sponsored

IoT and AI: Taking Customer Satisfaction Measurement to New Levels

istock 615833522
Credit: i Stock

Customer satisfaction is the true measurement of business success, right? After all happy customers are more loyal; they buy more stuff. They even spread the word, and positive word-of-mouth is everything. So what can you do to measure that all-important KPI?

There are several methods currently in use, including net promoter scores (NPS), customer effort scores, and customer satisfaction scores (CSAT). Typically these approaches are fed by customer surveys—quite often the results are too little and too late. But what would happen if we disrupted customer satisfaction measurement using the Internet of Things (IoT) and artificial intelligence (AI)?

The first step to this new disruptive approach is to have customers opt in and provide the data streams that could be fed into an IoT Complex Event Processor (CEP engine), which is a collection of technology components that can process millions of events from mobile devices, connected products, website clicks, social media posts, and pretty much any type of message that can be generated by computers. All of these events can be pushed into an AI engine and it can learn from the massive amounts of data as it is pre-processed and passed into the CEP engine.

Standard AI techniques can be used to interpret text messages from the consumer and/or the event streams. AI is also great at data comparison and duplicate elimination — for example, what is similar about these messages? These pictures? What is different? What does a happy customer look like? What does a stream of messages that resulted in a positive outcome look like?

As the filtered and augmented information such as customer lifetime value, their service level agreement, or even their geographic location, gets passed to the CEP engine, it will act as a “state machine” to switch between conditions like good experience, bad experience and neutral experience. Each state in the CEP’s state machine could act on the entire stream of data mentioned above and intelligently invoke additional AI techniques. For example, if the customer is in a bad experience state, what next best activity should be offered to move that person into a neutral experience state, or even a good experience state?

With this approach, companies could, at a glance, see how many customers are having a good experience at any given moment. A marketing campaign could be launched by a company and they could see immediate feedback on the success of the campaign. When combined with activities such as purchasing a product, real-time feedback could also influence salespeople, store clerks, and any customer-facing employees.

If you use a state-based approach you could measure customer satisfaction by providing a higher score the longer the customer spends in good experience state vs. neutral experience state or bad experience state. You could focus on specific activities from the data streams based on which state the customer was in. For example, if they are in a bad experience state, focus on their social feeds and make sure that someone is responsive to angry tweets, for example.

If AI is used as a preprocessor to a CEP, items triggering a good experience state could be suggested as recommended actions for the next customer purchasing a product.

For example, Acme Motorcycles (a hypothetical electric motorcycle manufacturer) discovered via AI that if it could get a prospective customer into a showroom and get them on a test drive then there was strong chance they would purchase the bike. Velocity data sent from the motorcycle test-drive experience in real time to the dealer and then pushed into the CEP engine indicated that if a customer drove the bike at over 50 MPH then the likelihood of purchase increased 10%. When the customer returns from the test drive, if the dealer’s analysis dashboard shows that the customer is in “neutral” state, then the dealer would be provided with a recommendation to advise the customer to go back on the road and try the bike on the highway.

When the customer journey ends successfully (the customer buys the bike) all of the new data gathered during this latest customer’s journey, like the route the customer drove on the highway that led to the positive experience, will be fed back into the AI engine to make more recommendations for the next customer to help repeat the happy experience.

This example was a very simple use case with only three states. Imagine if you added more states? For example, the customer was on a product journey and gathered this information and this resulted in a purchase and a happy customer. This would allow companies to accumulate purchases for each customer and gauge their satisfaction as they went through the entire product lifecycle. Were they happy in the information-gathering phase of the customer journey? Did their experience “go south” after they bought the product and tried to get customer support?

Imagine how powerful it would be to be able to have this information:

“40% of our customers get angry enough to want their money back if their device is reporting error code 134A and they have to wait longer than 2 hours to get a phone call.”

Within each state of the CEP system, you could trigger activities and generate analytics to show the entire customer journey instrumented for not only customer satisfaction but also customer lifetime value.

In the near future, combining AI methods with massive customer data streams that include clickstream data and connected devices data can yield extremely valuable information about the satisfaction of your customers. The days of guessing about customer satisfaction and forcing our customers to endure customer satisfaction surveys will be over.

Want more tips? Get inspired by these 12 IT trailblazers who are leading their whole business forward by building on the Salesforce platform.

Related:
NEW! Download the State of the CIO 2017 report