Cracking the Customer Feedback Code with AI

What to Read Next

1. Shoppers Have Low Tolerance for Errors in Holiday Ecommerce

2. Magnite Expands Audience Activation Solution to India, Southeast Asia

3. Jasper Launches AI Studio

4. Compulse Teams with Placements.io

The six steps you need to win the customer experience challenges with AI consistently.

Companies are still getting a critical business growth tool woefully wrong. Most have no way to truly understand (and react quickly) to customer feedback spanning support inquiries, social media, app/product reviews, surveys, forums, and more. Relying on manual analysis, human tagging, NLP, and survey-centric customer feedback initiatives simply can’t keep pace with the current customer feedback surge—a huge competitive concern in today’s business climate, where consumers’ expectations are high and generally unforgiving. In fact, according to Zendesk, more than six out of ten consumers will stop buying from a company and switch to a competitor after just one poor customer service experience. 

In real-time, advances in artificial intelligence (AI) to categorise and quantify all types of customer feedback (including unstructured feedback from emails, chats, voice, reviews, and social media channels) power better business decisions. Not only does it help improve customer satisfaction by helping brands get more out of their customer data, and it cuts support costs, complaints and issues. 

According to Harvard Business Review, the corresponding cost reductions can be substantial as efforts to improve customer experience can decrease care costs by up to 33%, while American Express has found that consumers are willing to spend 17% more to do business with brands with strong customer service reputations. With US businesses losing $62 billion following bad customer experiences, it is time to reshape how companies address customer experience. 

Forward-thinking brands like Instacart, Thrive, FabFitFun, and Pinterest now rely on AI tools to translate millions of customer feedback data points from various digital sources into easily understandable insights. So, how can you wring what you need out of your customer data to create positive experiences, rectify complex customer issues and save money in the process? Six steps to make sure you can consistently win the customer experience challenge with AI:

1. Create a strong customer feedback loop. When you don’t have a tight customer feedback-product loop or customer feedback-operations loop, your customers notice that you aren’t acting on their feedback. You wind up responding to the loudest customers instead of listening to the slow-burn issues affecting customers most. To strengthen the customer feedback-product loop, Pinterest created a new product roadmap and turned feedback into features. Pinterest’s product team maximised product value and customer experience by quantifying and prioritising the “why” behind customer feedback. Customers crave feeling heard, and Pinterest earned a rousing 93% positive launch rating by basing new features on its AI-driven feedback loop. 

 

1. Break the standard feedback label moulds. Historically, companies have had their support agents label cases. To simplify the task, companies often make these labels high-level to reduce the options for agents. Make sure your machine learning and AI tools can create a custom set of labels for unique data sets and calibrate sentiment analysis to surface missed trends and unify real-time insights across the board and via specific channels; this will make the insight actionable. Knowing negative sentiment associated with the “Login” category is not actionable; however, knowing how many users reset their passwords but did not receive a password reset email tells you exactly how big the problem is to decide whether to take action. With AI labelling, you can make your labels much more specific. 
2. Include humans alongside AI tools. Enable machines and humans to work together in tandem—the result is always much more powerful than anything done in isolation. For example, machine learning instantly recognises the customer’s question and triggers an auto-response which gets more information the agent will need to know to answer the customer’s question. This provides the ability to automate repetitive tasks, optimise large amounts of information and act rapidly and allows human supervisors to be quickly looped in to provide empathy, an extra human touch, or context when needed, especially for new issues. This is especially true with chatbots, which need to be programmed to know when emotional escalations or nuanced questions should be passed to an agent. Technology like chatbots can be detrimental when it gives generic answers that can’t help a customer. AI and humans working together can achieve the end goal of taking care of the customer and freeing up agent time for the most strategic purposes. 
3. Transform qualitative feedback into quantitative data. Leverage automation in place of anecdotal ideas to review all customer interactions. This will transform freeform qualitative feedback into quantitative data. The best AI solutions will look at all interactions from the ground up to assess the root cause of each interaction, identify trends and questions that should be asked, and supplement product usage data with anecdotal behavioural information. This could mean looking at how often features are mentioned and how customers describe their problems. One example is looking at tweets across product launches to identify sentiment. On a larger scale, FabFitFun needed a scalable, data-driven way to translate the voice of the customer cross-functionally. Idiomatic analysed and categorised FabFitFun’s text survey responses and support contacts in real-time—leading to a 250% increase in product satisfaction. 
4. Measure customer support on the right metrics. Faster is not always better in the world of customer experience. The reply time and customer satisfaction are dated metrics that don’t require customer support agents to step outside the box. To incentivise the highest levels of customer service, measuring Net Promoter Score (NPS) and reducing customer churn are more compelling metrics that help customer support teams focus on the bigger picture versus the speed of responses. When teams think of themselves as responsible for the full customer service experience, they can leverage the data they are collecting to make a much bigger influence and become truly responsible for the customers’ experience, not just support. 
5. Use real-time ticket categorisation. When users self-select ticket categorisations, it can lack the precision needed to connect them with specialised agents. AI-driven, real-time ticket categorisation can optimise efficiency and improve the support experience. As an example, Idiomatic’s integration with Zendesk helped Instacart manage a 116% increase in contact volume during the pandemic by analysing and categorising support contacts in real time. By assigning Idiomatic’s customised AI categorisations to tickets in Zendesk, Instacart streamlined support workflows with ticket routing, agent specialisation, and spike notifications. Idiomatic’s real-time insights dashboard empowered Instacart to uncover nuanced customer pain points and make quicker customer-driven changes.

Leveraging AI in customer support helps companies move away from gut instincts and fragmented perspectives to conclude real customer conversations and data. Hard problems are involved in improving customer experience and customer teams’ workflows. AI is helping brands solve these problems and crack the customer feedback code.