Partner Feature
Who decides where to make the next 5G upgrade? It’s usually the network engineers. But what if business teams could get involved? What if MNOs could upgrade sites not purely on questions of infrastructure – but on the potential to boost revenue…?
5G is costing mobile network operators a lot of money. According to GSMA estimates, MNOs will invest $1.1 trillion in network infrastructure between 2020 and 2025 – and 80 per cent of that will go on 5G.
That’s just over $800 billion in CAPEX.
In this first phase of the transition MNOs are moving to a ‘non-standalone’ 5G architecture, which anchors 5G radio to the existing 4G LTE network. The non-standalone approach lets carriers roll out 5G without a massive infrastructure overhaul. It’s an interim step before the migration to a fully cloud-native 5G core that will support industrial applications.
Clearly, it will take time for MNOs to monetise next-gen markets such as connected car, smart city and augmented reality that emerge from the high-speed, low latency 5G Core.
But what about today’s non-standalone launches? Can carriers monetise these 5G investments better?
As of Q2 2020, 87 MNOs had already launched 5G services across 39 markets. According to Ericsson’s Mobility Report, there will be 190 million 5G subscribers by the end of this year.
These non-standalone 5G launches bring faster data speeds and greater mobile coverage. Surely this is an opportunity, since some customers will respond to these service improvements by spending more.
But which customer segments? In which regions? And what packages would deliver the best results?
If MNOs knew the answers to these questions, they could make the most of existing deployments – and target their cell site upgrades more productively.
The good news is that they already can.
Lynx Analytics offers a tool – the Customer Happiness Index (CHI) – that lets MNOs link operational metrics to business outcomes.
The CHI ingests data sets about a customer from across the business: number of app downloads, product purchases made, network quality, dropped calls, payment method, number of customer service calls and more.
On a simple dashboard, MNOs can use filters to select a group of subscribers, and display the key metrics for that group. Then, they can apply filters to see what impact a small change of behaviour would make on those key metrics. The CHI uses machine learning to model the results.
For example, how does a rise in dropped calls affect lifetime value? How does the number of customer care calls correlate with NPS score? Does a change of payment method impact churn rate?
Crucially, MNOs can also filter results by 4G and 5G enabled sites and devices.
Which brings us back to the key theme of this paper. In short, MNOs can use the CHI to evaluate the specific impact of 5G on business outcomes. They can then make better decisions about upgrading sites and cells.
A number of MNOs are already applying the CHI in this way. One is Hong Kong Telecom. In most other MNOs, decisions about 5G deployments are driven by engineering and backhaul concerns. HKT was keen to add commercial impact into the mix. To do so, it began using the CHI to guide its 5G investments.
In this paper, we will explain more about the CHI and the results of HKT’s three-year project.
Case study: how Hong Kong Telecom used the CHI to make better decisions about 5G roll out
HKT is a longstanding CHI user. It uses the tool to organize data feeds on customer satisfaction, engagement, complaints and mobility patterns across network technology tiers.
In 2020, the MNO decided to see if the CHI could make its 5G upgrade strategy more effective. Colleagues from engineering, technology, and marketing worked together on the project. Dr. Chung Ng, SVP Technology Strategy and Development at HKT, says the project observed the following process:
Select the group of customers whose network quality was having the biggest impact on NPS, CLTV and predicted churn levels.
Discover which variables (complaints, latency etc) were most affected – and could be mitigated by a 5G site upgrade.
Find out where the targeted customer group was consuming most of the data and voice traffic.
Match these locations to site clusters/towers.
Decide which site upgrade could have the biggest impact on groups.
Upgrade the site.
Follow up with micro-targeted 5G upgrade campaigns offering packages linked to video, gaming, and 5G enabled handsets – and communicating service improvements to customers.
Analyze the impact of these campaigns. Repeat the roll out for lookalike audiences.
The 5G rollout started in April 2020 with the first customer promotions starting in June. Thanks to the highly targeted nature of the campaigns and HKT’s teams best-in-class use of CHI, the results were impressive: double digit take-ups from all contacted customers.
And the improvement was self-perpetuating. By September, HKT could see the impact of its campaigns inside the CHI. The CHI dashboards showed improvements in engagement, satisfaction, ARPU and app usage.
This new data proved valuable. HKT used it to define lookalike audiences for the next batch of site upgrades. In effect, the CHI became a kind of ‘recommendation engine’. It could define the ideal sequence for finding a group of target customers, matching them to cell sites, upgrading the sites and finally devising 5G campaigns.
