The Power of Language

“Language work remains competent without being outstanding – better marks are possible”.

As far as my secondary school English teacher, Mr. Edwards, was concerned, a career built on any level of mastery of the English language was looking unlikely.

I agreed. What use is an understanding of sentence structure and the meaning, in context, of different words anyway? Shakespeare? Chaucer? I was good, thanks.

Numbers were my friend. So, I studied maths and arrived in the world of work with a detailed knowledge of the Greek alphabet but was largely unable to string a sentence together using my own.

But that didn’t matter where I was going, right?

Wrong. It didn’t only take the events of the past year to show me how wrong I was.

Most data and analytics professionals start out plying their trade through the application of coding languages and statistics. These are languages in themselves, of course, and can take many years to master to a professional level, but their beauty and logic only create value when converted to plain English.

Words are a form of data too, yet in our field, we too often downplay their importance in our success.

The far more nuanced, imprecise, and context-dependent skills of storytelling and influencing are where the real impact is created.

It doesn’t just start and end here, though. From our ability to annotate code so that someone entirely unfamiliar with our intent can comprehend it, to the skill required to compose an email conveying difficult news with sensitivity, to the flair needed to convey some game-changing good news to your whole organisation, the right words matter. Who can’t remember an example of where either a particularly good, or poor, choice of words has had a meaningful impact on outcomes for the people that we work with?

Some of the most effective people I know in data and analytics have studied a language to an advanced level, English or otherwise. These skills allow them to innately connect with colleagues outside of our teams on a human level. Similarly, the principles of a considered story structure can elevate our work from ‘just another deck of slides’ to something that unlocks value by engaging hearts and minds.

Many storytelling frameworks exist. One of the better known is Aristotle’s triangle of influence which combines logos (facts and data), ethos (credibility and trust) and pathos (emotion and purpose). His dramatic structure starts with a backstory (establish credibility and a ‘character’ that we care about), then an inciting incident (uh-oh, we have a problem) which creates rising tension to a climax (we must act!), followed by falling tension (we can see a way forwards), resolution (we agree what needs to happen) and a wrap-up. This can work as effectively to create value from data as it does to create box office returns for Hollywood movies.

What does this mean in practice? Well, we can apply the same dramatic structure to the process of delivering value to the business through data.

Stage one is understanding ‘the backstory’. We establish rapport with the  business stakeholder and develop an understanding of their work. In the second stage, ‘the incident’, we work closely with the stakeholders to understand the problem that needs to be solved.

In the third stage, ‘rising tension’, we define the stakes—what do we stand to lose if the problem is not solved? This might include outcomes like lost revenue, additional costs, or damage to the customer experience. Stage four, ‘falling tension’, is when the actual work happens — the data pipeline, model build, or the analysis, for example. In stage five, ‘the resolution’, we restate the problem and its impact. This creates an imperative to act that will unlock the desired outcome. In the final stage, ‘the wrap-up’ we successfully achieve the outcome.

Note that the actual work doesn’t happen until stage four. Stage five (the resolution), and stage six (the wrap-up) form the basis of the story. The ability to tell these stories demonstrates commercial awareness and advances careers in the long run. That is why they are powerful skills to acquire.

The use of words in this way has always mattered but now, more than ever, we need to take language seriously.

The potential of large language models (LLMs) has captivated boardrooms and galvanised the attention and interest of society like few other emerging technology trends in recent times. Whilst opinions differ on their long-term impact, what is undeniable is that the winners in the current generative AI arms race will be those that understand (all) languages, and all of their endless subtleties, the best. Increasingly, this understanding is being democratised to us, the end users, by rewarding the most articulate user prompts with the most useful content.

Isn’t this ability for us to directly engage with this technology, and the words that we choose to put into it, the secret to its success? Why it might yet sustain a level of mainstream popularity that has ultimately eluded other technologies like blockchain and crypto? It circumvents the abstract notions of the data that powers it and brings it into all of our hands through the use of our common language.

I believe that we can all learn from this.

In my day job at the Financial Times, I have the privilege of working with some of the best writers in the world. That I have a ringside seat as the next act of the story of this fascinating technology unfolds is not something that I take for granted.

I hope Mr. Edwards would approve.

*Note: no LLMs were involved in the creation of this article.

This article was featured in the second Edition of our Driven by Data Magazine. You can download the magazine and read more articles like this by clicking here.