AI copywriting: The opportunity and the bull$#!%

Whether it’s LinkedIn posts, politics, or the latest reality TV show, one thing seems to be true: middle-of-the-road, sober opinions and views don’t get the clicks. Unfortunately, that’s what this series is all about, so I had to at least add an edgy title.

This is the first in a series that will explore different aspects of AI application—where there is opportunity today, where I believe it will emerge tomorrow, and where to look out for people slinging BS.

You will learn about the current state of AI in copywriting, the benefits it offers, the hurdles it faces, and realistic steps for leveraging AI without falling for the hype.

What AI copywriting can do NOW (and it’s not just hype)

AI isn’t writing your brand’s manifesto yet, but it’s damn good at cranking out copy that sells (sometimes). We’re talking about clear, benefit-focused copy that’s consistent in tone, optimized for SEO, and usually even respects your brand guidelines. 

Think of it as your tireless intern who never complains about writing another description of a pair of socks.

Concise, persuasive product descriptions 

AI can generate product descriptions at scale, focusing on benefits, tone consistency, and SEO optimization. Less time spent agonizing over descriptions, and more time focused on strategy.

• Companies to watch: Copysmith, Describly, and even some AI-powered Shopify apps are tackling this problem head-on.
• DIY approach: The DIY approach here isn’t too shabby either, fine-tuning an open-source model to better understand your brand’s value proposition and voice can help drive more differentiated compelling copy. 
• Things to watch out for:  You’ll still need to take a look at the descriptions generated, a set-it-and-forget approach will likely lead to unlucky hallucinations showing up in some of the descriptions (especially if the descriptions are fairly long)

Dynamic email subject lines and body copy

We all know that crafting a unique email for every single customer on your list isn’t exactly practical. But guess what? You don’t have to. AI can help you move beyond generic blasts by tailoring subject lines and email content to specific customer segments. We’re talking about groups based on purchase history, browsing behavior, demographics – whatever data points matter most to your business. It’s not quite one-to-one personalization, but it’s a heck of a lot better than “Dear Valued Customer.”

• Companies making it happen (and how they’re helping you win): Cordial, and honestly every mainstream ESP. But head on over to Cordial AI to get a glimpse as to why we do it best 🙂 
• DIY approach: That good old tab over to your favorite GPT4/Gemini etc, given the nature of this workflow it still leads itself to the age-old business process of copy and paste.
• Things to watch out for: 1:1 personalization of large send sizes can still be cost and time-prohibitive. Going straight to the customer with LLM-generated content can have risks, the shorter the content, the less risk (transactional SMS is a great place to start).

Smart ad copy variations

A/B testing is the bread and butter of digital marketing, and AI is here to make it even tastier. Let AI generate multiple versions of your ad copy – different headlines, calls to action, you name it – then let the data tell you what works best. It’s like having a whole team of copywriters working round the clock, without the payroll.

• Companies to watch: Anyword, AdCreative.ai, and Google’s own Responsive Search Ads.
• Why this matters: You can’t afford to waste ad spend on copy that doesn’t resonate. AI helps you find the winning formula, so you can put your budget to work where it matters most.

Real-world examples (because we like proof, not promises)

• Stitch Fix: They’re using AI to personalize product descriptions and style recommendations.
• eBay: Their AI-powered email campaigns have shown significant improvements in open rates and click-throughs.
• Cosabella: This lingerie brand leveraged AI to generate ad copy variations, resulting in a 38% increase in click-through rates.

Behind the curtain: How AI is crafting your copy (and what marketers need to know)

You don’t need a PhD in computer science to understand how AI copywriting works, but a little background knowledge can go a long way. Here’s the gist, in terms even a CMO can understand:

Large Language Models (LLMs): The brains of the operation

At the heart of AI copywriting are large language models (LLMs). These are massive AI systems trained on gargantuan amounts of text data – think books, articles, websites, the whole enchilada. By analyzing patterns in language, LLMs learn grammar, syntax, even some semblance of “style.” But remember, they’re not sentient beings. They’re sophisticated pattern-matching machines.

If you’re curious about what LLMs are the “best”,  this leaderboard is a great answer.

Context windows: The AI’s short-term memory

LLMs don’t process entire novels in one go. They have a “context window,” a limited amount of text they can consider at a time. Imagine it like a sliding window that moves along the text. The larger the context window, the more information the AI can “remember” while generating copy. But even with large windows, there’s a limit to how much context the AI can truly grasp.

Fine-tuning: Teaching AI your brand’s voice like a well-trained parrot

Out-of-the-box LLMs are generalists. They can write a decent email or product description, but they won’t sound like your brand. That’s where fine-tuning comes in. By feeding the AI specific examples of your brand’s copy – marketing materials, past (performant) emails, even internal style guides – you can teach it to mimic your unique voice and tone. It’s like training a parrot, but with fewer crackers and more algorithms.

