Saturday, May 22, 2010

Library twitter account - what tools are you using?


Introduction


By now Twitter accounts for libraries are old news, I  run a Twitter League of Library accounts, and it easily topped over 600 accounts  (Note the service is quite unreliable , statistics reported there often lag, so I'm not actively maintaining it, besides adding libraries that request to be added). For more libraries on Twitter lists see Twitter lists by Lindyb for Academic libraries &  Public Libraries 

While there have being short pieces discussing what libraries are doing with their Twitter accounts (see for example this early piece), as far as I know there is little describing what work-flows are put in place for managing Twitter accounts.

Moreover, due to Twitter's open ecosystem there are hundreds of Twitter related services that have sprung up that enhance the basic Twitter experience, so there are many possible ways to run a Twitter client for your library beyond just going to Twitter.com and answering from there.

Of course, what tools or services you eventually end up using should depend on the circumstances of your library. I'm no social media guru, so I don't pretend to be an expert on this, as usual this is a long rambling post trying to clarify my thoughts.

There are many issues to consider when managing a Twitter account for your library but  the way I see it, the answers to the following questions will affect the tools you will use and vice versa.


1. What are you using your twitter account for?


2. Will you be tweets updates from other accounts/services or manually updating every tweet?


3. How does Twitter fit into the library's Social Media stream? Will Twitter updates be pushed to other accounts (e.g. Facebook) ?


4. How do you want to be alerted to and respond to Tweets?

5. How do you track click-through on links tweeted?


6. How do you archive tweets?




1. What are you using your twitter account for?

Clearly, not every library uses Twitter in the same way. For instance an account such as
NLC_Reference which is geared to tweet reference questions received is clearly going to be handled differently from one like grpl which not only pushes library news updates, but is also used as an account to proactively scan for tweets about the library and to respond with them.

And of course, these differ from Twitter accounts used primarily for internal communication.

If for instance, you intend to use your Twitter account to proactively scan for tweets and to answer them, you definitely need to use tools that can pick up such tweets in real time, compared to a twitter account that tweets only reference questions.


2. Will you be tweets updates from other accounts/services or manually updating every tweet?

At the very least, most libraries use Twitter as a broadcasting channel. Typically library news, as well as blog posts are tweeted.

While one can manually tweet these updates, I believe many libraries have chosen to automatically push these updates.

There are many ways to push auto-updates, but the easiest ways involve getting the information you want to push in RSS format, and pushing it through Twitterfeed , a free service that allows you to push RSS streams into your Twitter account at hourly intervals.



RSS is pretty much ubiquitous these days, so one can push all sorts of information into a Twitter account (See also
Using RSS feeds to distribute library news - 6 ways) . RSS feeds of recently added books would just be one example. Another example would be recently deposited thesis in your Dspace repository.

That said, I'm not quite sure if it's such a good idea to use such auto-updaters. Firstly, while such auto-updates are a convenience, there is a danger that this can lead to tweets that have little context.

For example, if you are feeding your Twitter account with a feed from your library blog, typically only the blog post header will be pushed with the link. As such the blog header has to be specially crafted to make sense to the users following your Twitter account.

Some services like Twitterfeed  allow you to perpend descriptors like [Blog] which helps.

Also manually crafting a tweet allows you to create more effective tweets, this not only includes tricks such as reducing message length to make it easy to be retweeted (less of a concern with new style retweets), but also allowing you to put all information in the tweet.

Take for example, a blog post announcing change of library hours during the vacation period. If you manually construct a tweet about this, it is likely you will try to squeeze the information into 140 characters. But if the information is coming from say a blog post, chance are the blog header will be only something like "Change in library hours", which will be pushed, and only by clicking the link will the user then see the new opening hours on the blog.

Not the end of the world, but a convenience for users who are receiving tweets only via sms, and have limited access to mobile web.

