We all conduct our own tests every time we tweet to see what gets noticed and what gets flushed into the ever-deepening pit of pointless, unseen slosh that is the twittersphere.

It’s therefore no surprise that all kinds of theories have emerged over the years that try to pinpoint what makes the perfect tweet. Some say that a tweet should never be longer than 100 characters. Twitter itself says that a photo URL in a tweet increases the retweet rate by 35%.

But what effect does the actual wording of the tweet have on the amount of retweets it gets? A 2014 study conducted by Cornell University, in collaboration with a research scientist at Google, attempted a scientific experiment to analyse this question.

The study is the first large-scale experiment studying the effects of author wording on information propagation.

The study took advantage of a surprisingly common phenomenon, where Twitter users tweet the exact same information, but worded slightly differently, a couple of hours apart.

It’s assumed they do this because they’re unsatisfied with the level of engagement the first time around.

The researchers analysed 1.77m tweet pairs to eventually find 11,404 that were virtually identical in terms of content and subject matter, with the only difference being the use of words. To reduce the effect that the amount of followers would have on the results, the tweet pairs belonged to users who had at least 5,000 followers.

Certain features were then tested to see whether their presence in a tweet increased the amount of retweets. These included the length of the tweet, the kinds of words used (verbs, nouns etc) and whether hashtags and mentions were used.

The study made the following findings:  

1. It helps to ask people to share

Unsurprisingly, tweets containing the words ‘RT’ and ‘retweet’ gained significantly more engagement, whilst those containing ‘spread’ and ‘please’ saw a noticeable improvement.

2. Be informative

More informative tweets got more retweets. Contrary to previous studies, the test found that longer tweets received higher engagement than shorter ones, presumably because they contained more information.

The inclusion of numbers also boosted the retweet rate, presumably because this meant the tweet contained hard facts.

3. Be like the community, and be true to yourself

Although distinctive messages may attract attention, messages that conform to expectations tend to be more easily accepted and, therefore, more readily shared.

In terms of the words you pick, a balance should be struck between using the general language conventions of Twitter and your own personal style.

4. Imitate headlines

Tweets phrased in a news headline style are often, by their very nature, more informative and attention-grabbing than other tweets.

Naturally, this results in better engagement and more shares.

5. Include positive and/or negative words

Using words that promote an emotional response in the reader increases retweets.

Using positive, negative, or a contrast of both sentiments resulted equally in higher engagement.

6. Generality helps

Tweets that aren’t tailored for specific audiences get more retweets. 

Hashtags had very little effect on what got retweeted, and mentions actually had a negative effect. The researchers suggested that this might be because users felt mentions indicated a personal message, and so were not for sharing.

The same was true for pronouns. Using the third person (he, she, they) resulted in more retweets than using the first (I, we) or second (you) person. 

The study concluded that the best way to get more retweets is by adding more information, making one’s language align with both community norms and with one’s prior messages, and by mimicking news headlines.

Why not check it out for yourself? The researchers have developed a simplified version of the tools they used for the experiment, allowing you to test two tweets to see which would, hypothetically, be retweeted more.