Sunday, 5 July 2015

The Best Time To Tweet For Clicks, Retweets and Replies

Updated 7/1/2015: Added some new insights and learnings into the blog post here. We’re so grateful for the chance to learn from the community on this!
Imagine removing all guesswork when you schedule your tweets, knowing the best times to experiment with tweeting for maximum clicks and maximum engagement.
As someone who shares frequently to social media, this info would be fantastic to have! We’re always eager to dig up new research into social media best practices—things like length and frequency and timing.
The timing element, in particular, feels like one where we’d love to dig deeper. And we just so happen to have a host of data on this from the 2 million users who have signed up for Buffer!
With a big hand from our data team, we analyzed over 4.8 million tweets across 10,000 profiles, pulling the stats on how clicks and engagement and timing occur throughout the day and in different time zones. We’d love to share with you what we found!
best time for twitter

The best time to tweet: Our 4.8 million-tweet research study

Our key learnings

Wow, we learned so much looking at the awesome stats from those who use Buffer! Here were some of the takeaways we came up with. I’d love to hear what catches your eye, too!
  • Based on all of the tweet data we have collected, the early morning hours appear to be the time in which tweets receive the most clicks, on average.
  • Evenings and late at night are the times when your tweets receive the most favorites and retweets, on average.
  • In some cases, times with the highest amount of average engagement are almost inversely related to the most popular times to tweet.
  • The most popular time to tweet and the best times to tweet for engagement differ across time zones, so it’s still important to experiment and find the times when your audience is most engaged.

What you might do with this data

In some of the results below, you’ll see specific times that we found to be the most popular times to tweet or the best times to tweet for clicks, for example.

What we love to do with specific takeaways like this is use them as the starting points for new experiments.
Over time we’ve come to learn that research studies like these are great for inspiration, not prescription. I’d pause slightly in suggesting that you change your whole Buffer schedule to align with these new Twitter times—unless your data and analysis says so!
What I’d love to suggest is that these new times perhaps give you ideas about what to test out next with your social media sharing, perhaps some counterintuitive suggestions about what to try—tweeting at non-peak times, tweeting at 2:00 a.m. for clicks, for instance.
And we’d love to hear back on how any of these tests turn out for you!

The most popular time to tweet:

Noon to 1:00 p.m.

We’ve taken the data from all tweets sent through Buffer to find the most popular times for posting to Twitter. Looking at all tweets sent across all major time zones, here is an overview of the most popular times to tweet.
  • Noon to 1:00 p.m. local time, on average for each time zone, is the most popular time to tweet
  • The highest volume of tweets occurs between 11:00 a.m. and 1:00 p.m., peaking between noon and 1:00 p.m.
  • The fewest tweets are sent between 3:00 and 4:00 a.m.

