Does the language used in online recommendations lead to worse decisions?

Word of mouth has become increasingly important for firms as consumers turn to one another online for product recommendations. This research looks at the language consumers use to endorse products, and how this language impacts persuasion.

What You Need to Know

When endorsing products, people tend to do so either explicitly (“I recommend…”) or implicitly (“I like…”). These minor language variations are an important cue that consumers use to assess source expertise in technology-mediated contexts (e.g. online reviews, forums).

This research reveals that consumers believe those who use explicit endorsement language are more expert than those that use the implicit style. This belief contributes, in part, to “recommendations” being more persuasive than “likes.” Paradoxically, however, novices are significantly more likely to use explicit endorsements than experts. This can lead people who use word of mouth information astray.

More Details

What Did the Researchers Do?

Grant Packard (Lazaridis School) and Jonah Berger (Wharton) asked:

1. Are there meaningful linguistic variations in how consumers endorse products to one another;
2. Whether category knowledge moderates the endorsement language consumers use; and
3. Whether these language variations affect the persuasive impact of word of mouth.

The researchers report the results of a field data analysis of over 1,000 online reviews and four experiments to answer these questions.

What Did the Researchers Find?

First, the authors analyzed over 1,000 real consumer reviews and the review writers’ purchase histories to identify common endorsement styles, and to test whether category experience impacted style preferences. Results reveal explicit endorsements (e.g., “I recommend…”, “I suggest…”) and implicit endorsements (e.g., “I like…”, “I enjoyed…”) as the two dominant styles. Consumers who had bought fewer products through the website were almost 4x more likely to explicitly endorse products than those who had a greater purchase history with the firm. This suggests that category experience may be linked to endorsement style.

Next, two experiments formally tested the relationship between category knowledge and explicit endorsement, and examined the mechanism behind this effect. Results confirm that category novices are significantly more likely to “recommend” products than experts, and awareness of variation in others’ tastes (preference heterogeneity) drives this result. In short, less knowledgeable consumers tend to imagine that if they like something, say a particular wine, everyone else will too. They therefore “recommend” it to others more frequently. In contrast, consumers who are more knowledgeable about wine appreciate that some people may like bold Cabernets, while others may prefer lighter Pinots. As a result, rather than recommend things to others, knowledgeable consumers tend to just describe their own preferences when endorsing something (e.g., “I enjoyed…”).

A set of additional experiments studied endorsement style’s impact, testing how explicit endorsements affect persuasion, and the mechanisms underlying this effect. Explicit recommendations (e.g., “I suggest…”) generated perceptions that the information source was more expert, and further, that they liked the product more than the implicit style (e.g., “I liked…”). These perceptions then led to increased purchase intentions for the reader.

Finally, a unique “yoked” study design demonstrated that, overall, the relationship between consumer knowledge and endorsement styles can lead users of online product information to make worse decisions. Specifically, more novices chose an objectively inferior (vs. superior) product, and then chose to explicitly recommend it in an online review. Participants in another study “yoked” to this one saw the actual distribution of “recommends” and “likes” for the products in the word of mouth condition. This led them to be even more likely to choose the worse product versus those asked to make the same decision without word of mouth information.

How Can You Use This Research?

Online retailers, media or review portals, and social media sites can use this research to assess how their rating scale labels (such as the “likes” used by Facebook versus the “highly recommended” scale end point at Amazon) may impact product rating quality and consumer persuasion. They may also change instructions on their website to encourage reviewers to share their own tastes by encouraging implicit endorsements. Firms pursuing increased use of “recommendation” language should consider whether doing so will enhance both firm and consumer welfare in the long run.

Marketing managers can use this research to help optimize the language they use in their own marketing communications whether in advertising, direct marketing, or paid (e.g., celebrity) endorsements.

Want to Know More?

Article citation: Packard, G. and Berger, J. (2017), “How Language Shapes Word of Mouth’s Impact,” Journal of Marketing Research, 54 (4), 572-588.