New Retweet Data: Tweet Character Length

More Retweet data continues to pour in from Dan Zarrella’s blog. In addition to recent posts on images, quotes & hashtags – Zarrella just released data that shows how character length in Tweets effects Retweet rates.

The data analyzed 1.4M tweets from 1.2M unique Twitter accounts. Zarrella found that tweets between 100-115 characters were 34% more likely to be retweeted compared to tweets with character lengths outside that range. The results were calculated within a 99.9% confidence interval.

As you can see, there is a nearly linear growth in Retweet rates from 10 to around 110 characters. At that point, there is a severe dropoff on Retweet occurrence that flatlines near zero around the 130 character mark.

The flatline area might be explained by the fact that most social media users typically leave some room at the end of a tweet for commentary. Therefore, the number of Retweets greater than 130 characters is probably very low simply because there are a much lower number of tweets within this character range. It would be interesting to see a graph similar to the one above that corrects for distribution of character lengths among the data set.

Zarrella doesn’t provide this data, but I would assume that many of these Retweets are “native” retweets (a Retweet from clicking the Retweet button, which does not allow additional commentary). If a significant number of native retweets exist, then Retweeting behavior would be independent of commentary. Therefore, it would be much more interesting to understand Retweeting behavior between types of Retweets (quotes vs. classic “RT” vs. native retweets).

It would also be very informative to understand the distribution of Retweets as a percentage of each character count total, which would correct for character length distribution as I mentioned above. For example, Zarrella might find that 135 character tweets actually get more Retweets as a percentage of all 135 character tweets when compared to 60 character tweets, which would go against Zarrella’s graphic above.

While the Retweet length information above is certainly interesting, there are too many independent variables at work to make any real informed decisions from Zarrella’s findings.