Can AI change how you listen?

AI is the talk of the town. Thanks to ChatGPT and Bard, AI has gone mainstream, and if media headlines are to be believed, it will soon take everyone's job. There again, they said the same thing about computers.

Looking beyond the Pepsi Challenge

So let's peel away the hype and scaremongering and look at the reality for marketers. Starting with some simple definitions. While ChatGPT and Bard are rapidly becoming the Coke and Pepsi of AI, the category is much much broader.

Some aspects of AI have been used in business for several years. For example machine learning is a branch of AI that teaches machines to instantly perform thousands of repetitive tasks. Usually ones that humans would rather not do.

Machine learning is not as cool as it once was, but it's robust and predictable. It's like milk or tea, if I'm going to keep the drinks analogy going.

The exciting fizzy drinks segment of AI is large language models (LLMs). LLMs like GPT4 and Bard are constantly learning. They are constantly drawing on vast data sets - basically reading lots and lots of text - and then using what they've learned to create new things. That's pretty cool. So for example, it could write you a new blog post on direct marketing because it's read millions of other blog posts on direct marketing.

AI for marketers

As a marketer, how can you put AI to work? Is it just a quick way to write bland content or produce slightly odd 'photographs'? Or does it offer practical value?

For me, the day-to-day value of AI is speed and efficiency. It's very good at accelerating the scale, speed or consistency of repetitive manual work. That can include:

• Generating ideas for marketing content
• Creating more engaging ways to answer repetitive customer queries
• Analysing large volumes of text data

We're all familiar with talking to chatbots on websites. While a quick search on Twitter will bring up 100s of threads on how to use ChatGPT to write content. So I won't go there, beyond saying that LLMs create content based on what already exists. So unless your prompts include a clear point of view on a topic, you'll get the same beige content as everyone else. That's why a new job title is rapidly appearing - Prompt Engineer - with eye watering salary numbers!

AI-powered text analysis

The use of AI I'm most interested in is how it replaces manual data tagging. Picture the scene. Your customer survey closes and you receive the spreadsheet full of verbatim comments.

You make yourself a strong coffee/tea/whiskey and settle down at your screen. Scrolling through the lines of text, you start looking for familiar themes or topics, tagging the text as you see it. Aside from the time and energy you're wasting, about half way down the list you start noticing a new theme. Is it new, or has it taken a few instances for you to notice it? The only way to be sure is to go back to the top and start again...

The big problem with manual text analysis, is that its manual. We humans can only keep a small number of topics and concepts in our conscious mind at once. In contrast, natural language processing (NLP) algorithms can instantly search for millions of concepts in large volumes of text.

Aside from the time saving, this has a powerful impact on the quality of your insights. NLP algorithms are constantly looking for everything within their scope. That means if they don't find something it's because it's not there. If humans don't find something in a big block of text, it could just mean we missed it or were looking for other topics.

How AI improves client listening

Imagine that your firm is trying to stand out in the market for its innovation, approachability and expertise. You conduct some key account interviews and send a survey to the wider client base. This results in a long list of verbatim comments. Quick as a flash, the algorithms process the text. They highlight all the positive comments about how approachable your firm is and how the sector expertise of your teams is shining through.

So far, so good. But, when it comes to your innovative thinking and approach, tumbleweed.

It's not that clients have complained about it, or said you're not very innovative. They've not mentioned it at all. If a human was analysing the text, this would likely get missed. We naturally look for positives and negatives, and spot instances of familiar words. So no complaints must mean everyone loves us!

I call this the Sound of Silence (I heard a lot of Simon & Garfunkel growing up). With consistent text analysis, silence becomes a signal. Why aren't your clients talking about your innovative approach? The two most common explanations are:

• Your clients don't actually care about innovation. They're here for your expertise, approachability or perhaps how quickly everyone responds to them.
• Your firm isn't as distinctive as you would like. The levels of innovation are no different from competitors, or the client's expectations. Hence your approach is not standing out to them.

Making insights actionable

Either way, your firm is spending time and money on it's brand and client experience and not getting the cut-through and returns desired. Me-too is a dangerous place to be when budgets are being cut. When clients get asked for testimonials they will mention your expertise and approachability, but not your innovation. Which makes it hard to win a pitch where that's a key criteria. Worse, when they are asked to recommend an innovative firm to a peer, or they look at their panel for a specific piece of work, they will go elsewhere.

Using AI to automatically analyse text comments can bring your firm three benefits:

• Stronger relationships. When AI analysis is combined with year-round listening, you quickly spot friction. The experiences and touch points that aren't meeting your clients' expectations.
• Stronger reputations. AI text analysis lets you discover the sound of silence. The areas where fee-earners aren't standing out to their clients.
• Growing revenues. McKinsey recently published an article on the importance of experience-led growth. In a nutshell this means driving incremental revenues from existing clients. This comes from discovering and responding to client needs faster than your competitors.

Automating text analysis is step 3 on the Feedback Flywheel by MyCustomerLens | always-on listening . If you would like to know more about how our clients are using the Feedback Flywheel to strengthen relationships, reputations and revenues, please drop me a line.