The Words and Phrases to Use — and to Avoid — When Talking to Customers

The key to any successful relationship is effective communication. In the business world, this means trying to understand what consumers and clients are saying, and responding to them in ways that reflect that understanding. For the most part, however, the way businesses have used language to persuade, satisfy, or rectify has been more art than science.

The retail world in particular abounds with catch-phrases, habits, and commonly copied templates: “Say it with a smile.” “Never say no.” “Sorry is a magic word.” “A person’s own name is the sweetest sound in any language.” But do these and other long-held tips about how to speak to customers really work?

Studying the effectiveness of the words businesses use to talk to customers is tricky, but the rise of digital communications, social media, and big data is producing massive amounts of text that researchers can analyze and interpret using sophisticated new techniques. By combining natural language processing, computational linguistics, and psychology experiments, we can now uncover the true importance of subtle variations in how customers talk to front-line employees, and how customers respond to the words chosen by those employees. This allows us to understand the ways people communicate in business settings with growing precision, and what language is most effective.

It is now clear, according to our research and that of others, that some of the time-honored truths of customer service interactions fail to hold up to scientific scrutiny. You can, for example, say “sorry” to a customer too many times. Even if you’re a member of the company’s team, it is often better to say “I” than “we.” And not every piece of communication needs to be perfect; sometimes, a few mistakes produces a better result than flawlessness.

Here is the latest on the fast-growing, insightful, and sometimes surprising new world of business language research.

Be Human…

The body of research analyzing language use among customers, and between employees and customers, suggests a personal touch is indeed crucial. This is particularly important given the growing frequency of conversations that happen via technology (the phone, email, text, or chats) rather than in-person. Here are a few tips based on the latest research:

Speak as an individual, not part of a team. While companies and employees believe they should refer to themselves as “we” when talking to customers, and actually do so in practice, our research shows this practice is less than ideal. In a series of controlled studies, company representatives who referred to themselves in the singular voice (e.g., “I”, “me”, or “my”) were perceived to be acting and feeling more on behalf of customers than those who adopted less personal plural pronouns (“we” or “our”). For instance, saying “How can I help you?” outperforms “How can we help you?”). For one company, an analysis of over a thousand email interactions with customers found that switching to first person singular pronouns could lead to a potential sales increase of over 7%.

Share the same words. People who mimic the language of the person they’re interacting with are trusted and liked more, whether this mimicry entails how they talk (pronouns like “I” or “we,” articles like “it” or “a”) or what they talk about (nouns like “car,” verbs like “drive,” adjectives like “fast”). For example, in response to a customer inquiry such as “Will my shipment arrive soon?” an agent would be better off saying “Yes, your shipment will arrive tomorrow,” rather than “Yes, it’s being delivered tomorrow.” Employees’ linguistic mimicry creates affiliation with the customer, and research in progress by Francisco Villarroel Ordenes, Dhruv Grewal, Lauren Grewal, and Panagiotis Sarantopoulos has linked mimicry to customer satisfaction. Rapport can also be created by asking employees to imagine the customer as similar to themselves (e.g., shared background, personal, or business interests), even when they may be thousands of miles apart.

First, relate. Researchers performing automated text analysis of hundreds of airline customer service transcripts found that, consistent with consumer self-reports in prior research, expressing empathy and caring through “relational” words was critical, at least in the first (opening) part of service interactions. Relational words are verbs and adverbs that demonstrate concern (e.g., please, thank you, sorry) as well as signal agreement (e.g., yes, uh huh, okay). While this finding may not seem surprising, what may be for some is that front-line employees shouldn’t necessarily offer a caring, empathetic touch over the entirety of the interaction.

… And Then Take Charge

While using words that establish a more personal rapport with customers is important out of the gate, a new trove of research suggests that the assumed importance of front-line “empathizers” may be limited. More sophisticated analysis of the language of customer interactions suggests that once they’ve shown they’re listening, front-line employees should quickly shift gears towards language that signals a more assertive, “take charge” attitude.

Move from relating to solving. The same research that examined airline check-in service transcripts found that after an initial period in which the employee demonstrates their empathy for the customer’s needs, hearing employees say “sorry” and other “relating” words had little effect on customer satisfaction. Instead, automated text analysis revealed that customers wanted employees to linguistically “take charge” of the conversation. Specifically, this research suggested a shift to “solving” verbs (e.g., get, go call, do, put, need, permit, allow, resolve) as the interaction unfolds was an important predictor of customer satisfaction. Similar results were found by Ordenes and his colleagues in their in-progress research that analyzed a major consumer product company’s online chat-based interactions with customers. Automated text analysis of these conversations found that customer satisfaction is higher when front-line employees dynamically shift from deferent words (e.g., afraid, mistake, pity) to more dominant language (e.g., must, confirm, action).

Be specific. Analysis of the language used in telephone and email customer service interactions at two major retailers found that after the introductory phase of a conversation, when agents must show they are listening, customers see employees as more helpful when they use more concrete language. For example, for a clothing retailer, “white turtleneck” is more concrete than “shirt,” and “sneakers” is more concrete than “shoes.” Lab experiments currently underway by one of us (Grant Packard) and Jonah Berger suggest that using more concrete language signals to the customer that the agent is psychologically “closer” to the customer’s personal needs. For one of the retailers, a field analysis of over 1,000 email service interactions found that moving from “average” concreteness to one standard deviation above the average (roughly speaking, at about the 68^th percentile of linguistic “concreteness”) was linked to a 4% lift in customer purchases after talking to the employee.

Don’t beat around the bush. Analysis of the language used in consumer and expert product reviews — plus lab studies — suggest that subtle variations in the words used to endorse a product or action can have substantial effects. For example, people are more persuasive when they use words that explicitly endorse the product to the customer (“I suggest trying this one” or “I recommend this album”) rather than language that implicitly does so by sharing the speaker’s personal attitude (“I like this one” or “I love this album”) towards a product or service. This is because explicit endorsements signal both confidence and expertise on the part of the recommender, a perception that could be particularly important in personal selling contexts.

As more and more consumer-firm conversation moves online or to other text-based media, the importance of utilizing language properly is greater than ever. However, firms are frequently teaching their employees to use language that doesn’t stand up to scientific scrutiny. Fortunately, new research using automated text analysis and methods from linguistic psychology offer some simple, actionable, and nearly cost-free solutions to improve the speaking terms by which companies engage their customers. What’s more, advances in natural language processing and machine learning seem primed to help researchers and managers uncover even deeper insights on how words really work for businesses.