Consider a manager advertising for a job, deciding whether to go with “Apply today to join a great team!” or “Join a great team, apply today!” These messages are similar, and both are inviting a job application, but they are formulated differently—or, in other words, they use different syntaxes. Can the manager tell which message will more successfully facilitate applications? How can the manager compare the effectiveness and efficiency of the formulation of these messages and decide which one to use?
Formulating the right message for an intended purpose is not a trivial task. Today, companies face this challenge of finding the right message on an hourly basis due to their permanent presence on digital platforms. Many utilize multiple communication tools and channels to spread messages of various lengths and types, reaching consumers via social media, television, radio, or newspapers, as well as email, brand websites, and blogs. While marketing messages must weigh various elements—including content, images, and choice of channel—language is one of the most critical aspects of effective communication.
In a new Journal of Marketing study, we explore how companies can formulate effective messages intended to reach a desired outcome. We focus our study on syntax, which is the arrangement of words in a sentence, and investigate a measure called syntactic surprise: the average unexpectedness in the syntax of a message. By using state-of-the-art methods in natural language processing, we demonstrate the role of syntactic surprise in effective writing.
The Power of Surprise
Studies have shown that people begin comprehending a sentence before processing it in full. For example, people generally anticipate a direct object relationship after a verb (such as in the sentence “Amazon delivers diapers”), in which case the syntactic surprise is low. By contrast, an adverb following a verb is less expected and would bring on high surprise (such as “Amazon delivers fast”). Taken together, syntactic surprise is the unexpectedness of the syntactic element occurring (i.e., object: diapers vs. adverb: fast) given the previous syntactic element (i.e., verb: delivers) that the individual encounters in the sentence.
We conducted four main studies using large-scale field data and a series of follow-up experiments on Facebook and Instagram to validate syntactic surprise. We then assessed the role of this measure in various forms (i.e., experimental and field data), contexts (i.e., donations, advertising, and product reviews), and relevant outcomes (e.g., likelihood to donate and click-through rate). We find that syntactic surprise is a unique aspect of syntax that accounts for the effectiveness of marketing messages beyond previously established measures.
Additionally, we show that the relationship between syntactic surprise and effectiveness of the message follows an inverted U-shape: Messages are most effective at a medium syntactic surprise level but less effective at low and high levels. Next, we collectively analyzed the findings from the studies to show how the measure of syntactic surprise can be used (i.e., by marketing professionals) to improve the effectiveness of various messages. To this end, we classify the range of syntactic surprise into four categories: optimal, effective, acceptable, and ineffective. Through a series of field experiments on Facebook and Instagram, we demonstrate how managers can use this proposed approach to modify the syntactic surprise of their ads to increase click-through rates significantly and improve performance.
Syntactic Surprise Calculator
To simplify the process of computing syntactic surprise and improving a specific text, we developed an easy-to-use, free online tool that automates the use of the metric: the syntactic surprise calculator. This tool calculates the syntactic surprise of any text at the message and sentence level and then provides recommendations. Managers can revise their messages sentence by sentence until they reach the effective or acceptable range. The proposed approach is automatic, scalable, and can be used without any machine learning expertise.
Overall, our findings demonstrate the importance of syntactic surprise in various forms, contexts, and relevant outcomes and shows how to use syntactic surprise to improve marketing messages. With the use of the syntactic surprise calculator, communicators can improve their messaging strategies. Regardless of message length, a practitioner can measure the syntactic surprise of any text, assess its syntax, and use the results to improve the message. Any communicator (e.g., retailers, brand managers, advertisers, politicians, educators, policymakers) can benefit from the findings.
Read the Full Study for Complete Details
From: A. Selin Atalay, Siham El Kihal, and Florian Ellsaesser, “Creating Effective Marketing Messages Through Moderately Surprising Syntax,” Journal of Marketing.
Go to the Journal of Marketing