As a result of this activity, HKT will be among the world’s first to achieve significant 5G coverage in its territory with a significant number of customers adopting the service with 5G-enabled handsets, and HKT can sequence the investments based on the speed of realizing the returns.
Paul Berriman, Group Chief Technology Officer of PCCW/HKT, says: “We always want to stay true to our promise of delivering the best customer experience in Hong Kong. Being fastest to 5G can obviously help us do that. Thanks to the CHI, we’ve been able to find out exactly which customers would benefit most from a service improvement – and know where those customers are. This insight has guided our 5G site upgrades. And the results have been fantastic.”
Customer Happiness Index: how it works
Every MNO is sitting on a vast trove of operational data. These metrics cover all aspects of customer behaviour.
There are hundreds of these diverse data points. Here are some examples:
Quantity of data consumed
Quality such as latency, throughput, packet loss, on different site technologies, down to customer level
Number of calls made, or received from customer service
Mobility patterns
Cross ownership of fiber broadband, household product mix
Dropped calls, reducing in importance, but still important for highest value customers and older customer cohorts
Bill disputes
Payment method type, digitally enabled direct debit, etc
Value-added service subscriptions
Roaming events
Customer support queries
The list goes on.
Meanwhile every operator has key business objectives, which can also be measured. The most important of these are:
Customer lifetime value (CLTV)
Propensity to churn
NPS satisfaction score
Loyalty engagement, campaign take-up, event attendance
The challenge for every operator is to link the two groups – to see what impact changing a specific operational metric (say, latency linked to specific applications) will have on a given business outcome (say, propensity to churn or recontract).
This is what the Customer Happiness Index does.
The CHI uses machine learning to model outcomes. And it displays the results in easy-to-read graphs and tables on a web dashboard, enabling automated selections of audiences whose behaviours are explained by the diverse data points.
Customer Happiness Index: a walk-through
To understand the CHI better, let’s take a hypothetical use case.
Let’s assume our fictional MNO has 8.4m customers, and that the average CLTV is $876.
Now, we want to know what happens to CLTV when we select only the customers who are predicted to have some complaints about network coverage.
We simply add the ‘predicted network complaints’ filter, and we see that 350,000 customers are predicted to complain within the next 12 months. The CLTV for this cohort goes down to $784
Let’s filter even more precisely – by geography. We can look at a colour-coded map to see which region is home to most predicted complainants. Again, we add the filter. Now we see that there are 96,000 customers in this group. The modelled CLTV falls to $633.
Now, let’s model what might happen if we compare this base with customers in the same geography who are not predicted to complain. We can see there are 332,000 customers in this cohort. Their CLTV was $880.
So, now we know that customers in this region who are predicted to complain about some aspect of the network service will spend an estimated $247 less over their lifetime than those that don’t.
The arithmetic is compelling. The collective CLTV of the complainers is around $60m. Turn those unhappy customers into non-complainers and the CLTV jumps to almost $84m. That’s nearly $24m in gains.
Obviously, this is just one example but a verified one. As the CHI constantly retrains its predicted network complainers and defines CLTV potentials based on real data, 5G upgrades offered with various tiers and handset or price plan vouchers are a great way to turn around customer happiness. During the process past grievances are discovered and resolved, verifying some of the predictions.
Eventually, operators can look at the modelled impact on churn and NPS, and CLTV will also correlate with those.
At the end of the project, operators can give every district and microcluster a rank – and organise these districts at cell site or tower level. They can then use this information to inform decisions about 5G roll-out.
Talk to us!
MNOs all over the world are now using the CHI to connect their operational metrics to measurable commercial outcomes.
They are using the insights to make informed decisions – not least about 5G roll-out.
Traditionally, network engineers made most if not all of these decisions. Now, sales, marketing and business development teams can have their say too.
It takes just a few months for an MNO to launch its own CHI, which Lynx will configure to its unique needs. Typically, they see results within weeks of using it.
Learn more about Lynx Analytics data science solutions by visiting their website at https://www.lynxanalytics.com/ [1]
About Lynx Analytics
Headquartered in Singapore, Lynx Analytics is a leader in artificial intelligence and data science solutions. With a strong expertise in predictive analytic models, Lynx Analytics help businesses enhance customer experience and predict outcomes such as churn rates, customer lifetime value, NPS and ARPU.
[1] https://www.lynxanalytics.com/
![\"IT電腦補習](\"http://itteacheritfreelance.hk/test/wordpress/wp-content/uploads/2016/05/logo2-2.png\"){kind=logo}
www.itsec.hk
www.itsec.vip
www.itseceu.uk
Leave a Reply