Which words make a brand’s social media post more shareable? What a Journal of Consumer Psychology study found
Posts that ask readers both to choose well and to get going earned more shares, but only when their language stayed abstract. Text analyses of brand and customer posts on Facebook and Twitter and three studies with US adults back the pattern, with a caveat for customers’ own replies.
Pham et al. find brand posts pairing be-right words like right with action words like get earned more shares when the wording around them stayed abstract.
Marketing in 3 · 2023 in research · No. 26 · Paper published 6 July 2023
The same skyline, close up on a screen and far away across the water. Photo: Frolicsome Fairy on Unsplash.
The short answer
Words that ask readers to get it right (right, perfect, compare) placed next to words that ask them to get moving (get, go, start) were linked to more sharing, but only when the rest of the post was abstract and big-picture. The pattern appeared in Facebook posts and tweets from 100 global brands, and in online studies where US adults first answered "why" rather than "how" questions, or pictured a place far from home.
Evidence level
★★★One study
★★★Several studies
★★★Meta-analysis
Key takeaways
Only 4.24% of 210,963 Facebook posts from 100 global brands contained both a be-right word and an action word, and only 2.88% of their 311,365 tweets did.
On Facebook, more of the paired wording went with more shares in the 42.83% of brand posts with the most abstract language, and with fewer shares in the most concrete 7.64%.
After answering "why" questions, people listing their thoughts on an ad with both kinds of words wrote 24.47 words on average, against 18.45 to 20.31 words for the other versions, a sign of closer attention.
Across 377,086 replies written by customers to the same brands, more paired wording was tied to fewer retweets for 86.06% of replies, and to more retweets only in the most abstract 0.45%.
What the researchers did
Thuy Pham and
Felix Septianto of The University of Queensland,
Frank Mathmann and
Hyun Seung Jin of Queensland University of Technology, and
E. Tory Higgins of Columbia University
asked what makes people pass a post on, which matters to anyone who runs a brand account. They started from two
motivations that psychologists call assessment and locomotion. Assessment is the wish to be right, to weigh the
options and pick the best one. Locomotion is the wish to act, to get started and get things done. Most posts appeal
to one or the other. The team looked at posts that appeal to both at once.
Their argument was that the two motivations make different demands on attention, and that abstract, big-picture
language lets a reader combine them. Someone thinking about why a post matters can take "choose well" and "get going"
as a single aim. Someone thinking about how, about steps and details, faces two separate requests.
Six studies tested the idea. Three applied text analysis to real posts from 100 large international brands, such as
Amazon, Coca-Cola and Nike, collected between January 2017 and October 2022. The other three were online studies with
US adults.
Field data
210,963 brand posts on Facebook, 311,365 brand tweets and 377,086 customer replies to the brands on Twitter
Brands
100 international brands taken from Interbrand's Best Global Brands 2022 or from earlier research on the language of brand posts
Online studies
399, 798 and 794 US adults, all from the Prolific panel; the first of these was preregistered
Word measures
A published dictionary of assessment and locomotion words, and a concreteness dictionary that rates how abstract each word is
Outcomes
Share and retweet counts in the field data; stated likelihood of sharing, or a choice of which version to share, in the online studies
How were the words scored?
Be-right words in the dictionary include right, perfect, correct and truth. Action words include get, go, act, start
and make. For every post the team worked out the proportion of be-right words and the proportion of action words,
then multiplied them, so a post scored above zero only when it held both kinds. This issue calls that paired
wording. The researchers' own example is Gillette's "Get the right shave for the right start to your day": eleven words, with right twice, plus get
and start.
Abstractness came from a second dictionary that rates how directly a word refers to something people can see, touch
or hear. Each post took the average rating of its words, and a higher score meant more abstract language.
What did the brand posts show?
In Studies 1a and 1b, the team modelled share counts with methods built for count data. The models also allowed for the account's age and
audience, the post's age, hashtags, mentions, links and length, its emotional tone, and the time and day it went out.
On both platforms, paired wording and abstract language worked together. The more abstract a post, the more the
pairing went with extra shares on Facebook and extra retweets on Twitter. For posts one standard deviation more
concrete than the average, the pairing made no measurable difference on either platform.
The Facebook result held when brand, product category and calendar timing were also taken into account. The Twitter
result kept its direction under those controls but was no longer statistically significant. The team then corrected
for brands choosing their words strategically. Both results kept a positive sign but fell short of significance, a
pattern the authors call directionally consistent.
Customers' own replies
Study 2 repeated the analysis on replies that customers wrote to the same 100 brands. These replies were
rarely passed on, with 0.22 retweets each on average.
