A hand extending from a computer screen writing on a stack of papers

Like many others, we’ve been observing AI’s progression over the past few years, exploring new developments and now witnessing the explosion of ChatGPT. Rather than duplicate existing content, this blog focuses on what ChatGPT means to B2B marketing. More specifically, it looks at the effect AI is having on B2B marketing and whether it is ready to be applied in meaningful ways.

It’s important to note:

  1. We don’t believe ChatGPT will replace all humans. 
  2. We also don’t think it should be ignored. 

Like other technologies, it has great potential (if you use it correctly). An iPhone, for example, is fantastic for a lot of things. But to use it as a hammer? Not so much. 

TL;DR: While ChatGPT is fascinating, it’s not quite there yet. But it highlights an even deeper problem we often experience in the marketing world: quantity versus quality.

Why are we hearing so much about ChatGPT?

ChatGPT and many other large language models (LLMs) give natural language responses to questions and in turn have the potential to change everything! It’s easy to imagine all the possibilities when we see them at work. It takes seconds to write an inquiry before your screen is filled with seemingly relevant text right before your eyes. It’s easy, convenient and a bit addictive when you begin to test the limits of what it can and can’t do.

New tech launches every day. So why is ChatGPT special?

Consider the time it takes to adopt a new platform: Netflix took 3.5 years to reach 1 million users; Instagram took 2.5 months. ChatGPT took just 5 days — now making it the fastest “platform” ever to do so. In the short time since its launch, ChatGPT is already being used in myriad ways across countless industries, from law practices to recipes. 

Venture capitalists and corporations (like Microsoft, who has invested $13B so far) are throwing money at the growing number of AI models. And it seems there’s something new practically every hour.

Everyone is rushing to stake their claim in an application in hopes that it will be the next Alphabet, Meta or Amazon. Even key personnel from founding organizations are jumping into new AI ventures. 

All of this suggests that LLMs will continue to improve and change the way we do things.

Two challenges with ChatGPT: input and output

Knowing how to build an LLM inquiry is critical and in itself an exercise in good writing. Usually referred to as a “prompt,” it takes practice. Getting it right can make all the difference. As we know from human nature, there are infinite ways to interpret even the simplest instructions. And when using ChatGPT specifically, each input nuance can create a different result. Even the order in which you add prompt elements can change the output. Writing prompts has become a full-time (and well-paying) job. 

Strategists, writers and art directors are uniquely suited for this. They have years of practice describing precisely what they need from teams, gathering input for concise direction, and developing clear rationale for their work. In essence, they are creating prompts. To craft a good prompt, a writer must have a clear vision for their expected output. Then, to achieve that vision, you must provide context and be as clear, specific and detailed as possible. You may be thinking, “Well, if you can do all of that, why do we need ChatGPT?”

ChatGPT’s outputs magnify the issues we consistently see as marketers. It’s great at quickly generating streams of generic filler, but it has the same issues as any other shortcut. It’s like having the ability to quickly microwave fish. You CAN, but it ruins the fish (and makes you unpopular in the office). And more is not better, either; ask your customers. They don’t want more stuff. But they WILL respond positively to the right messaging at just the right time, delivered in a compelling and emotionally engaging way.

Great uses of ChatGPT

  • Offer a contrarian point of view for positioning and claims
  • Generate lists of potential challenges for your brand
    • So you can be ready to answer them
  • Provide a synopsis of meeting notes with next steps and key points
  • Create a first draft to edit
    • We all know it’s sometimes easier to respond and edit versus creating something new

So, what’s not to love?

It’s important to remember that we’re still using ChatGPT in its beta release phase. The creators and developers do not intend for this to be its final iteration. Instead, the platform continues to evolve and change as it responds to more prompts. ChatGPT still struggles with “hallucinations,” accuracy, math and biases, but these are all being updated as we speak. ChatGPT is based on GPT-3, and GPT-4 (which is rumored to have exponentially more functionality) is set to release soon. Others are creating “tests” to check if outputs are generated by AI or a human. Questions also arise about what “original” means, including the intellectual property legality of using AI to write in the style of someone else. 

In addition, its popularity has caused the ChatGPT website often to give you only a “we’re at capacity” message. They’re testing a paid version now that will allow people faster access. And if you didn’t suspect that the times are changing, a waiting list sign-up was posted on a founder’s Twitter account and has quickly gained popularity.

ChatGPT has magnified the need for creativity

Creative is a key variable in making your marketing successful, because your programs are only as good as the creative you feed into them. If you don’t make campaigns that convert, you won’t be successful. And you won’t convert unless you can engage your audience. Creative is quickly becoming one of the only things that can catch someone’s attention. When the wave of generic content generated by ChatGPT hits, you’ll want to stand apart. While ChatGPT has the potential to become a valuable tool and resource, it’s just not there YET. Below is a demonstration of why this is our belief.

Our ChatGPT experiment

We entered this company definition (their designation redacted to protect the innocent). Like many corporate definitions, it isn’t focused on the customer.

“At [redacted], our power to engineer solutions is the result of combining local expertise with a global team of the best that the industry has to offer. Our employees run the full gamut of processing technology and experience. Their expertise covers: [redacted], process control, and life-cycle management. We have built a team that can deliver the entire spectrum of services that the [redacted] industry can offer.”

We then asked ChatGPT to rewrite it, emphasizing the customer benefit, but in the same voice:

"At our company, we understand the importance of creating solutions that not only meet, but exceed your expectations. That's why our team of experts is dedicated to using their deep understanding of the industry to engineer innovative solutions that will take your business to the next level. With a unique blend of local expertise and industry-leading professionals, we are able to provide a comprehensive range of services, [redacted]. And with our focus on process control and life-cycle management, we can ensure that your products will be of the highest quality, and stand the test of time. We know that choosing the right partner is a big decision, let us show you the difference working with true professionals can make.”

This attempt was too long and still more from the company’s perspective than the customer’s. So I asked it to rewrite it again with instructions to make it shorter and more compelling for a buyer. Interestingly, it took far longer to generate this copy, which is what it’s like to write shorter copy IRL:

"Unlock the full potential of your business with our unparalleled expertise. Our team of local and industry-leading experts engineer innovative solutions to exceed your expectations. From design to delivery, we ensure the highest quality for your products. Experience the difference working with true professionals can make. Partner with us today.”

This time, ChatGPT acted as a good tool for knowing what not to write. It performed as asked and improved the copy. But the algorithm selected the most used phrases to technically answer my query. This collection of jargon lacks creativity and emotional appeal. The resulting copy is so overused and generic, it reads as meaningless noise, providing no differentiated reasons to work with them.