What you might do with this data

The most popular time to tweet is also the time when there is the highest volume of tweets, perhaps making it a bit more difficult for your tweet to stand out in someone’s timeline.
What might be great to try here is tweeting at non-peak hours, the times in the early morning and late evening.
Another thought is that the most popular times to tweet could very well correlate to the times when most people are on Twitter. Perhaps it’s worth testing also to see if tweeting during a popular time is worthwhile simply for the amount of people who are online.
(One great stat to look at with this is tweet impressions, which you can find inTwitter’s free analytics.)
Here’s the chart for the most popular times worldwide, taken from an average of 10 major time zones (the times represent local time).
Most Popular Time to Tweet Worldwide
Here is the graph for the most popular times to tweet in each of the four major U.S. time zones. 
Buffer social media science study - US popular times to tweet
(We normalized the data to account for daylight’s savings in the U.S. as well.)
Here are the charts for the major time zones in Europe and Africa.
Most Popular Time to Tweet Europe
(Note: The London (GMT) time zone used to be the default time zone for new Buffer users, so our data for GMT is not as clean as we would like it to be. We’ve omitted any takeaways for GMT from the research results here.)
Here are the charts for the major time zones in Asia and Australia.
Most Popular Time to Tweet Australia Asia
It’s interesting to see how the most popular time to tweet varies across the time zones. We’ve shared Buffer’s 10 most popular time zones in the charts above. Here’s a list of each most popular hour for the 10 major time zones.
  • Los Angeles, San Francisco, etc. (Pacific Time): 9:00 a.m.
  • Denver (Mountain Time): noon
  • Chicago (Central Time): noon
  • New York, Boston, Atlanta, Miami, etc. (Eastern Time): noon
  • Madrid, Rome, Paris, etc. (Central European): 4:00 p.m.
  • Cape Town, Cairo, Helsinki, etc. (Eastern European): 8:00 p.m.
  • Sydney (Australian Eastern): 10:00 p.m.
  • Hong Kong (Hong Kong Time): 8:00 a.m.
  • Tokyo (Japan Time): noon
  • Shanghai, Taipei, etc. (China Time): 9:00 p.m.
For any clarification on this or the other research throughout this article, feel free to leave a comment and we’ll get right back to you.
Takeaways & thoughts:
  • The most popular time to post could be due to a number of factors: This is when most people have access to Twitter (perhaps at a work computer), this is when online audiences are most likely to be connected (see Burrito Principle), etc.
  • Should you post during the most popular times? That’s one possibility. Also, you may find success posting at non-peak times, when the volume of tweets is lower.
  • If you have a large international audience on Twitter, you may wish to locate the particular part of the world where they’re from, and adjust your schedule accordingly. You can find the times when your audience may be online with tools like Followerwonk and Crowdfire.

The best times to tweet to get more clicks

We were excited to dig into the specific metrics for each of these tweets, too, in hopes of coming up with some recommendations and best practices to test out for your Twitter strategy.
First up, the best time to tweet for clicks.
Looking at the data, we found the following trends for maximizing your chance to get more clicks:
  • Tweets sent between 2:00 and 3:00 a.m. earn the most clicks on average
  • The highest number of clicks per tweet occurs between 2:00 a.m. and 4:00 a.m., peaking between 2:00 and 3:00 a.m.
  • The fewest clicks per tweet happen in the morning (when tweet volume is particularly high), between 9:00 a.m. and 1:00 p.m.

What you might do with this data

We were fascinated to see that the best time to earn the most clicks on average was the middle of the night. It’s a quite counterintuitive results!
One interpretation here is that with this being the average, there is the possibility that outliers can have a large impact on the data—for instance an tweet that gets 4,000 clicks at 2:00 a.m. would raise the average significantly.
So what I might take away from this is that tweeting at 2:00 a.m. would likely not mean that every 2:00 a.m. tweet will see really high click numbers but that every once in awhile a 2:00 a.m. tweet could really take off. 
** Scroll to the bottom of the post for an updated explanation of this data (as well as some alternate views). We’re very grateful for all the helpful comments on this! **
The data in the below chart is the worldwide average, calculated for the local time in each time zone. So the peak at the 2:00 a.m. hour would hold true as the overall top time no matter which time zone you’re in—2:00 a.m. in Los Angeles, New York, Cape Town, Hong Kong, etc.Best Times to Tweet for Clicks Worldwide

For the specifics on each of the best time to tweet for clicks in each of the major time zones in Buffer, here’s a breakdown.
  • Los Angeles, San Francisco, etc. (Pacific Time): 2:00 a.m.
  • Denver (Mountain Time): 7:00 p.m.
  • Chicago (Central Time): 2:00 a.m.
  • New York, Boston, Atlanta, Miami, etc. (Eastern Time): 11:00 p.m.
  • Madrid, Rome, Paris, Berlin, etc. (Central European): 2:00 a.m.
  • Cape Town, Cairo, Istanbul, etc. (Eastern European): 8:00 p.m.
  • Sydney (Australian Eastern): 2:00 a.m.
  • Hong Kong (Hong Kong Time): 5:00 a.m.
  • Shanghai, Taipei, etc. (China Time): noon
  • Tokyo (Japan Time): 8:00 a.m.
Best Times to Tweet for Clicks - by time zone
Takeaways & thoughts:
  • Clicks was far and away the largest engagement metric that we tracked in this study (compared to retweets, replies, and favorites).
  • Some of the recommended best times for individual time zones show thatnon-peak hours are the top time to tweet for clicks. This data may reflect some particularly high-achieving posts—some outliers—that bring up the average when the volume of tweets is lowest. Still, it’d be a great one to test for your profile to see what results you get.
  • One neat thing to keep in mind is that a non-peak hour in, say, Los Angeles may correspond to a peak hour in London or Paris. The worldwide audience is definitely one to consider when finding the best time to tweet.
  • The 2:00 a.m. recommendation in the worldwide chart is made by looking at the average of all the data and hence may include the effect of certain outlier accounts. One way I like to look at this is that the potential exists for great leaps in engagement for your tweet by posting at 2:00 a.m., however it may be unlikely to expect that posting at 2:00 a.m. would bring consistently higher click rates on each and every individual tweet.
For more context on this, see our note from Julian, Buffer’s data scientist, below.

