We are used to seeing updates from companies that compile crowd-sourced data related to mobile network performance. The likes of Ookla, OpenSignal and Tutela release regular country reports and comparisons, with headline-grabbing stats on who is the fastest, best for video and so on.
Perhaps it is less well known that these companies also sell analyses and slices of their data to mobile operators, often for benchmarking purposes. Operators buy more detailed reports than the public versions to see how they are doing compared to their competitors, and to cross-check what they think they know about their own performance.
These app and analytics vendors themselves have slightly different methodologies. Something like Ookla relies on a user actively installing an app and then opening up the app to run a speed test or to ping the network. OpenSignal is mostly accessed via a direct app download (OpenSignal has a small number of app partners) but it runs in the background, monitoring certain metrics such as signal strength and data throughput. Tutela also runs in the background, but it doesn’t require users to download a Tutela app – instead it pays developers of popular apps and games to include a small amount of code in their app, and this becomes the reporting mechanism.
There are other companies such as P3 and RootMetrics by IHS that conduct their own more structured walk and drive tests, sometimes with a complementary crowd-sourced element. Then you have companies such as Metricell which have a very operator-centric analytics platform. Plus of course, operators have a host of network monitoring tools themselves, allied to an increasing capability to deploy active test agents across the network that can simulate and emulate device interactions and measure KPIs. The data from these instruments and tools is fed into data analytics platforms to expose poorly performing apps, services or devices or geographical areas.
You might call all of this a route towards “customer experience visibility” – in that the goal is not only to describe the network state, but to have a more graded view of whether customers have the latency, throughput and access they need to achieve whatever it is they are trying to do at that moment in time. This means you need to know if you have the coverage, capacity, throughput and latency required to deliver a certain app as the user accesses it. That is what you really need to know, not just whether a cell is showing as a green light on the screen in the NOC or SOC. That avoids a situation where the technical team is telling the CEO that everything is great, while the CMO says that customer satisfaction scores are troubling, and the CFO doesn’t know where to invest next.
When you have that customer-level understanding then you can then take action. That might indeed mean taking corrective action in the NOC, but it might also entail doing something creative within the marketing or customer care departments. For example, if you can also tie those data analytics outputs to your CRM then happy days, because you can start to see which customer segments are getting poorly served, where they are, how important they are to you, and what you then want to do about it.
Perhaps the most important thing to note is that there is no one approach here that can give operators the whole picture. Instead, they need to take samples from each method and build up an overall view. Gaining as near a real-time view as possible of customer experience, integrated with customer care, marketing and network operations, has been something of a holy grail in telecoms. If you could deliver it, you could really start to differentiate and add customer value beyond what your customers can deliver, both for business and consumer customers, but also for content and app partners.
A new entrant arrives
So, in this crowded market, is there room for another player? Netradar, a Finnish start-up, says yes. It says that because it claims it has developed a new method, based on years of research, to enable mobile operators to assess the actual user experience of their customers when they are using specific apps and services.
Netradar is a privately-owned startup that was founded in late 2017, with the aim of commercialising research work originally carried out at Aalto University. It has developed software that it offers via an SDK to operators to install on the applications they install on customer devices. This reporting software is allied to a back-end platform running Netradar’s analytics algorithms, and is marketed to operators as Netradar Suite.
With its first customers acquired in early 2018, Netradar claims its solutions are in use by multiple customers internationally, including Tier-1 operator groups. In a series of blog posts and articles on its website, it outlines how operators could use such insight to plan and deploy 5G network rollouts and develop services, and to uncover issues related to mobile use of cloud services, for example.
Chief Research Officer and co-founder Jukka Manner, also a Professor at Aalto University where he is Director of Communications and Networking, answered a few questions from TMN about what Netradar does differently, what benefits it can bring to operators and how its technology works.
Manner outlines the issues with existing technology that Netradar is addressing. He says that as most network quality tests are active speed tests, the tests themselves introduce a subjectivity. That’s because users tend to run a test either when they want to show off a very high speed, or to prove a poor experience. Or some customers might not want to run too many tests because they use up a lot of data. As for those tests that run in the background, Manner contends, “Such tests are not tied to situations where people really would need data.”
And passive tests – such as probes in the network – do not tell you why the consumer is getting a certain bit rate, just that they are. “Is it because the network is congested, or the cloud server is only sending at a slow rate, or something else” Manner asks.
We can also tie the data to applications and services and can distinguish network performance for different mobile apps. This allows a totally new angle to understanding consumers’ network and app experience
So what is Netradar doing differently?
“Our fundamental research challenge many years ago was the following: ‘Can we measure consumers’ network quality without killing their data plans and incur costs, and get information 24/7 while they use data?’ Pretty much an impossible task, right? It took us many years of algorithm development and measurements, but we finally got it right,” Manner claims.
Manner says that Netradar uses a hybrid technology where the consumer’s own Internet traffic is used as the measurement payload, with Netradar adding simple latency measurements and pings.
“Say a user is only getting 5Mbps throughput. With our algorithms and data analysis we can identify if the network is congested or if the user in a bad radio environment, is the issue with the cloud service itself, or is it that a particular device model, OS or even a single device is broken, or we can ask if this speed is actually normal for this particular app?”
Manner says that as Netradar analyses consumers’ own data traffic, it gets a full understanding of consumer needs and how networks support them, all day long.
“We can also tie the data to applications and services and can distinguish network performance for different mobile apps. This allows a totally new angle to understanding consumers’ network and app experience, and how to enhance that.”
How it works
So how does it work? Netradar’s client-side technology is a small SDK than can be included in Android apps. It sleeps when there is nothing happening on the radio, and wakes up when an app starts up. That traffic is then analysed for performance, and reported from time to time to the database. “The whole data gathering is extremely lightweight, in terms of data usage and battery consumption,” Manner adds.
If you compare this to other approaches, it means measurements are being made when an app is currently active, and are specific to the app experience itself.
And this look at data and app traffic can be tied to operators’ network monitoring systems to understand root causes more deeply. So, say Netradar automatically finds a problematic area in a cellular network where many users have issues, an operator can then invoke its own network tools to investigate and correct.
Operator engagement
Unlike some of the other players, which go direct to market and then share data with the operators, Netradar prefers to engage privately with operators.
Manner: “Typically operators take our SDK, put it in any number of their own apps and start collecting data on the network quality. So they don’t need to have yet another app to push to their customers, they can simply update their current apps and are ready. Our SDK can tie the performance to all apps on the device, so for example we can see how well Netflix works in an operator’s network without having our technology in the Netflix app itself.
“We bring in our software and install it on the operator’s own servers. Operators are free to choose their hardware and operating system and we can even push the data to an existing database service, big data platform, or use Google Big Query.”
“As the data is the operator’s, it can take the raw data and combine it with existing sources, like network planning and customer support. That is what most of our customers do, use our dashboards and AI engines, but also do their own data analysis with existing tools and systems.”
Netradar thinks it has something different in a crowded market, something that operators can add to their capabilities as they continue to search for the customer experience holy grail.
