Customer Interactions: An Untapped, Overlooked Source of Marketing Insights

A marketer’s job is a pressure cooker. They have to establish awareness, attract attention and stimulate demand for a company’s products and services—all with scant data. They compete with their marketing peers for mindshare in an environment that gets more competitive by the second.

To even scratch the surface, they spend millions of dollars on research, surveys, interviews, and panels to arrive at these loose predictions. Then they use these ‘insights’ to inform ad spend with a hope for a return on investment. And one small misstep can cost their organization hundreds of thousands of dollars. 

But there is one massively underutilized resource right under their noses: customer interactions. Customer support teams talk to thousands of customers every month expressing how they are experiencing your product or service. That’s a goldmine of information coming directly from their end user; the very people your company is built to serve.

Sadly, most companies don’t think about this valuable asset beyond conducting surveys, in large part because customer interactions have historically been the job of CX teams and contact centers.  

In some ways, it’s understandable. Manually reviewing each conversation is a Herculean task, and relying on people to extract actionable data just doesn’t scale. Not only that but deriving valuable information from conversations is a nuanced endeavor, requiring loads of time. The result is marketers don’t even bother to try.

The good news is technological advances in generative AI in the past couple of years have dramatically accelerated the field of conversation intelligence (CI), offering organizations a powerful, viable, tool for deriving valuable information from customer interactions.

LLMs Unboxed

Most marketers are already using large language models (LLMs) to create content via ChatGPT. Applying LLMs to data analysis is just another application of the same technology.

Imagine you could have a “virtual agent” side by side essentially acting as a data analyst. One that could break the barriers of speed and volume processing. One that could quickly synthesize customer sentiment, intent, and nuanced expression just as fast as ChaGPT fires back a response to a query. This is not a work of science fiction; it’s happening now.

Before you ask, no, LLMs are not replacing analysts or marketers. They are a force multiplier, a really adept assistant, if you will, making it possible to review every call and text interaction with unprecedented depth and speed.

So let’s get down to it. This is what you’re missing by not leveraging LLMs in your marketing operations:

You don’t actually know what customers want. Customers write and call with issues and questions every day. It’s part of doing business. But it’s not just about the question. Normally there is much more to it than that: There’s intent. That could be intent to churn. For example, a customer might call and say, “For something so expensive, it shouldn’t break so often.” That could be a missed opportunity to do whatever you can to save that customer, or to simply prepare for the loss.

On a more positive note, a customer might divulge information that could help upsell. For instance, if a customer gives information that they have issues with a product where an upgrade exists (think internet/phone plans) or disclose that they are moving to a new house (and therefore need new service). This is clear intent that could be immediately converted to retention or dollars.

You’re not really ever sure you’re targeting the right audience. Every piece of information a customer mentions can be used to make ad buying more efficient. But how do you know whether your customer is a homeowner versus a renter, single versus married, have children, etc? All of this information could be synced back to lookalike audiences to make Google and Facebook ad spending more efficient, saving millions of dollars.

Understanding target audience plus gauging intent could ensure resources are allocated toward customers with the highest likelihood to upgrade and get ads, not on ones having a bad experience, for example.

You’re not understanding how your customer is engaging with your brand. Customers often contact companies with questions and issues about promotions or discounts, or because they weren’t able to do what they wanted to do on the marketing site. Sometimes there is checkout confusion; it happens. But, again, if you don’t have feedback from the customer, you can improve promotions and your website—which could ultimately to improve conversion rates (and decrease inbound contact rates).

To sum it up, stop hesitating and start acting. If you’re not sure about ROI/purpose, tap into an LLM API and experiment in-house before you buy an existing product. This is actually possible today given the broad availability of open source and proprietary models.

So move past the guessing game. It’s expensive. Even if experimenting costs you a bit at first, it’ll pay off in dividends. You’ll target the right audiences more efficiently, and enhance how customers engage with their brand. You’ll open opportunities for saving customers on the verge of churning and capitalizing on upsell opportunities. You will, for the first time, get to know your customers.

Alex Kvamme is the CEO of Echo AI, a genAI-native conversation intelligence platform. Echo AI helps enterprises discover hidden insights, opportunities, and risks deep within their customer conversations to reduce churn, increase conversions, be compliant, and much more. Kvamme was the cofounder of SeatMe, which was acquired by Yelp in 2013. Kvamme is also an active investor and adviser in Silicon Valley for a number of fast-growing startups.