• 21/12/2023 11:59:57 PM

Maximise dataset efficiency with machine learning: A guide for MVNOs

Two woman working at desk looking at data from their computers

Managing real-time data across multiple systems and locations is a major challenge for any business. In the competitive and data-rich world of MVNOs, it’s particularly hard for MVNOs who are concerned about data quality, accuracy and consistency, as well as value, resourcing, and scalability. In this post, we explore five big questions being asked by MVNOs right now and explain what you can do maximise your dataset efficiency – all without huge investments in labour and upskilling.


1. How can scaling MVNOs better manage multiple large datasets in real-time to make a direct impact on their bottom line?

Most leaders today recognise the value of data and the importance of leveraging it to its fullest potential within their operations. However, many leaders are also fearful that they don’t have the right expertise or resources in-house or the technical capabilities to effectively gather the many data sources that exist coherently. There’s also a genuine concern about the investment it will take to introduce the technology that is needed to make it happen.

The good news is that machine learning and decision intelligence tools now exist that are specifically designed to solve these concerns within the telco sector. They rarely require a real-time flow of all data to support dynamic decision-making and they deliver positive returns almost immediately. Thanks to machine learning, the capability of using behavioural signals to predict future behaviour can be done with smaller data sets and very quickly. 

This means MVNOs can identify the customers likely to churn or the ones who will readily respond to upsell and cross-sell. And, after identifying these groups of customers, they can implement service interactions and marketing journeys that align with those behaviours.

It works because the machine learning capability takes real-time behaviour and combines it with a long-term understanding of the subscriber base. In other words, ‘knowledge’ derived from behavioural data built up over the life of the brand. Using a scoring approach like this means you don’t need 10TB of raw data and a quantum computer to process it to find patterns you can act on. You only need 10G of slowly changing data to identify patterns and inform strategies.


2. What if I don’t know what data I have or what state it’s in?

First-class data quality and consistency are an absolute must for machine learning. However, not necessarily in the way you might think. While it is true that if data quality and completeness are poor across the board it will be difficult for machine learning models to be effective. However, a machine learning model can cope with some gaps and sub-par quality data. The thing it needs is consistent data. Model ‘drift tools’ (tools that can spot a change in the usual pattern) are growing in their capability. But statistical reporting and alerts – the new field of data observability – to catch problems as early as possible is where the future truly lies.


3. How is data observability changing things for MVNOs? 

Imagine if you were able to identify where a spread of data is abnormal – for example, a change in patterns, unusual customer behaviours, or decaying referential integrity across systems. Having the power to pinpoint this would mean you can immediately alert the data and analytics teams to that anomaly, who will then review the potential impact of the change, communicate it to the broader user community, and strategise an appropriate plan of action. Better still, this can be done before the anomaly turns into a bigger, more critical issue. This is the power of data observability.

For example, if you notice a trend where customers are leaving three months after they sign up, you know you have a joiner problem. Similarly, if customers aren’t responding to an offer you’ve put out into the market – before this turns into a critical issue, you can pause it and refine your strategies based on real-time data insights. 

And, at the heart of this, is machine learning. With a good set of historical data, it can decipher daily and seasonal differences, predict them before they appear and then highlight and alert where the issues are. This only needs to be run hourly or daily as any major business or technical issue will be picked up by people within the team before that check happens.


4. What are some of the challenges in visualising this data? 

While the capacity to visualise data is hugely advantageous for MVNOs, challenges do exist. This usually stems from the data literacy of those using the data and the limitations of the business in terms of using and absorbing complex and changing information quickly.  One of the biggest challenges in visualisation is the ability to display the complexity that exists without oversimplifcation – too much simplification doesn’t expose the opportunity, not enough presents a challenge of comprehension. Training on data literacy and the right focus on the explainability is the only way to resolve this challenge.


5. How can machine learning overcome these challenges? 

Imagine this scenario: you ask your team ‘How are we doing on signing up new customers?’. The processes and workflows you have in place will influence the answer to that question. Firstly, is the question, ‘How many people have signed up?’, or is it ‘How many have made payments?’ or is it ‘How many additional people signed up because of a marketing activity?’. And, if you only run reconciliation twice a day, how can you derive a specific and accurate answer? Is it net new customers or just new services? Will the team only be able to know if someone is truly a brand-new customer when the dedupe process runs at 2AM? There are many complexities and layers.

While you could upgrade systems to increase the frequency of reconciliation and improve the matching process at the front end, that would mean a significant cost to the organisation – and for what value? Adding machine learning to this situation to reframe the question, you are able to get a different set of answers. For example:

  • ‘What’s been the average payment rejection percentage for the last week/month/year?’
    If we can see a specific number of new sign-ups we can then make valid predictions on the x% of paid customers.
  • ‘What’s the average percentage of net new customers vs cross-sells for the last week/month/year?’

In these instances, historical real-time data has led up to a pre-set alarm going off to see if there are any identifiable trends. Thus, the model moves from reactive to proactive. This is valuable for any customer-facing business, but particularly for MVNOs for whom churn is a huge market issue. It is an opportunity to implement a 24x7 machine to ‘listen in’ to things it’s told are important and raise an alarm when things go off-kilter. This is decision support and human augmentation at its best. 


The competitive MVNO landscape is constantly evolving, and machine learning is a key tool for overcoming data challenges as well as optimising performance. Telco-specific platforms, such as SourseAI’s Atlas, enable MVNOs to efficiently manage multiple datasets without huge investments in labour or upskilling, as well as detect trends and anomalies using historical data and patterns, uncover actionable insights, rapidly solve issues, and identify new opportunities. In doing so, it empowers them to gain the edge on the competition and power ahead.

Ready to maximise your data efficiency and unleash the full potential of machine learning within your MVNO? Contact the SourseAI team today

Back to Blog