Hallucinations: When AI gets a little too creative 

You might have heard the term “hallucination” thrown around in the AI world. It’s not as spooky as it sounds, but it’s something to be aware of. In AI terms, a hallucination is when the model generates output that’s factually incorrect, nonsensical, or just plain bizarre. Think of it as the AI equivalent of a brain fart.  

Why do hallucinations happen? 

Hallucinations occur for several reasons:

• Insufficient training data: If the AI model hasn’t been exposed to enough high-quality data, it might struggle to understand the nuances of language and generate accurate responses.
• Ambiguous prompts: If your instructions are vague or open-ended, the AI might fill in the gaps with its own “creative” interpretation, which could lead to hallucinations.
• Over-reliance on patterns: LLMs are excellent at recognizing patterns in language, but sometimes they get a little too eager and apply those patterns even when they don’t make sense in the current context.

Where to be wary:

• Factual accuracy: If you’re using AI to generate product descriptions, marketing copy, or any other content that requires factual accuracy, be sure to double-check the output for any hallucinations. A small error could damage your brand’s credibility or even lead to legal issues.
• Brand voice: Hallucinations can also undermine your brand’s voice and tone. If the AI starts generating wacky or off-brand content, it could confuse or alienate your audience.  
• Sensitive topics: Be extra cautious when using AI for topics that require cultural sensitivity or ethical considerations. Hallucinations in these areas could lead to offensive or harmful content.

When hallucinations can be a feature (embrace the weird):

• Creative storytelling: If you’re looking to generate creative story ideas or brainstorm out-of-the-box marketing campaigns, a little bit of AI-generated weirdness can actually be a good thing. Hallucinations can spark unexpected ideas and lead to truly original content.
• Humor and wordplay: Want to inject some humor into your copy? Let the AI loose and see what it comes up with. A few well-placed hallucinations could lead to some genuinely funny puns or unexpected turns of phrase.

The bottom line is to use your judgment:

Hallucinations are a natural part of the AI copywriting landscape. The key is to understand when they’re a problem and when they can be an asset. By using your judgment and carefully reviewing AI-generated content, you can harness the power of AI while minimizing the risk of embarrassing mistakes or unintended consequences.

Other technical tidbits (for the truly curious):

• Transformer architecture: The underlying architecture of most LLMs is called a transformer. It’s a neural network design that’s particularly good at understanding relationships between words and phrases, even in long sequences of text.
• Tokens: LLMs don’t process words directly. They break text down into smaller units called tokens. These can be whole words, parts of words, or even punctuation marks.
• Attention mechanisms: Transformers use attention mechanisms to weigh the importance of different parts of the input text when generating output. This helps the AI focus on the most relevant context and create more coherent copy.

What’s on the near-term horizon (No SkyNet just yet)

Hold onto your hats, folks, because AI copywriting is about to get a whole lot more interesting. We’re not talking about flying cars or robot butlers (yet), but the advancements on the horizon are seriously game-changing for retail marketers. Here’s what’s coming down the pike:

Hyper-personalized copy at scale 

Remember that dream of truly unique copy for every customer? It’s not as far off as you might think. As AI models become more sophisticated and data collection gets even more granular, we’re moving towards a world where every email, every product description, every ad can be tailored to an individual’s preferences, purchase history, and even mood that day. Imagine a world where “Dear Valued Customer” is replaced with copy that feels like it was written by a personal shopper who knows you better than you know yourself.

• Why this matters: This isn’t just about being fancy; it’s about driving serious results. Studies have shown that personalized emails can increase open rates by 26% and click-through rates by a whopping 14%. Imagine the impact that level of personalization could have across all your marketing channels.

Natural language generation (NLG) advancements (AI that writes like a human, almost)

Let’s be honest, some AI-generated copy still sounds a bit… robotic. But that’s changing fast. Advancements in NLG are making AI-generated copy more creative, nuanced, and downright human-sounding. We’re talking about witty product descriptions, email subject lines that make you chuckle, and ad copy that tugs at your heartstrings.

• Why this matters: Authenticity is key to building strong customer relationships. As AI gets better at mimicking human writing, it opens up a whole new world of possibilities for engaging and delighting your audience.

AI-powered storytelling: Where data meets narrative

Imagine AI that doesn’t just write product descriptions or ad copy, but crafts entire brand stories, weaving together data-driven insights with compelling narratives. This could involve analyzing customer reviews, social media conversations, and even cultural trends to create stories that resonate deeply with your audience.

• Why this matters: Storytelling is a powerful tool for building brand loyalty and emotional connections with customers. AI-powered storytelling could take this to the next level, generating narratives that are both personalized and relevant to each individual.

