A common question we get is “Why don’t you credit more Twitter users in your Tweets?” The main answer is because we rarely find tweets through Twitter users.
One reason we don’t read a lot of tweets to find interesting things to share is that Tweets don’t show the date the information linked to was published. So 5-year old articles can be tweeted next to 5-minute old articles and you can’t tell the difference.
But also, sometimes a user thinks that because they tweeted about it before we did—in some cases including our username in their tweet, such as “/cc @TweetSmarter”—that we found the tweet from them. We didn’t. When we DO find an article because of a tweet, we credit the person(s) who (re)tweeted it.
Tip: If you’re going to suggest a tweet, at least read our tweets to see if we’ve tweeted it already. We regularly get people asking why we haven’t retweeted their suggestion—when it was something we had ALREADY tweeted.
Where do we find things to tweet?
UPDATE: In 2013, we source a lot of tweets from customizable services such as Flipboard and Zite.
We don’t read every tweet with our username—we get too many retweets (100’s of thousands/year) and spammy callouts to make it possible to look at them all. We find most of the posts we tweet from custom searches looking for recent, relevant Twitter articles. Here’s an example of one of our Google searches, set to search blogs only for the last 24 hours:
twitter | tweep | tweet “how to” | ~tip | ~tutorial | ~help | ~strategy -money
We do searches on a variety of services besides Google, such as Topsy.com.
In fact several times a week, we add someone’s username to a tweet we have already found before we saw their tweet. That’s right—we give credit to someone who had nothing to do with the tweet. We do it because it actually saves us time from having to respond to a complaint later. Also even if there would be no complaint it prevents someone being disappointed.
What tweets do we look at?
In rough order of what we look at first:
- Tweets that may include a question to us as described here
- Tweets addressed to us, filtered to exclude certain users. Why filtered and not blocked? Among other reasons, some users who appear to have mental disorders sometimes write us many tweets all at once. (Yes, we have experimented in the past by replying to them.) Others are just “celebrity baiters”—always addressing miscellaneous banalities to users with a lot of followers.
- Tweets that include our username that we just happen to notice. We scan other tweets with our username as time allows, and occasionally notice things that we respond to.
How do we schedule tweets?
Although many tweets go out within minutes of our finding them, typically 3-5 times a day we schedule a group of tweets.
On average, a tweet is scheduled 3-4 hours before it is actually tweeted. Only a few tweets are ever scheduled more than 12 hours in advance, and we very rarely schedule anything more than 16 hours in advance. We have tried many programs for scheduling tweets, including but not limited to HootSuite, SocialOomph, EasyTweets, FutureTweets and others. We sometimes use multiple interfaces to get around downtime on any one service.
Tip: You may want to check out 15 Free Twitter Tools For Scheduling Future Tweets.
Our strategy for scheduling tweets
Some of the guidelines we follow include:
- Urgent or extremely timely topics we try to tweet immediately (such as info about new Twitter errors or downtime)
- We try to tweet what is likely to be most popular at the time most people will see it. The time that something will be popular is completely dependent on our research of when OUR users respond, not when Twitter as a whole is most responsive. So a “hot tweet” that we find at a slow time of day we will delay tweeting until more people are likely to see it.
- We sometimes delay our tweeting about extremely popular topics until a time of day that is well-read in another country. For example, if interesting but non-timely information comes out via @Mashable late afternoon U.S. time, we have found that our tweeting about it right away is redundant—most of our followers checking tweets at that time have already seen the story. We even get complaints! So we wait a few hours until more people are checking tweets in Australia and post the information then.
- Tweets that got a lot of clicks or retweets we will sometimes repeat once (prefaced by “r/t “) at a less busy time of day within 8-16 hours after the original tweet went out.
- Certain kinds of tweets (highly technical, business-oriented, humorous) are more popular at certain times of day, so we sometimes take that into account.
What about crediting authors of articles?
We rarely include the username of the author of a post in the tweet itself because we are already linking to the author’s article. Authors who want their usernames visible to readers should make them visible to readers of the post, and not rely on others to find and include their username in tweets. In general, anything that makes tweets shorter makes them get retweeted more frequently (which is what is most beneficial to the author anyway).
For more ideas about who to credit and when, go here.
Should we suggest articles for you to tweet?
That’s actually a tough question to answer. We realize that with over 200,000 followers, MANY people are likely to have tweeted about something before we do.
First, realize that we only want tweets that are less than 24 hours old, preferably less than two hours old. But, people say, what if I have found something great that is a week old. IF you have looked at EVERY ONE of our tweets in the past week to KNOW that we have not tweeted it, and IF it is really a great tweet for us, then fine. But we’ve found people just send stuff and say “I don’t think you tweeted this.” In EVERY case so far, we DID already tweet it. EVERY case. Sometimes it is a very popular tweet that we have already planned to retweet once or twice more in upcoming months and so we tweet it and credit the person suggesting it, even though we already found and tweeted it several weeks previously. Again, we prefer harmony rather than explanations or rejections where possible.
Second, we only want suggestions for tweets that are urgent and timely, or very clearly about helping people use and benefit from Twitter. Yes, we tweet about social media in general and issues that affect large numbers of internet users. But we don’t want any help finding those tweets! Those are either easy to find because they are so generally popular, or sometimes just barely significant enough to be tweeted.
Third, we ALWAYS have some tweets scheduled in advance. The odds of you finding something that we haven’t seen or heard of already are not as high as you might think. Your best chance is to tweet us something just before we are doing one of our daily searches. But there is no set time we use, so it’s random luck when that happens.
But what if you are already sending us tweets? A few people do, and though it usually doesn’t help us, we mostly maintain a gracious silence. There are about one or two people in any given week that send tweets we are happy to see, totalling usually only 3-6 tweets altogether. One user is always sending us his own off-topic articles and saying things like “This is right up your alley!” Dude—it isn’t. And you’re obviously just self-promoting. Drop the hype. Why don’t we block him? Well, we have blocked many users like that. His is a borderline case. Again, this is someone we’ve given feedback to and he just doesn’t listen. We may yet block him.
We have tried a few experiments asking people to find tweets for us. Each time it has created more, not less work for us. It’s also a lot of work for our tweet-finders. They have to at least click and look at every one of our tweets to see if we have already tweeted something, unless they have found something ultra-current. What we have learned is that there are always people in any group who listen poorly.
We had one user we contacted six separate times (one document of guidelines, four emails, two DMs) about serious errors they were making in finding tweets for us in one of these programs. Errors such as suggesting things we had already tweeted, irrelevant tweets that had nothing to do with Twitter or relevant social media, etc. They never changed their tweets and we ended the tweet-finding program in order to stop dealing with this one person. Why didn’t we just end their participation? Because it was clear they weren’t listening, and any explanation we could give about why they were being rejected was also unlikely to be heard clearly. With an account as large as ours I’ve found maintaining harmony simply saves time. We are still planning to restart a program of folks who help us find tweets…but we won’t be inviting them to it.
Are we being a little too “nice?”
We have tried being more direct. We have blocked users who don’t listen when we ask them to stop what amounts to harassing us. We have sent clarifying emails, had others talk to them on our behalf, etc. We have repeated ourselves over a long period of time before eventually blocking or being more and more blunt. We have found that some people Just. Don’t. Get it. And more than once, we have groups—yes groups—of people contacting us saying “Why did you do that to ____?” Or hearing people spreading misinformation. Having experienced that enough times, we decided to be nice to almost everyone—because it’s easier. We do LIKE being nice to people, but we like even more making our lives easy (and peaceful).
And to those of you who makes our lives easy—we appreciate you more than you will probably ever know.