The best times for overall engagement with your tweet

We define engagement as clicks plus retweets, favorites, and replies. When looking at all these interactions together, we found the following trends formaximizing your chance to get the most engagement on your tweets:
  • Tweets sent between 2:00 and 3:00 a.m. earn the most total engagement on average
  • The highest amount of engagement per tweet occurs between 11:00 p.m. and 5:00 a.m., peaking between 2:00 and 3:00 a.m.
  • The smallest amount of engagement happens during traditional work hours, between 9:00 a.m. and 5:00 p.m.
Best Times to Tweet for Engagement
Takeaways & thoughts:
  • The best times to tweet for engagement are quite the inverse of the most popular times to tweet. (The late-night infomercial effect—tweet when fewer people are tweeting—seems to be the case here.)

The best times for retweets and favorites on your tweets

Adding together two of the most common engagement metrics, we found some interesting trends for maximizing the retweets and favorites on your tweets, especially for those with a U.S. audience.
Looking at 1.1 million tweets from U.S. Buffer users from January through March 2015, here were some of the notable takeaways we found:
  • Tweets sent at the 9:00 p.m. hour in the U.S. earn the most retweets and favorites on average
  • The highest number of retweets and favorites occurs between 8:00 p.m. and 11:00 p.m., peaking between 9:00 and 10:00 p.m.
  • The lowest retweet-favorite engagement happens at 3:00 a.m.
(Interesting to note, the takeaways from this data compared to the worldwide engagement data differ slightly for a couple reasons: 1) clicks represent a huge portion of overall engagement, and 2) the worldwide vs. US datasets vary.)
Best Times to Tweet for Engagement USA
We’d love to make it easy for you to share these results with your audience, your friends, your clients—anyone you think might benefit from them.
>> Download every chart from this post (.zip) <<

The methodology for our research

We studied all tweets ever sent through Buffer—4.8 million tweets since October 2010!
Based on this sample set, we looked at the number of clicks per tweet, favorites per tweet, retweets per tweet, and replies per tweet, in accordance with the time of day that the tweet was posted to Twitter.
Further, we segmented the results according to time zones, based on the assumption that the learnings might be more actionable if they could be specific to exactly where you live and work.
We had an interesting opportunity to consider whether median or average would be the better metric to use for our insights. It turns out that so many tweets in the dataset receive minimal engagement that the median was often zero. For this reason, we chose to display the average.

Thanks for all your great comments on this data. We’re so grateful for your help in improving these results. See below for an update from our data scientist Julian. :)