Here the paired wording was linked to fewer retweets overall. The more abstract a reply, the weaker that negative
link became, which is the same interaction the brand posts showed. A positive link appeared only in a very small set
of highly abstract replies. On this evidence, the case for paired wording is clearest for posts that brands write
themselves.
What the online studies added
Study 3 measured people's own tendencies. Its 399 US participants chose between two slogans for a city named
Echuca. One slogan stressed what was desirable about the place, an abstract appeal, and the other what was practical,
a concrete one. Participants said which they would rather share and then completed a standard 24-item
questionnaire on assessment and locomotion. The more inclined people were to act, the more they preferred to share
the abstract slogan, but only among those also strongly inclined to evaluate. Among people low on evaluating, the
inclination to act made no difference.
Study 4 set the wording itself. Its 798 US participants first answered a series of either why or how questions about
sharing information on social media, a standard way of putting people into an abstract or a concrete frame of mind.
They then saw a Facebook ad for plant-based meat written with be-right words (think, alternatives), action words
(momentum, change), both kinds, or neither, and rated their likelihood of sharing it. After why questions, the
version carrying both word types came out ahead of each of the other three. After how questions, the four versions
did not differ.
Likelihood of sharing a plant-based meat ad on Facebook (1 to 7), by ad wording and the questions answered first
Likelihood of sharing a plant-based meat ad on Facebook (1 to 7), by ad wording and the questions answered first
Group
Value ()
Why questions: both word types
2.45
Why questions: be-right words only
2.04
Why questions: action words only
1.78
Why questions: neither
1.92
How questions: both word types
1.93
How questions: be-right words only
1.97
How questions: action words only
2.0
How questions: neither
2.06
Study 4, 798 US adults, group means reported in the text. After why questions the ad with both word types beat each other version; after how questions the four did not differ.
Attention is the link
Study 4 also measured engagement, meaning how absorbed people were in the ad. On an eight-item scale, the
paired version scored 3.48 out of 7 after why questions, against 2.99 to 3.03 for the others. Tests
of the indirect path showed that the sharing advantage ran through this higher engagement.
Study 5 checked the same idea from another angle. Its 794 US participants read a short Twitter review of a new restaurant
located either close to their home or far from it, since distance tends to make people think in more abstract terms.
The paired review ended "Compare other alternatives for yourself! I definitely recommend you quickly get here!", while
the plain version used neither kind of word. Half the participants also wrote three or four sentences about their
thoughts on the post before rating it, a task that raises attention for everyone. Without that task, the paired
review drew higher retweet ratings when the restaurant was far away and no higher ones when it was close. With the
task, the advantage disappeared, which is what the researchers expected if attention is the mechanism.
Likelihood of retweeting a restaurant review (1 to 7), by distance, review wording and a writing task
Likelihood of retweeting a restaurant review (1 to 7), by distance, review wording and a writing task
Group
Value ()
Far away: paired review
2.8
Far away: plain review
2.16
Close to home: paired review
2.17
Close to home: plain review
2.32
Wrote first, far: paired review
2.4
Wrote first, far: plain review
2.67
Wrote first, close: paired review
2.74
Wrote first, close: plain review
2.61
Study 5, 794 US adults, group means reported in the text. Only the first pair differs significantly; writing about the post first removed the difference.
The authors' advice for brands
The authors' first recommendation is to write posts that combine the two word types in abstract language,
and to check drafts against the word lists before they go out. Their second is to look at the words
customers use. For people who use both kinds of words themselves, they suggest ads that stress what a product makes
possible over how easy it is to use. Their illustration is a home improvement tool that "will make your bathroom look
fantastic" rather than one "anybody can fix their bathroom with". That example comes from the discussion and was not
itself tested.
Marketing in 3 steps
Ready. Set. Market.
1
ReadyDecide and prepare
Count the two kinds of words in past posts. Pull your last 50 posts. Mark every word that asks people to choose well, such as right, best, perfect or compare, and every word that asks them to move, such as get, go, start or make. Count how many posts carry at least one of each.
Write down the reason behind each post. Before drafting, finish the sentence "This matters to our customers because..." in one line. That line is the big-picture purpose the post should carry, not the steps or the specifications.
2
SetBuild it
Put one choosing word next to one moving word. Give each caption at least one word from each list, as in "Get the right gear and go". Keep to one or two of each, so the line still sounds like something a person would say.
Lead with why and move the how elsewhere. Turn feature lists and instructions into the result they lead to. Swap "three settings and a quick charge" for "make every morning easier", and put the details behind the link or in the first comment.
