Preventing Connected-Home Support Calls With Artificial Intelligence

SSupported by cloud service provider DigitalOcean – Try DigitalOcean now and receive a $200 when you create a new account!
Listen to this article

Here’s a math quiz: If there are 1 billion devices operating in today’s connected homes and some of these devices work alone, others in tandem (smart speakers and smart TV), and others in groups of three or more (smart doorbell, surveillance cameras and smartphone app), how many support calls are Internet Service Provider (ISP) Customer Care Centers likely to receive?

The answer is: Far more than they can possibly handle.

Snap, Crackle and Pop

ISP Customer Care staffs are already stretched beyond their breaking point. The support operation is out of control, breaking expense records every month, and not even coming close to coping with the onslaught of support calls emanating from their subscribers’ acquisition and use of multitudes of new devices and services.

Let’s not even concern ourselves with the impossibility of any hypothetical population of ultra-trained Customer Service Reps (CSRs) having access to enough scripts and information to deal with the innumerable permutations of devices working—and not working—together. There are many times more permutations there than dollars in the US national debt!

Sure, ISPs can tell all their subscribers who call for support that they bought their smart devices on their own and, even though the devices are connected to the Internet via the ISP router, subscribers shouldn’t expect the ISP—who never sold them any of this stuff anyway—to be the address for their support needs.

But that is a great way to set new subscriber-churn records.

Here’s a better answer to the math quiz: Artificial Intelligence (AI) can prevent numerous support calls, improving the cost structure and efficiency of the ISP support operation while contributing mightily to subscriber experience and satisfaction.

That’s a great way to reduce churn and cut costs at the same time.

Location, Location, Location

Think of a Netflix movie streaming from cloud servers, through the Internet, into the home through the router, into the air via the WiFi and, finally, to the smart TV and speakers. All those packets of information have to traverse that entire trip to be of any use to the movie viewer. If the stream fails anywhere along that service-delivery chain, the experience becomes miserable and risks turning into yet another support call from a frustrated subscriber.

Artificial Intelligence can prevent that from happening.

The primary task of the AI is to monitor all the devices and services all the time. If the AI can (1) detect a problem as it happens and (2) determine where along the service-delivery chain the problem is occurring, it can immediately decide if the problem is the territorial responsibility of the ISP (router, WiFi) or beyond the scope of the ISP (Netflix cloud service, Internet, or faulty TV).

Detecting and locating problems is the basis for preventing support calls. But the story gets even better.


Upon detection and analysis, the AI is able to repair the problem by itself (in many cases), especially if the problem is related to the WiFi, the main (but not only) territorial responsibility of the ISP. Noticing saturation, interference and other WiFi conditions, the AI can switch channels or take other measures to solve the problem on the spot. The subscriber won’t even notice and won’t think to make a support call.

That’s certainly welcome news to the ISP Customer Care Center. The ISP Marketing Department will be pleased as well. There’s not much risk of churn when problems are fixed automatically, and the superlative subscriber experience is maintained.

Notify, Notify, Notify

If, in real time, the AI detects a problem, automatically locates it along the service-delivery chain, and determines that it is not the fault of the ISP, it can take other actions that also will prevent a service call to the ISP.

Here’s an example. Let’s say the AI determines the problem to be related to the Netflix cloud service—it’s slow at the moment. The AI can instruct the subscriber, via smartphone or voice assistant, to direct her call for help to Netflix and not to the ISP.

If the problem is found to be within a device, that information is also valuable to the subscriber who, once again, will not bother the ISP to solve it for her, but instead, will chase after the device manufacturer or the retail outlet that sold the device to her. That will prevent a call to the ISP while helping the subscriber obtain a proper solution.

But, in the real world, many subscribers will call ISP Customer Care for support no matter what. Millions of them do. What can the AI do in these cases?

Upon detecting a problem, the AI can notify ISP CSRs of its analysis, alerting them to the location of the problem. If the CSR knows that the problem isn’t in the ISP’s realm, he can direct the subscriber where to look. If it is in the ISP’s realm, the AI has already analyzed the problem—located and diagnosed it—for the CSR, eliminating all that time-consuming and irritating dialogue: “Turn off the router, count to ten, and turn it on again. Make sure all the cables are connected, etc.” The AI’s notification arms the CSR, slashing Handling Time of the support call and boosting First Call Resolution.

Now, let’s take a case where the AI has determined that the problem does, indeed, belong to the ISP. For example, the AI’s automatic analysis has determined that the location of the problem is in the WiFi within the home and the diagnosis reveals that there is intermittent neighbor interference. The AI’s notification to the CSR will inform him that an expensive router replacement is not a feasible solution and he won’t order one. He will instead concentrate on an effective solution to the interference, the real root cause of the problem.

AI will prevent millions of support calls to ISP Customer Care.