How Twitter Usage Differs Between Humans and Company Accounts
0 likes522 views
The document analyzes differences between tweets from company accounts and personal accounts that one user follows. It finds that company accounts tweet much more often, with their tweets making up around 80% of tweets in the user's timeline even though they only represent around 46% of followed accounts. Company tweets also tend to receive more engagement in the form of favorites and retweets. In contrast, personal accounts interact with each other more by mentioning each other in tweets. Finally, company tweets are much more likely to include a URL, with around 95% doing so compared to a lower rate for personal accounts.
![Wrote quick script to pull attributes of
interest:
# basic structure is based on this:
timeline = client.home_timeline( options )
# the attributes pulled:
headers = [
"tweet.id",
"tweet.user.id",
"tweet.user.screen_name",
"tweet.created_at",
"tweet.full_text",
"tweet.favorite_count",
"tweet.retweet_count",
"tweet.user_mentions?",
"tweet.hashtags?",
"tweet.media",
"tweet.url",
"tweet.user?",
"tweet.lang",
"tweet.source",
"tweet.user.name",
"tweet.user.contributors_enabled?",
"tweet.user.verified?",
"tweet.user.website_urls.first.expanded_uri.to_s",
"tweet.urls?"
]
# then for each Tweet in timeline, run and store values from headers](https://image.slidesharecdn.com/howtwitterusagediffersbetweencompanyaccountsandpersonalaccounts-150916075418-lva1-app6891/85/How-Twitter-Usage-Differs-Between-Humans-and-Company-Accounts-4-320.jpg)
How Twitter Usage Differs Between Humans and Company Accounts
- 1. Based on those I follow How Twitter Usage Differs between Company Accounts and Personal Accounts @alexweberk
- 3. https://github.com/sferik/twitter Tweets: ● Between 8/22/15 12:34 and 8/23/15 14:33 ● Total of 1276 Tweets from accounts @alexweberk follows Sample is limited, and by no means representative of all tweets. Used Twitter Ruby Gem to pull data @alexweberk
- 4. Wrote quick script to pull attributes of interest: # basic structure is based on this: timeline = client.home_timeline( options ) # the attributes pulled: headers = [ "tweet.id", "tweet.user.id", "tweet.user.screen_name", "tweet.created_at", "tweet.full_text", "tweet.favorite_count", "tweet.retweet_count", "tweet.user_mentions?", "tweet.hashtags?", "tweet.media", "tweet.url", "tweet.user?", "tweet.lang", "tweet.source", "tweet.user.name", "tweet.user.contributors_enabled?", "tweet.user.verified?", "tweet.user.website_urls.first.expanded_uri.to_s", "tweet.urls?" ] # then for each Tweet in timeline, run and store values from headers
- 6. 1. Company accounts tweet a lot more than humans @alexweberk
- 7. Company Accounts Tweeted 4.6x More @alexweberk
- 8. Company Accounts Tweeted 4.6x More Company Accounts simply tweeted a lot more on average than Personal Accounts. Almost around 4.6x more, in this case. @alexweberk
- 9. @alexweberk
- 10. Distribution of account type of accounts I follow Distribution of tweets by account type @alexweberk
- 11. Company accounts tweet a lot more than personal accounts @alexweberk
- 12. 46% of the accounts I follow are Company Accounts, but they represent 79.8% of the tweets in my Timeline. @alexweberk
- 14. Filled with Tweets from Company Accounts
- 15. 2. Company accounts have more tweets with engagement. @alexweberk
- 16. @alexweberk
- 17. @alexweberk
- 18. Company Accounts are better at getting engagement with their tweets. @alexweberk
- 19. Why? @alexweberk
- 21. They Use More Hashtags (=easier to get found by searches) @alexweberk
- 22. They Use More Media in Tweets (images, etc) @alexweberk
- 23. And, They Tweet A Lot More. Same chart from earlier -> @alexweberk
- 24. Summary: Ways of getting more engagement: 1. Use Hashtags (Easier to get found via search. Get on trends) 2. Use Media (Images are more eye-catching. It takes up more screen real estate.) 3. Lots of Tweets (At least some of them have to get attention, right?) @alexweberk
- 25. 3. Humans interact a lot more with each other @alexweberk
- 26. @alexweberk Humans mention each other a lot more
- 27. @alexweberk DOIN’ IT RIGHT. Humans mention each other a lot more
- 28. 4. Humans don’t have a link in every single tweet @alexweberk
- 29. Likelihood to Include URL(s) in Tweet Personal Accounts Company Accounts @alexweberk
- 30. Likelihood to Include URL(s) in Tweet Personal Accounts Company Accounts @alexweberk 95%
- 31. Likelihood to Include URL(s) in Tweet Personal Accounts Company Accounts @alexweberk This is a really huge difference across the two User Types, so much so you can probably determine whether an account is a bot or not by the frequency of links in its tweets.
- 34. Distribution of favorites are heavily concentrated to the top performing tweets. Winner Takes All - Favorite Count @alexweberk
- 35. Distribution of favorites are heavily concentrated to the top performing tweets. Winner Takes All - Retweet Count @alexweberk
- 37. You can tell the twitter client used via the Twitter API Buffer popularity is notable. SocialFlow has large companies as users. @alexweberk
- 39. Thank you! @alexweberk This is a personal account! :)