Real-time copy optimization: The never-ending A/B test

What if your marketing copy could constantly evolve and adapt based on real-time customer interactions? Imagine AI that continuously analyzes how customers respond to different headlines, calls to action, and product descriptions, then automatically tweaks the copy to maximize engagement and conversions.

• Why this matters: Real-time optimization could eliminate the need for manual A/B testing and ensure that your copy is always performing at its peak. It’s like having a 24/7 copywriting team that’s constantly learning and improving.

What’s overhyped

Alright, let’s get real. AI copywriting is powerful, but it’s not magic. Some of the claims you hear are just plain ridiculous. Let’s cut through the BS and set the record straight:

AI as a complete copywriter replacement 

Here’s a reality check: AI isn’t coming for your copywriter’s job. Sure, it can handle the grunt work, but it can’t replicate the nuanced creativity, strategic thinking, and brand understanding that a skilled human brings to the table. AI is a tool, not a replacement. Think of it as your trusty sidekick, not the superhero.

• Why this hype is dangerous: Believing this myth could lead to disastrous results. You might end up with bland, generic copy that lacks the spark and personality that makes your brand stand out. Or worse, you could alienate your customers with tone-deaf messaging that misses the mark entirely.

The “perfect” AI-generated copy 

Don’t be fooled by the marketing fluff – AI-generated copy isn’t perfect. It can be formulaic, lack originality, and even perpetuate harmful biases if the training data isn’t carefully curated. And let’s not forget the importance of brand voice. Even the most sophisticated AI can’t capture the nuances of your brand’s unique personality without significant fine-tuning and human oversight.

• Why this hype is misleading: Expecting perfection from AI sets you up for disappointment. Instead, focus on using AI to augment your human copywriters, not replace them. Let AI handle the repetitive tasks, freeing up your team to focus on the big picture, the creative concepts, and the strategic messaging that truly connects with your audience.

The BS meter (because some things are just silly)

Alright, let’s have some fun. AI is impressive, but it’s not about to take over the world (or your marketing department). Some claims out there are so outlandish, they deserve a good laugh (and a healthy dose of skepticism):

Sentient AI copywriters 

Let’s be clear: AI doesn’t have feelings, consciousness, or a soul. It can’t experience joy, sadness, or the creative spark that drives truly original work. So, when someone tells you that AI is “just as good as a human,” or even better, just smile and nod. They’re either trying to sell you something or they’ve watched too many sci-fi movies.

AI that can “predict the future of marketing” 

AI can analyze trends and make predictions based on data, but it can’t predict the future with certainty. The marketing landscape is constantly evolving, and human intuition and adaptability will always be essential.

AI as a replacement for all human creativity 

Sure, AI can generate catchy headlines and persuasive product descriptions. But can it craft a brand story that tugs at your heartstrings? Can it write a Super Bowl commercial that goes viral? Not without a whole lot of human help. Creativity is a uniquely human trait, and it’s not something that can be easily replicated by algorithms.

The bottom line: Don’t be a sucker

It’s easy to get caught up in the hype surrounding AI, but don’t let it cloud your judgment. Be critical of the claims you hear, ask tough questions, and don’t be afraid to call out the BS. By understanding the limitations of AI, you can make informed decisions about how to use it effectively in your marketing strategy. Conclusion: The Future of AI Copywriting is Bright and it can reduce time within the organization significantly today, but it’s not foolproof.

So, there you have it – the good, the bad, and the downright ugly of AI copywriting. We’ve explored what it can do today, what’s on the horizon, and what’s just plain BS.

The bottom line? AI copywriting is a game-changer for retail marketers. It can help you:

• Save time and resources: Automate repetitive tasks and free up your team to focus on higher-level strategy.
• Personalize at scale: Deliver more relevant and engaging content to your customers.
• Boost conversions: Craft persuasive copy that drives results.

A philosophical aside 

AI is a mirror reflecting our past. It learns from what we’ve created, a master mimic of human expression. Yet, humans are not merely echoes of yesterday. We are the architects of tomorrow, the weavers of dreams yet undreamt.

Our creativity isn’t just a remix of old tunes. It’s a symphony of the unseen, a melody composed for a future that doesn’t yet exist. AI can analyze, predict, and optimize, but it cannot conjure the spark of true innovation. It cannot fathom the depths of human emotion that fuel our greatest creations.

In the age of AI, remember this:

Human creativity is the compass pointing towards a future only we can imagine.

AI is the mirror reflecting the brilliance of our past.

Let AI be our tool, our collaborator, our muse. But never forget that we are the ones who will shape the world to come.

Next up: AI and images – the visual revolution

In our next installment, we’ll dive into the world of AI-generated images. We’ll explore how AI is transforming the way we create and use visuals in retail marketing, from product photos to social media graphics. Get ready to see how AI is reshaping the visual landscape and learn how you can use this technology to captivate your audience and drive sales.

Stay tuned, folks!