Published: Mar 25, 2022
large-scale personalisation (AI/ML) – what is it, how does it work and why would you use it
Large-scale Personalisation – What is it, how does it work and why would you use it?
Have you ever wondered how large Digital Native Organisations (DNO’s) like Google, Apple, Uber, Airbnb, WeChat, Facebook, Amazon, Weibo, LinkedIn and others seem to read your mind, amazing you with delightful suggestions and new ideas that are just the kind of things that you’re interested in? Sometimes, it’s strange how it works. You may often find yourself thinking, ‘What a coincidence - I was just wondering about that’.
It can be a digital content provider prompting you to watch a movie that appeals to you, or alerting you to an upcoming sports event that you love. Maybe, it’s a social media platform pushing out content that hits the spot – being of the right topic, time and perspective. The list goes on and on – exciting shopping suggestions, perfectly timed entertainment options, a wellness retreat just when you’re feeling burnt out, and many more targeted, attractive and personalised suggestions. Does it work? Do they get you engaged? Yes, quite often – but not always. When the suggestions work, they work extremely well. Even if they don’t motivate you to action, they get you thinking about things you may never have considered before.
Where does personalisation data come from?
What’s going on here? How do DNO’s have the knack for consistently knowing what you need at the right time? The short answer is, they know because we tell them! We reveal huge amounts of information about ourselves – sometimes intentionally and often unintentionally. We tell them what we like and don’t like. They know our schedules and habits, and they know who we talk to, how often and for how long. They know where we go, at what times and what routes we use. Additionally, DNO’s are incredibly aware of interesting things happening in our world – things that we might not know. When they combine their knowledge about our interests, concerns and passions with a vast awareness of what’s happing in our world, they make magic happen by creating an engaging 360-degree picture tailored to each of us.
When we subscribe to any digital service, we give permission for them to collect, store and analyse all kinds of information in order to synthesise a model of us. DNO’s have access to unimaginably large amounts of technology to conduct real-time analysis of the data they collect. As technology costs trend ever cheaper, the DNO’s are enabled to use more technologies to build predictive patterns of individuals using Machine Learning (ML) models in conjunction with other Artificial Intelligence (AI) capabilities. The ML models cycle though iterations of learning until they can accurately provide predictable outcomes for certain situations. They then execute accurate scenario analyses predicting individual future behaviors instantly. Once the ML-predicted behaviour is forecasted, other AI systems suggest the best course of action to sway the situation towards a positive direction for the DNO.
Why collect and analyse data for large-scale personalisation?
Why go to all the trouble of personalising and harmonising subscriber profiles right down to the individual level? It’s done to engage us – to encourage us to interact with DNO’s and remain their customers for longer periods. They do this for several reasons. First, retaining existing customers is often many times cheaper compared to attracting new ones. Second, engaged customers can be targeted with pinpoint accuracy for upsell offers. Third, AI/ML can monitor subscribers and detect usage inflection points caused by changes in their behavior. These points often represent an opportunity to upsell, or identify a subscriber behaving in a way that suggests they are considering leaving the service. Inflections trigger AI processes to analyse the best course of proactive intervention to best engage with the subscriber. If the AI informs us of an upsell opportunity, an automated robot is connected with the subscriber to interact and attempt to close the deal. On the other hand, if the AI tells us that the subscriber is likely to leave (churn) soon, either a robot or an employee may contact the subscriber and utilise retention strategies suggested by the AI. Obviously, organisations need the necessary people and strategies in place to be able to act immediately on the insights that the AI/ML provides.
The benefits of large-scale personalisation
Does this level of large scale personalisation make good business sense? A recent study by McKinsey found that by using AI/ML tools to engage subscribers, it’s possible to increase revenue by up to 10 per cent, while at the same time improving customer engagement by 20 to 30 per cent. So, yes it makes a lot of business sense to have this capability (McKinsey & Company, February 24, 2022). However, not all organisations are progressing on a journey to scale-up their customer-360 projects because they are trapped in old technology siloes that don’t directly provide the data that AI systems need.
Mass-personalisation is applicable to any organisation that has a sizable subscriber base such as utilities, banks and communications service providers (CSP’s). As an example, CSP’s often find themselves in a situation where they have vast amounts of information about their subscribers. However, it’s not stored in a way that allows them to build a 360-degree picture. Indeed, only a small percentage of CSP’s have successfully developed their full 360-degree capabilities and even fewer are using them effectively to intervene, upsell and reduce churn. This is despite the huge commercial upside opportunities being more than enough to fund development of these capabilities.
If mass-personalisation capabilities lead to improved customer satisfaction, revenue and profitability – what options are available to organisations like CSP’s trapped by legacy silo systems? First, AI/ML, cloud storage and compute are standard technologies and readily available to anyone. Further, data scientists can provide the expertise to draw up a customer-360 picture to provide coherent personal subscriber recommendations at scale, ensuring the capability is accessible by everyone. Second, technologies like a data-mesh or data-lake can ingest information from all kinds of systems and transform islands of information into new synthetic representations such as a customer-360 picture. Third, while some organisations may not be collecting full 360-degree subscriber data, this is not an unsolvable problem. Doing something is better than nothing, and a data-mesh or data-lake can be continuously extended over time to build a more complete picture.
The next frontier in personalisation for organisation competing with digital natives
The next frontier in managing customer value over their lifetime is to bring in external data to augment the ML models and AI recommendations. As an example, what would ML and AI predict across your subscriber base if the competition introduced a new service that was superior to your offer? Which subscribers would be impacted, what would be the risk of churn, and what is the best intervention strategy to adopt? Now add regulatory change impacts, international impacts, emerging challenger competitive threats and anything else going on in the world that might get your subscribers considering a change in their service. Market leaders have already conducted these kinds of analyses to build resilient customer relationships, yielding much higher than average revenues, profits and customer engagement.
With great power comes great responsibility
Although this saying about power and responsibility has ancient roots, today it is ever more relevant, often used by figures such as Winston Churchill, Spiderman and many others. Recently, there have been several highly-publicised abuses of power of large-scale personalization. These cases have targeted susceptible cohorts for the purpose of gradually shaping their thinking to intentionally drive behavioral outcomes that may have not otherwise occurred. As an example, much has been written and debated about on efforts to use personalised social media data to shape the outcomes of electoral contests in several geographies. The abuse of AI/ML mass-personalisation for nefarious purposes has raised awareness that organisations that collect and analyse data must uphold their responsibility towards their data donors by using their power to do good. Many social media organisations are constantly reviewing how they utilise their significant AL/ML resources in ways that do not cross ethical boundaries, as should any organisation aiming to deploy this powerful piece of technology.
NCS – the largest Asian based technology developer and system integrator
NCS is a major Asian-based technology services firm with deep experience in technologies required to create and launch a successful customer lifetime value management process. We have deep telco domain knowledge including design skills and advanced analytics experience. NCS’s Telco+ Strategic Business Unit was created to specifically help telcos to become true experience companies, especially for their digitally savvy customers. Talk to us if you are interested in better customer engagement and improved revenues.