At the core of every meaningful conversation there is an emotional connection.
Without emotional bonding, there is no flow. People don’t remember what you say, they remember how you made them feel.
Naturally, the words we choose influence the outcomes. One of my favourite quotes from George Bernard Shaw says:
“The single biggest problem in communication is the illusion that it has taken place.“
How often do we struggle to clearly communicate our message to someone? I realised the awesome potential of our tech when someone at a conference, having seen my demo, asked me if our platform can help them analyse what went wrong with their ex partner.
Emotions stir up language and alter conversational intelligence — our ability to communicate clearly and effectively — and we are left to deal with unintended consequences.
Conversational intelligence has been transformed using Artificial Intelligence (AI) and Natural Language Processing (NLP), which will improve exponentially over the next ten years. Those not adopting AI into their business model within five years may never catch up.
In this post, we’re exploring the relationship between sentiment — the emotional engagement and the language we use — with the outcomes. In particular, sales outcomes.
What is sentiment?
According to the Cambridge Dictionary sentiment is “a thought, opinion, or idea based on a feeling about a situation, or a way of thinking about something”. Sentiment analysis has been a popular term in NLP, usually referred to algorithms able to categorise opinions to determine whether the attitude towards a particular topic is positive, negative, or neutral. In very simple terms, negative sentiment would impose frustration, irritation or anger, while positive would typically imply satisfaction or happiness.
However, if we put a hundred people in a room, they will largely disagree on whether a sentence contains negative or positive emotions. We’ve spent years refining these definitions within the domain of AI and contact centres, analysing outcomes to derive what we call a smart language.
What Is the Smart Language?
Smart language is a clear and effective language driving positive outcomes.
The lessons below are derived from analysing over a million phone conversations using Sentient Machines platform in order to understand what is the relationship between sentiment and positive sales outcomes.
Lesson One: Smart Language Drives Sales
It seems that the positivity of language largely impacts the outcomes.
80% of successful sales calls are deemed positive — this means that the agents were employing the smart language, and the reaction found in the language and tone of voice from their clients was very positive. The chart below demonstrates that higher sentiment drives higher sales combining smart language and optimised call length. The secret appears to be identifying the unique smart language for each company and product. One size does not fit all, which is why AI is so powerful.
Negative language tends to be found in longer unsuccessful calls — In fact merely by monitoring the length of calls together with sentiment can improve sales, and quickly remove ‘lost causes’. As James Newell from Clear Sales Message says Don’t chase them, replace them. See also how to optimise time taken to win a deal.
Figure 1: Relationship between Sentiment and Sales Outcomes
Lesson Two: Empathy Doesn’t Help with Sales.
This one has found me by surprise, as empathy is usually one of the highest indicators of the agents good performance. However, more empathy is found in unsuccessful calls than successful calls. Digging deeper, we discovered that most empathic calls are found in cancellations — this is due to the agent showing empathy to the customer’s frustration with the product. In successful sales, there is rarely any frustration (as we’ve seen above), hence often no need to be empathic.
Figure 2: Relationship between agent empathy and successful sales
Lesson Three: Emotional Dynamics and Upward Turn in Sentiment Drives More Sales.
One of the measures we use at Sentient Machines to help with agent training is: Are they able to turn an unhappy customer into a happy one? Are they able to turn a neutral uninterested customer into an excited one? We call this sentiment change. Here is what we discovered in the relationship between sentiment change and driving positive sales outcomes:
Agents improving sentiment sell more. Agents ability to change negative language into positive leads to 36% more sales, however, even more valuable is the skill to turn an impartial customer into an excited one, resulting in 340% more sales. Our platform quickly identifies where agents need additional training to optimise their performance.
No negative sentiment change in unsuccessful calls. If a customer starts the conversation frustrated, and the situation hasn’t improved, we call this ‘negative change’ or ‘Negative to worse’.
Figure 3: Relationship between sentiment change and sales outcomes
Lesson Four: Higher Sentiment Index Drives More Sales
For each conversation our platform analyses, our algorithms attach a Sentiment Index — a score between -1 to 1 — indicating how well the conversation went. This is calculated by monitoring the flow of all participants in the conversation and their use of emotion, behaviour and language dynamics.
We often call this score an Anomaly score, because it also indicates how ‘different’ this call is from the others when it comes to language dynamics and usually can’t be determined by the speakers themselves.
Sentiment Index for successful sales is higher by 24% than in unsuccessful sales — the higher the index, the more successful sales are. Interestingly, a significant difference of almost 13% between retention and successful sales. Digging deeper, we discovered the reason for this: the retaining customers have called, often frustrated, with an intention to cancel and have been retained thanks to well-trained, kind agents who manage to neutralise the caller’s concerns using smart language.
Sentiment Index vs. CSAT — CSAT is a very popular measure and can be very useful for improving your product or service. Arguably, Sentiment Index could be even more objective, indicating transparently the health of your business in real-time. While you can collect customer feedback in a smart way (see a good article here), according to the same article the most common customer response rate is 15%, and in reality they can be as low as 2%. Sentiment Index on the other hand is measured on 100% of the customer interactions.
Figure 4: Relationship between sentiment index and sales outcomes
Lesson Five: Smart Language Drives Not just Sales, but Enhanced Customer Satisfaction.
What’s exciting about NLP is the fact you can mine Smart Language from large datasets. You can define your own unique language that optimises the outcomes you want. Do you want to increase your revenue? Do you want to increase customer satisfaction, or both? By mining the conversations with a high Sentiment Index or/and those that resulted in more sales, you could discover what use of language works for your audience.
Therefore, defining the Smart Language depends largely on your desired outcomes. The Smart Language should correlate with your objectives maximising desired outcomes. For example, depending on who your customers are, you can adjust your vocabulary and tone of voice to theirs, helping to create an instant rapport with clientele as well as reflecting the appropriate image for the brand. The language you use with Millennials will be ever so slightly different to that used with Generation X, but the difference in outcome will be significant in terms of sales performance. AI knowledge is power in an age where rapid evolution drives rapid growth.
Do you want to improve your own customer satisfaction and sales utilising AI-derived smart language?
Book a free demo today to learn how to use AI to quickly drive sales, improve customer satisfaction, reduce costs, and leverage new unseen opportunities. Seamlessly plugging into your existing telephony system, utilising the latest AI breakthroughs — no obligation, no commitment, and no knowledge of AI required.