Update: Context & clarifications on the Twitter timing data

Some of the findings in this study are quite counter-intuitive. In particular, our finding of 2:00 a.m. as the best time to tweet for clicks really stands out.
I would love to take this opportunity to try to explain why.
The early morning hours were when tweets received the most clicks, on average, in several of the time zones we analyzed. There are several factors apart from the hour of day that may affect the amount of clicks that a tweet receives, many of which we’ve been grateful to learn from you in the comments and via social and email.
  1. The number of followers a Twitter account has can have a very large influence on the number of clicks, retweets, or favorites a tweet receives.
  2. The type of content, day of week, and messaging can also influence engagement.
When analyzing our tweet data, I did not control for these extraneous factors, and I’m sorry for the confusion this has caused. Analyzing the data as-is means that, when you see a result of 2:00 a.m. to 3:00 a.m. as the hour in which tweets receive the most clicks, this is an average amount of clicks for all of the tweets in that particular timezone.
This is an important consideration for a couple reasons:
  1. It means that large accounts with many followers have an unusually large influence on the average amount of clicks per tweet.
  2. Especially in the early morning hours, these large accounts have a disproportionate effect on the average amount of clicks because of both the high follower count and the relatively low tweet volume from many of the other Twitter accounts.
For example, the graph below shows the total number of tweets posted from accounts in each follower tier in the Eastern European Timezone (EET).
Twitter Timing - CET broken down by follower tier
The largest accounts, those in tier T06 with over 5,000 followers, tweet far more often than all of the other accounts and tweet mostly late at night, when other accounts don’t tweet as much. This raises a red flag in my mind. What I could do here is segment the data, so that data from the large accounts doesn’t confound the findings from the rest of the accounts, which may represent a more typical Twitter user. (I did this below, if you’d like to take a look!)
Another great thought from those of you who read the post is that we could look at the data with another convenient statistic, the median (the middle value in a series of values), to compare the amount of engagement received by tweets sent out at different hours of the day.
The main reasons I chose to stick with the average is that so many tweets in our dataset received no clicks, retweets, or favorites. As a result, most of the medians for the different hours of day turned out to be 0!
More complicated transformations of the data could have also been employed, but the use of such methods make the findings a bit more difficult to interpret, and we’re very keen to provide actionable insight to all of our readers!
I totally understand that, while including all of the accounts gives a full view of all of our Twitter data, it means that some of the data might be skewed by these large accounts tweeting in unusual hours. I definitely don’t want skewed data, and I want to be fully transparent about this and what we’re doing to correct this.

The best time to tweet for clicks: An alternate way of looking at the data

To control for the effect of the large accounts on average engagement, I looked at the total number of tweets posted from accounts in seven separate follower tiers as well as the average likes received from those tweets.
In some cases an outlier in the data was also rather obvious. For instance, seeing tweets from accounts with fewer than 100 followers receivve thousands of clicks on average raises a red flag in my mind that it could be an outlier. In other cases, I found that large accounts tweeted much more than other accounts during unusual hours and received an unnatural spike in average clicks.
I found that it might be useful to remove these segments from the data.
An example of a case such as this can be seen in the graph below.
Twitter timing example - outlier
After filtering out accounts with a disproportionate amount of weight that were skewing the data, here is what we found. I’d love to hear your feedback and would be happy to respond to any questions in the comments! :)

Best Time to Tweet for Clicks, Worldwide: 6:00 to 7:00 a.m.

(excluding outliers with abnormally volatile average click counts)
  • Los Angeles, San Francisco, etc. (Pacific Time): 10:00 p.m.
  • Denver (Mountain Time): 7:00 p.m.
  • Chicago (Central Time): 2:00 a.m.
  • New York, Boston, Atlanta, Miami, etc. (Eastern Time): 2:00 p.m.
  • Madrid, Rome, Paris, Berlin, etc. (Central European): 5:00 p.m.
  • Cape Town, Cairo, Istanbul, etc. (Eastern European): 5:00 a.m.
  • Sydney (Australian Eastern): 1:00 p.m.
  • Hong Kong (Hong Kong Time): 5:00 a.m.
  • Shanghai, Taipei, etc. (China Time): 7:00 a.m.
  • Tokyo (Japan Time): 8:00 a.m.

Over to you: What are your takeaways?

We’re so grateful for the chance to dig into the stats from the many tweets that people choose to share with Buffer. The data is super insightful, both for sharing with others and for impacting our own social media marketing plans!
What did you notice from the stats here?
Did any of the results surprise you or get you thinking about your plans in a different way?
I’d love to hear your take on this! Feel free to share any thoughts at all in the comments!
Image sources: IconFinderBlurgroundsDeath to the Stock PhotoUnSplash
To view the original article Click Here

No comments:

Post a Comment