a continuación.

    1. Comprender: las empresas tienen la mayor parte de la información necesaria para calcular las ganancias, pero se encuentra dispersa en varios sistemas y departamentos. En este paso, se realizan reuniones con diferentes partes interesadas para analizar los modelos comerciales existentes y comprender sus gastos generales de ingresos y costos, la metodología actual de cálculo de márgenes y las fuentes de datos disponibles.

    2. Desarrollar: una vez que se identifican todas las partidas relacionadas con los ingresos y los costes, se desarrolla una lógica

    a. para alinear los elementos en categorías significativas. Por ejemplo: ingresos por uso (dentro / fuera del producto), ingresos por interconexión (nacionales / internacionales), costo de la red (por ejemplo, alquileres / energía), seguros (teléfonos / servicios administrados), etc.

    b. definir el modelo de costos para determinar el costo unitario. Por ejemplo, el costo total del sitio de la red se divide por el uso total del sitio de la red (duración) para derivar un costo por segundo, y luego el costo se multiplica para obtener el costo de la red por transacción. De la misma manera, los ingresos de un suscriptor se asignan de forma inversa a los sitios de la red en la proporción de uso del suscriptor.

    3. Procesar y analizar: una vez que se desarrolla la lógica, se procesan y analizan diferentes elementos de línea de ingresos y costos al nivel granular posible. Se instala el motor de hardware y computación de última generación, que combina múltiples fuentes de diferentes sistemas y perfiladores en ejecución para obtener el margen final de una manera programada.

    4. Calcular: Este paso implica la configuración de un sistema con características de roll-up y slice-dice para derivar el margen final contra diferentes parámetros. Las reglas se definen para incluir o excluir las líneas de pedido según el tipo de nivel. Por ejemplo, el margen se puede calcular con diferentes parámetros como Cliente, Producto, Segmento, Estado, etc., como se muestra a continuación.

    Fig. Ilustración de la vista de pérdidas y ganancias del cálculo del margen.

    Casos de uso

    Algunos de los casos de uso en los que Subex ha entregado utilizando la metodología dada son los siguientes:

    1. Visibilidad de los componentes del costo de la red en la rentabilidad del producto: En el compromiso dado, el equipo de la red quería responsabilidad del equipo del producto al diseñar nuevos productos mientras el equipo de la red se esfuerza por cumplir con los requisitos de la red para los productos. Subex creó una contabilidad detallada de vista de pérdidas y ganancias en todos los componentes de costo de la red para los productos en función de su uso y puntos de precio de corte derivados para los productos que se ofrecen en los servicios ofrecidos. Pudimos identificar algunos productos con altos ingresos por ventas, pero con un margen negativo o poco profundo después de las asignaciones de costos de la red. Todo el proceso trajo más transparencia, buscando la coordinación entre ambos departamentos.

    2. Identificación de clientes altamente rentables: En este compromiso, el cliente quería cambiar los criterios de los clientes más valiosos, de ingresos a ganancias. Subex ayudó al cliente a comprender la contribución exacta que cada cliente hizo al resultado final. El cliente reconoció que el método de ingresos había sido engañoso e inexacto, ya que dos clientes con ingresos similares pueden tener costos variables. Además, ayudó al cliente a identificar los componentes clave del costo, convirtiendo a los clientes de altos ingresos en clientes de bajo margen.

    Conclusión

    En una situación en la que los precios son cada vez más difíciles de intervenir, optimizar las ofertas de costos y servicios se vuelve esencial para impulsar los márgenes y requiere redefinir el criterio del éxito desde los ingresos y las ventas hasta la rentabilidad. La vista de pérdidas y ganancias holística y granular propuesta aumenta la transparencia, lo que genera más información procesable y una mejor comprensión de la estructura de costos, lo que ayuda a identificar los impulsores de la causa raíz.

    Vea cómo Subex Analytics Center of Trust puede ayudar a su empresa.

    Programe una demostración