Your first day in telecom you learn that the two fundamentals that matter most are preventing churn and adding subscribers.
We’ve been analyzing detailed customer records with machine learning for over seventy million subscribers over the last seven years and have collected insights across a range of scenarios. We learned the rules of engaging customers and leveraged our insights to deliver hundreds of millions in uplift.
In March, we saw a lot of those rules change.
Before March, a subscriber who suddenly spent their afternoon at home probably had just lost their job. This was a leading indicator of credit risk, but meant churn was less likely.
Since March, huge numbers of subscribers spend their afternoons at home, and they’re at serious risk of churn—if their home coverage is not business quality.
That rule changed.
The same way you’re re-thinking your network for the new normal, you need to re-think customer engagement.
With these sudden changes in customer dynamics, intuitions based on old insights aren’t helpful. Old patterns are gone or are outright misleading.
What has worked for us is machine learning, when done right.
In our machine learning work, we immediately saw that old patterns had shifted, and where new patterns were emerging. This wasn’t an accident. Over the past 16 years we have built sophisticated levels of quality control, to ensure that insights were meaningful—from data source integrity to the insights generated. And that level of quality paid off when COVID-19 sent demand and supply shocks into the economy. We were alerted that behaviors across the map had changed and were able to adjust our recommendations in hours.
What we see now is a new customer, one presenting different risks and rewards. And this has raised dozens of new questions—some even questioning long held fundamental truths.
What has changed in the credit-worthiness of your customer, and to what extent are you now willing to risk a post-paid plan? In the pre-COVID-19 world, that shoreline drifted pretty far from the beach—and has come crashing back as consumer credit risks of all kinds have skyrocketed.
Do you know who is likely to add lines as college students on independent plans are consolidated to a single account? Do you know who is at risk for churn because of network problems during critical business hours? Do you know who is unlikely to buy 5G because they’re using wired broadband at home, and will this prevent a handset upgrade when the new iPhone comes out?
These are the types of insights—along with many others—that we track on a subscriber-by-subscriber basis. Knowing this level of detail about your customer—and their new behavior—is critical to maintaining a healthy business.
Current engagement methods are lacking.
I have two phones from two different providers. One sends me weekly messages informing me I won’t be charged for overages due to COVID-19, even though I already have an unlimited plan. Naturally, I’ve tuned them out. This means I have no idea if they’re making a legitimate offer that might entice me to spend.
The other provider has made no effort to contact me, even though I am an obvious candidate for churn. I have virtually no coverage in my home office and cannot make a business quality call. This could easily be resolved with a $100 signal booster. The same company was willing to install boosters in my corporate office, prior to social distancing. I’m gainfully employed with excellent credit and a spotless payment record—so there’s no reason to churn my account. There’s just a lack of coordination, with the marketing team having no idea I’m at risk, because they don’t incorporate call quality into their models.
What about subscribers who aren’t employed, have poor credit, or poor records of paying bills on time? We now ask if the rules of customer engagement have changed enough to question the foundational telecom fundamentals.
Are there subscribers you should allow to churn?
Is every subscriber acquisition worthwhile, given new patterns in credit risk?
Will customers purchase a 5G handset in Q4 if you don’t have 5G in suburbs? Are enough people returning the urban centers where you’ve added capacity?
Is your balance sheet prepared for a massive write-down when unemployed people can’t pay their bill?
Rethinking fundamental industry truths requires reliable, relevant insights into the subscriber, pulled from all corners of your company. That requires machine learning done right and done at scale.
As customer engagement has grown in complexity and the sands are still shifting in the new normal, the need for automated, intelligent insights has never been greater. ElectrifAi has deep experience and pre-built machine learning models that can provide tremendous, proven uplift in a few short weeks.
Join the world’s leading brands and contact ElectrifAi to help you understand your customers, even as your customers’ needs continue to change.