Dr Andy Tiller is vice president of product marketing at AsiaInfo. Here, he tells VanillaPlus about one of the most exciting capabilities of big data analytics: its ability to provide real-time contextual awareness. All the unstructured big data that CSPs generate needs to be non-disruptively turned into structured, actionable context-aware intelligence that they can act upon immediately to the benefit of their customers and partners and, critically, themselves
VanillaPlus: What is real-time contextual awareness, and why are we hearing so much about it these days?
Andy Tiller: One thing we’re seeing is that a lot of CSPs are interested in understanding their customers’ contexts better. Context incorporates many different elements, from static data such as the user’s gender or ARPU to more dynamic data such as location, current balance, and which type of device the customer is using. Increasingly, it’s the dynamic data that CSPs are turning to in order to get a real picture of the user’s context.
Big data allows the dynamic data to be extended to a much greater level of detail, incorporating what a customer is actually doing on their phone right now. An example would be if a customer is watching a video on their phone and their data balance is low. It might then be a good time to offer them a one-day pass for unlimited video access. That’s a simple example where an understanding of the customer’s context enables the CSP to offer the customer something highly relevant to their needs – the right offer at the right time.
It can be a win-win: the CSP gets to at least make the customer happy, and may even be able to charge for something they offer that is relevant to the customer’s context. Much of the drive towards improved understanding of the customer’s context is about improving customer experience. It’s about making customers happy.
VP: How complex is the environment in which a customer’s true context is constructed?
AT: Amidst all the variables involved the important thing is to understand where you can get the data from in real-time. In the video scenario I outlined, you need to know that the customer is actually trying to watch a video right now. You have to interpret the bits and bytes on the mobile data network and convert this into knowledge about what the customer is actually doing.
In addition, you have to access BSS data to see if the customer is nearing the end of their data allowance. You might also want to tap the OSS to find out if you have the network capacity available in that location to support the offer of a day pass of unlimited data for video consumption.
Only if all three factors match – in other words if the context is ideal – is the offer is made.
So there are at least three sources of dynamic data that need to be provided in real-time to support that simple example. We can also see the need for a complex event processing engine that looks out for those triggers and creates the appropriate response. This is what contextual analysis is all about. There are, of course, rules that have to be inserted into the process. The trigger might necessitate an outbound call from a customer care centre to make the offer or trigger an SMS with a click-through link to accept the offer.
Some CSPs already have the technology to do this, but we are at an early stage in understanding how to use these new tools effectively to create good customer experiences.
VP: How is contextual awareness enabled by big data analytics?
AT: There is typically masses of data that you need to analyse in real-time to find the context you are looking for, to find the fleeting moment in which to make a relevant, context-aware offer. Static data remains the same but the user’s location changes and their bill balance alters continually. Even more importantly, what the customer is doing on their phone changes all the time, and that’s where vast amounts of big data must be processed.
Our Veris C3 big data appliance processes a copy of those bits and bytes and converts them in real-time into structured data. So a set of bits and bytes corresponds to a specific user on a Samsung Galaxy S3 who is watching a YouTube video, for example. C3 structures the data and watches out for a pre-defined context to arise.
There is a lot of big data technology in there, including massively parallel and inmemory databases and deep packet inspection technology. All the signaling and user traffic from the mobile data network is processed. That structured data is combined with BSS data to identify context triggers in real time.
The structured data can also be analysed offline later to identify trends and patterns in how customers use the mobile network. That’s also useful for identifying customers’ interests for market segmentation and campaigns. Big data technology is at the heart of enabling this and it is only recently that we have had the technology and the horsepower to make this possible. The early adopters are experimenting with it to see how it can change the world.
VP: What technology is required, and what are the integration points?
AT: The integration points are the data sources that reveal the context. For the mobile data network, optical splitters are used to take a feed of data, non-intrusively from the network. It’s actually raw data collected straight from the pipe. A copy of everything is streamed through the C3 system.
We can identify the user’s location by getting the cell ID from the data network and we’d have another integration point with the BSS. For data sources it’s not a difficult integration job because we’re just taking a copy of data in non-disruptive way. Getting static data from a BSS is very straightforward; you just take a daily import, for example.
The voice network is separate, and typically the OSS systems will provide the source data. One example of a context trigger might be that a user has experienced three dropped calls in the last hour. The CSP can use that information to make an apology and pre-emptively offer some free minutes. To do that you’d want data in real-time from the OSS platform.
VP: What are some example use cases for realtime contextual awareness?
AT: There are three key areas. One is upselling for a CSP that knows the customer context and has something relevant to offer at the right time, such as the video day pass we talked about. Another example could be with a CSP that has a music streaming partner. They can use context-awareness to know when people are listening to music – possibly on a different music app – and make an offer to attract them to their partner’s streaming service. That sort of upselling is a key application area for real-time contextual awareness.
A second area would be in help and support. A poor voice quality experience could lead to an offer of free minutes; or at the moment when a customer changes their phone, the CSP could provide relevant apps, services and help for the new device. Knowing the exact moment is important. CSPs talk about ‘moments of truth’ when the customer is particularly susceptible to being dissatisfied or motivated. If you can delight them at these times you can win their hearts and minds.
A final area is advertising. This relates to upselling but really knowing the user’s context means you can target advertising better and everyone wins. The advertiser gets greater take up, the user only sees relevant advertising and the CSP is able to charge more to advertisers.
AT&T’s sponsored data service allows the data consumed by users watching advertising videos to be paid for by the advertiser, but it’s a blunt instrument – it simply removes a barrier to the customer watching the ad, but doesn’t provide any incentive. A step further would be for the CSP to incorporate context awareness in its pitch to the advertiser. The advertiser could then target the ad to relevant customers at appropriate times. It could also provide a reward to people that watch its advertisements – this time the context you need to watch for becomes whether the customer actually watched the advertisement to the end. If so, the reward is given.
VP: What are the issues around personal data privacy, and how can these be mitigated?
AT: There are two things here: what you’re allowed to do legally and whether it’s socially acceptable in your market. Are you allowed to monitor what a user is doing, and respond to it? For example, with our Veris C3 product you can even know what people are typing into search engines on their phones; China Mobile is using that today to target its own search-based ads ‘over-the-OTT’ (CSP over Google over the network), and it works. China Mobile is getting twice as good a response as Google from doing so.
You’d think this would be illegal in other countries, but it turns out that it’s legal to do that in Europe as well. More things are legal than you might think, but the bigger issue is whether they’re socially acceptable. For instance, if you’re searching for a new car and you get a message from your CSP about a third party service to help you sell your old car, you might find that unsettling or intrusive.
The key thing is to focus on what is socially acceptable in your market and how to create experiences that customers like. To achieve that you need to understand the customer’s context really well. It’s not just about making sure that the offer is relevant to the customer – you also need to be aware that a particular campaign might not work well if a customer is asleep or busy. It has to be delivered at the right time – hence the need for contextual awareness.
VP: What is AsiaInfo’s role here?
AT: Our starting point has been to build the technology platform and package it into the Veris C3 product. C3 is a specific big data appliance which tells you about how people are using the mobile data network. We can deliver C3 as a standalone solution, or we can build broader capabilities which take data from multiple sources (OSS, BSS, the CSP’s website… and others) feed them into a Complex Event Processing engine which watches for context triggers. Events can take place anywhere and we can track them.
The early adopters in China and worldwide are giving us interesting insights into what works and what doesn’t. We’re building up not just the technical platform, but also the experience of how to make big data analytics deliver real value from context-aware actionable intelligence.
