.svg)
As a kid in the late 1970s, wary of getting bullied by the sixth graders on the school bus, my mother shared one of the hoary cliches of childhood. Sticks and stones may break your bones, but words can never hurt you. While it helped me back then, how out of touch the sentiment seems today when words are, often quite rightly, seen as dangerous weapons. They can wound deeply, hurting more than mere sticks and stones. When used well, words can also change how you see the world.
That’s why communicators and marketers like our firm spend a lot of time carefully choosing words to explain what matters to consumers, businesspeople, and shareholders. Whether to define a brand, sell a product, or manage a crisis, we use research, intuition, and experience to create narratives and put together memorable phrases. Yet strangely, at most companies and agencies, the most technical and data-driven component of word choice—search engine optimization (SEO)--often gets pushed to analytics teams and is not a primary consideration.
Does this make sense?
Shouldn’t strategists spend more time thinking about specific word choices? Not just because everyone now acknowledges the power of narratives. But because machine learning is now defining which narratives win via context-driven search algorithms and sentiment analyzing LLMs.
Words, in fact, matter more in the Age of AI. Here are some key facts to consider:
—Most public companies have powerful associations with only a handful of emotion-laden words in Google search.
—Wikipedia, arguably a crowd-sourced equivalent to the LLM’s scraping and summary process, is likely the primary source for ChatGTP’s prompt responses.
—Sentiment analysis, a foundational tool used to rank content by search algorithms and LLMs, has long been used by top hedge funds and asset managers to deliver alpha, i.e., above market returns.
In our experience, these crucial facts and trends tend to fall through the cracks at most companies and agencies, lost between big announcements (earnings) and mundane marketing (keyword buys). With the evolution from the multiple but generally identifiable factors influencing Google search to the black box of LLMs, we get the temptation to throw in the towel and rely solely on human experience and intuition. But it’s a losing path.
Machines don’t read, write, and score content the way humans do. They analyze content in context using tools to understand each word’s relation to other words and their relative importance, i.e., their sentiment. With the rise of LLMs, brands and businesses have a current window to put sentiment analysis and other tools to use. Done correctly, these tools can help smart teams choose the rights words in the right context to shape how machine learning understands their narratives. Others who ignore this opportunity will likely face the hard and perhaps impossible task of trying to change the LLM’s “minds” later.
So, while sticks and stones may break your bones is great advice for dealing with cowardly childhood bullies, when it comes to the digital world, words can hurt you very much indeed.
