Ride the wave of AI: Tortuous as the road is, the prospect is bright
By: Han Zheng, Vice President of Huawei Software Business Unit, Huawei
The customer relationship is the “golden asset reserve” of the telco. To mine the treasure of customer data is the pivotal motive of CIO and CTO. But how could AI be used in the broad lode to achieve business success? Here is the exploration of both mind and practice from the perspective of BSS & CRM.
AI technology’s promise in ICT is reachable, but not inevitable
As per a McKinsey study and TM Forum’s prediction, AI technology will bring about 1.2% of additional GDP growth a year before 2030. Over 70 percent of companies will adopt at least one form of AI by 2030, and a significant portion of large firms will use a full range of AI technology.
AI could propel the global economic growth as a revolutionary power, just as the electricity and steam engine did before. Gradually available technology of machine learning and deep learning is developing on the base of neural networks which are regarded as one of the most beautiful software paradigms ever invented.
In the conventional approach to design and run the program, IT experts have to give the commands to the computer about what to do, breaking big problems up into many small, precisely defined tasks that the computer can easily perform to calculate the result. By contrast, in an AI problem the idea is not to tell the computer how to solve the problem. Instead, the system learns from the observational data and work out its own solution to the problem. That is why the AI has become the star under the neon lamp and many telco projects have great expectation for its application in telco space.
Machines can learn from data and solve those unsolvable ICT operation issues, which sounds promising. However, in the last decade of the AI boom, many architects and engineers haven’t found how to use AI to solve traditional problems, except for a few specialized problems, such as video recognition. Theoretically, telcos should play a key role in the AI application since its tremendous customer data base. Until now, this promise has already been proven as attainable and it is inevitable. At this stage limited increase of the productivity can be attributed to AI in ICT operation and customer value management.
This dilemma urgently calls for an answer.
From mirage to monetization: Churn prevention and business recommendation
Acquiring a new customer costs 5x more than retaining one. As such, customer retention saves a company money and reduces its marketing expenditure by keeping existing customers who have already interacted with the company’s products and services.
The value of a customers is that they don’t just buy one product or use the service once – they come back again and again for more. Customer retention can increase the customers’ lifetime value and boosts revenue.
To retain the customer, churn prediction is critical for the business management who needs to essentially understand which clients are most likely to cancel a subscription i.e. ‘leave a company’ based on their usage of the service.
Considering the signals from the customers before they decide to leave, are often buried in the noise of overall customer activity. To identify such customer in a short time is extremely hard using the traditional manual way. Preventing a customer from leaving requires an IT team to have a number of advanced signals which are obtained through the careful examination of large volumes of historical data, this is something for which machine learning models are ideally suited.
Some relatively classic techniques have been explored in this domain, such as the use of logistic regression or decision trees, for proactive churn detection may be insensitive to events such as churn which occur with low frequency. Furthermore, an array of innovative AI techniques such as neural networks and gradient boosted trees are being applied and they seem more capable of recognizing the subtle changes in patterns that indicate churn, but require careful configuration and evaluation.
Armed with the AI-driven Churn Prediction, the customer relationship retention still needs more tool from the “swiss army knife” of AI to entice the customer. The next step is just the adoption of the “next best action” (NBA). This is another playground for AI. NBA can be traced back to ODDA, an incredible strategic tool – Observe, Orient, Decide, Act – also nominated as the essence of winning and losing. ODDA and its evolved idea of NBA has been widely accepted. The nations and states around the world and business organizations are using this theory as part of their military or commercial strategy. It has helped the companies thrive in a volatile and highly competitive economy.
In the meaning of telco industry, NBA is a learning system, a method for dealing with uncertainty of customer behavior, and forming a strategy for winning white-hot contests and competitions of ICT market. AI algorithm, such as collaborative filtering, has shown the effective recommendation success rate in this scenario.
Thanks to AI & ML technology, during customer engagement process, IT system can immediately match the customer’s personalized demand and provide exactly suitable solution which meets customer business needs.
“Open Sesame”: Lightweight AI engine is the key
AI implementation in telco IT system can never be done without an AI engine. An AI Engine is a software tool that helps IT team build an AI algorithm and calculation system. It helps to reiterate the tasks that are repetitive and difficult to achieve by the O&M staff. There are a variety of tools that are currently present in the market. However, many of the AI solutions rely on the heavy investment and overly complex deployment which causes more cost than the benefit of applying those engines.
The best engine must be a lightweight, embedded and not-more-and-not-less solution in IT system. What the telco is looking for is an AI engine as part of CRM with constructed and trained churn prediction and NBA.
A good AI engine should be low-cost and useful at the same time. Some critical features are summarized from the experience of Huawei CRM:
- To predict Telco customer behavior hence it is considered for industrial use.
- To create a data model that makes the prediction more effective and fast.
- To interact with the customer relationship and engagement system based on context.
- With the user to select their model of choice.
- With the developers to create their model visually.
- With the tools for consulting and developing services are very helpful
- With the AI algorithm to process the data and easily automate the process.
Since AI for customer value management is a relatively new and exciting realm, there will be a lot more to try and discover. Only one thing can be made sure – there is yet no limit of bring the power of AI to Telco – in the TM Forum and in the foreseeable future.
More information of Huawei BSS & CRM can be found here.