How to build fast, effective feedback loops in an AI-driven world

According to the experience management company Qualtrics, 63% of customers believe companies need to get better at listening to their feedback. This aligns with my experience as both a customer and a marketer.

As a customer, I’m regularly bombarded with requests to “rate my experience” or “provide my feedback.” But as a marketer, I see how rarely this feedback is integrated in a cross-functional way. With the advent of marketing AI, we now have the ability to gather and process vast amounts of customer feedback through surveys, social listening, sentiment analysis, and more. However, many organizations lack the necessary structures, technologies, and roles to quickly and effectively act on these insights.

In this article, we’ll explore what is needed to create fast, efficient, and effective customer feedback loops in an AI-driven world.

(Read the Qualtrics research here.)

The AI-powered feedback explosion

AI has the power to generate more valuable feedback as well as analyze the vast amounts of feedback being created.

On the creation side, AI can make traditional approaches, like surveys, more valuable by infusing them with advanced capabilities. For example, AI-driven surveys can adapt questions in real-time based on customer responses, creating a more personalized experience and yielding more nuanced feedback. AI-powered social listening platforms like Brandwatch and Hootsuite Insights can collect and analyze public conversations to gauge public opinion. The proliferation of chatbots and virtual assistants generates vast amounts of data that can be mined for insights.

Many of the tools that generate this information also analyze it. However, marketers often find themselves overwhelmed by the sheer volume of data they receive daily. This data is often stored in disparate systems, creating silos that are difficult to manage and integrate. Sorting through AI-generated data to find critical insights can become a specialized endeavor in itself. Without a holistic view of the customer feedback landscape, AI has the potential to compound, rather than improve, customer feedback loops.

Identifying the gaps: Structures needed to act on feedback

The same gaps that exist in any marketing organizations can equally impact AI-driven marketing organizations: technology gaps, people gaps and process gaps.

While AI both produces and consumes data, data silos limit marketers’ ability to mine that data for insights and actions. To create meaningful feedback loops with AI, marketers must break down barriers with their IT counterparts. The two groups need to address data silos, identify missing technologies—such as AI-driven analytics platforms, automated tagging and sentiment analysis tools—and work together to narrow those gaps.

On the people side, marketers need to become much more technical. This means upskilling individuals and giving them the time to learn new tools. Every marketer should be able to analyze data, integrate low-code systems, and help bridge gaps between departments — essentially, creating marketing technologists.

As a marketer, I’ve found the more I dive into “no-code” and “low-code” marketing AI tools, the more technical I am forced to become. The tools are advanced enough to seem like they don’t require deep technical skills, but this can be deceptive. Just like WYSIWYG editors allow you to create lightweight assets, AI tools let you create lightweight automations. However, going beyond the basics requires identifying deeper technical skills that may be missing on the team. These skills involve understanding APIs, being able to edit code to debug automations, and knowledge of technical terms. This is the type of upskilling required of marketers in the near-to mid-term future.  

While AI can produce and analyze customer data, if processes aren’t in place for that data to be integrated into rapid testing and learning, customers will continue to feel that companies aren’t listening to them. Use process mapping to identify how learnings are integrated into your teams and spot any gaps that might persist as you start to automate your feedback loops.

Starbucks’ AI-driven feedback system in action

Compared to SaaS companies, coffee might seem fairly low-tech. However, since 2019, Starbucks has used its Deep Brew AI system to manage over 100 million weekly customer interactions in 78 markets worldwide. Deep Brew delivers data-driven coffee experiences, offering personalized recommendations and gathering valuable feedback to further optimize the customer experience.

Deep Brew knows its customers well enough to suggest coffee based on the time of day, weather and ordering history. The “My Starbucks Barista” chatbot allows customers to place orders, ask questions and get drink suggestions using voice commands. Data seamlessly flows from mobile apps to espresso machines and labor management systems, recognizing the interdependencies between a series of rainy days, a nudge to increase customer demand, the need for more staff to serve the increase in customers and a larger supply of coffee beans on hand. All of this is handled through the Deep Brew AI system. 

Here’s the interesting part: By automating routine tasks like ordering, Starbucks baristas can focus more on customer interaction, enhancing the human element of the transaction. In a world starved for connection, Starbucks has dialed in the use of technology to create more seamless and memorable moments for its customers.

Getting started with AI-driven feedback loops

You may not be Starbucks, but you probably have some ways to implement AI to improve your customer experience. With shiny new technologies abounding, shopping for new technology might seem like a great place to start, but that is a red herring. Instead, look at your existing customer feedback sources and pick a single use case — like using the weather to drive coffee recommendations at Starbucks. See how you might implement AI-driven feedback loops around that use case. 

Ask yourself how the data in that use case flows between systems, then work with your technical counterparts to get your data and integration processes in order. What opportunities exist to better collect, clean and store the data? Who needs access to the data? How, what or who could analyze the data in real-time? 

Next, look at your processes for that use case. You are now collecting clean data and (hopefully!) analyzing it in real time. What journey does that data take to influence marketing, product and technical decisions. How can you set up your processes to optimize the flow of this data to the right teams? 

Conway’s Law posits that technology follows the communication structures of an organization. This means that if your communications are siloed, your technology will be siloed. So get your desired communication structures optimized before implementing technology solutions. 

As mentioned previously, look at your teams and ensure you have the skills needed to manage and interpret the data. Taking an incremental approach by managing a single use case helps keep the overhead of upskilling manageable. Allow your marketers the time and training needed to learn a few skills at a time. We’ve been here before with social media — learning on the job, one platform at a time. 

Lastly, but most importantly, use your first use case to help develop and reinforce a culture of learning. Part of the shift to AI-driven insights is acknowledging that, as Bill Bullard said, “Opinion is really the lowest form of knowledge.” The move to AI grants marketers a huge opportunity to move from a battle of opinions to data-driven insights. Use those insights to drive ongoing cycles where cross-functional teams act, measure, and refine experiments, rather than debate merits in endless meetings.

Dig deeper: How to un-silo your organization and be more customer-centric

Truly listening pays dividends

Truly listening to customers, even in seemingly low-tech environments like Starbucks, has the potential to pay huge dividends. But it turns out that listening to customers at scale can be hard. Disparate data sources, an overload of information, and a lack of the right skill sets lead to missed insights and delays in decision-making.

If marketers implement AI technology haphazardly, AI has the potential to compound these issues, rather than solve them. By taking a single use case and looking at it holistically through the lenses of technology, people, and processes, marketers have the opportunity to use a gradual, incremental approach to build fast and effective feedback loops in an AI-driven world.