3
MarketLaunch and measure
Run the new wording against the old. For two weeks, alternate posts in the new style with posts in your usual style on the same platform, at the same times of day. Compare shares per 1,000 impressions, not likes.
Review the results every three months. Sort the quarter's posts into four groups: both kinds of words or not, big-picture or detailed. Keep whichever style earns more shares relative to its reach, and drop the pairing on any channel where it does not help.
Found this useful?
Marketing in 3 is free and ad-free, with new issues every Monday and Thursday. A coffee keeps it that way.
Small effects. Average ratings across the conditions of the two experiments ran from 1.78 to 2.80 out of 7, so even the best wording left most readers unlikely to share.
Abstract language is the condition. Paired words in detailed, step-by-step posts brought no benefit in any of the tests.
Attention matters. The advantage vanished once readers were made to think hard about the post anyway. An inference from this, not a tested claim, is that the wording matters most where people scroll past quickly.
Platforms differ. Abstract language on its own went with more shares of brand posts on Facebook but with fewer retweets on Twitter, so test each channel separately.
Where it does not work
Field data show links, not causes. Brands that write in this style may also differ in ways the models could not capture.
Stated intentions. Participants rated their willingness to share or picked a preferred version; no real sharing was recorded.
Text only, in English. Images and video, now central on most platforms, were not analysed, and other languages remain untested.
US participants. All three online studies recruited US adults through one panel.
Some topics may not benefit. The authors expect that disgusting or violent posts would gain nothing from this kind of wording, a possibility left for future work.
Questions marketers ask
What wording makes people more likely to share a brand post?
Pairing a word about getting it right, such as right or perfect, with a word about acting, such as get or go, was linked to more sharing when the surrounding language was abstract, focused on why rather than how (Pham et al., 2023). In concrete, detailed posts the pairing made no difference.
How does abstract language differ from concrete language?
Abstract language describes the purpose behind something: what an action is for and why it matters. Concrete language describes the steps, features and details a person could see or touch. In the experiments, answering why questions or thinking about a distant place was used to put people in an abstract frame of mind.
Does it work for customer reviews and replies?
Partly. In one of the online studies, a restaurant review that combined both kinds of words scored higher on retweet likelihood among readers told the restaurant was a long way from home. Among real customer replies to brands on Twitter, the combination was mostly tied to less sharing, and only the most abstract replies gained.
Why does the pairing need big-picture wording?
Evaluating and acting compete for attention. The researchers argue that thinking about the larger goal lets readers merge them into one aim, which keeps them absorbed in the post, and that this absorption leads to sharing. When attention was raised by a writing task instead, the pairing no longer helped.
How was the abstractness of a post measured?
With a published concreteness dictionary of 85,863 words, in which words tied to direct sensory experience count as concrete. A post's score was the mean across its words. A rougher check for everyday use, our suggestion rather than the researchers', is whether a draft says why something matters or only how it works.
Research on whether people would rather rely on an algorithm or on another person mostly draws on participants from wealthy Western countries. This paper includes people from more than 30 countries on every inhabited continent. Those born or raised in countries with high perceived corruption were far less averse to algorithmic decisions, in some studies not averse at all, and making corruption salient in an experiment also reduced the aversion. Acceptance of automated decisions varies by market, so the argument for an algorithm may need different framing from one country to the next.
Ads that make people curious tend to be read more closely, which might be expected to make readers more critical. Across four studies, raising curiosity and then answering it had the opposite effect: the answer brought a pleasant feeling, scepticism towards the product fell, and attitudes and purchase intentions rose. The information itself was held constant, so the way it was disclosed made the difference, and it worked even when the curiosity came from an unrelated source. A teaser helps only if the campaign delivers the answer it promises.
Source
Pham, T., Septianto, F., Mathmann, F., Jin, H. S., & Higgins, E. T. (2023). How construal–regulatory mode fit increases social media sharing. Journal of Consumer Psychology, 33(4), 668-687. https://doi.org/10.1002/jcpy.1375
This brief is written for practitioners from the published article. Open access (CC BY-NC-ND 4.0): free to read at the publisher. Numbers are as reported by the authors; read the article for full methods.
Thumbnails with a middling amount going on won the clicks at 5.48 out of 7. Middling colour harmony lost them at 3.41. Two dials, pulling opposite ways.
Schivinski's machine learning study of 1,356 brand followers found five engagement patterns driven by commitment, and they still matter in the age of AI.
Bruno Schivinski is Associate Professor of Marketing at Gdańsk University of Technology. He studies how consumers engage with brands in digital and social media.