Another danger of auto-pushing updates is the lack of control. If this is overdone, you can cause a flood of tweets (though some services like Twitterfeed have a throttle switch that smooths out updates ), and gives the users the impression that "there is no one at home" and everything is on autopilot so the user won't be encourage to tweet you questions.

This isn't a problem of course if this is exactly what you want to accomplish.

Do note, there is no reason why you can't combine both, for example using Twitterfeed  to tweet blog posts, while manually updating when necessary to show that there is a real human behind the account!


3. How does Twitter fit into the library's Social Media stream? Will Twitter updates be pushed to other accounts (e.g. Facebook) ? 

Most libraries today are on multiple web 2.0 accounts. At the very least, most are on Facebook, YouTube  & Flickr. Many are also on Delicious, SlideShare etc on top of being on Twitter.

This means that there is quite a bit of social activity/updates that can be pushed out to users.

In fact, there are a bewildering number of ways to push and aggregate updates from and to these accounts.

A simple minded approach would be to push these updates all to Twitter.

While one can of course use the already mentioned Twitterfeed, many of these services can link to Twitter natively already without external services. For example you can link your YouTube account to Twitter, so it automatically tweets when you upload a video. 



But for a more general solution you can experiment with lifestreaming services like FriendFeed &   Posterous that allow you to aggregate all these accounts together and/or push to Twitter.



You can choose to aggregate various services into  FriendFeed  and choose to push them into Twitter. See my blog post "Libraries on Friendfeed" for more details.


Of course there is no reason that Twitter has to be the end point of all activity, one can easily push Twitter updates elsewhere as well.


In particular, with facebook being so dominant these days , a facebook centric strategy, where Tweets are being sent to the facebook fan page rather than vice versa might be a good idea.


This needs quite a bit of thinking since there are many possible ways to setup the  stream of information and if you do it wrongly, you end up with the same updates appearing multiple times.

5 Good Examples for Managing Your Lifestreaming Data Flow  shows how 5 top social media gurus organize the flow of information generated with their social media tools. Below shows one example.


Source: 5 Good Examples for Managing Your Lifestreaming Data Flow
 
For a library related example see how cmbUMCG does this.


Technically Posterous is not so much a lifestreaming tool  as a tool that allows you to post updates through multiple means (email, sms, bookmarklet, IM) to multiple channels at the same time (different blogs, twitter, facebook etc).



4. How are you alerted to Tweets? How do you respond to Tweets? 

Chances are though even if you use a method to autopost tweets, you probably also want to use your Twitter account to handle queries and even if you don't encourage this, you probably still need a system in place to be alerted when a tweet is directed to your account whether via reply or direct message (if you follow accounts).


Some options (there are literally hundreds) include

1. Desktop clients - Tweetdeck, Seesmic Desktop etc

2. Web-based clients  - Twitter itself, Brizzly, Seemic web, CoTweet, Hootsuite, Postling etc

3. Instant Messenging clients that handle Twitter - DigsbyPidgin , Trillian , LibraryH3lp

4. Misc - TwInbox (works with Microsoft Outlook), Echofon for Firefox/Twitterfox (Firefox extension), LibAnswers 



There are many more of course, but these clients essentially alert you almost instantly whenever you receive a reply or direct message,  and allows you to respond.


I have left out the mobile Twitter clients, because I doubt librarians will be answering tweets on mobile phones, not officially anyway.  Except perhaps those doing sms reference service with Android phones?

Tweetdeck is probably the most popular Desktop Client with power users, allows you to monitor multiple accounts at the same time. You get near real-time alerts when a reply comes in, or when a search alert is met. The last can be handy if you proactively scan for tweets about the library and to respond with them. Do note that like almost all Twitter clients, it is not capable of  searches that pick up tweets based on location, so you can't pick up in real time, for instance a tweet that mentions library that is within 1 km of your location.  




Another feature that you might want to look for in your choice of Twitter clients would be the ability to schedule tweets.  This can be very useful, for example, you might learn of say opening hours on Christmas eve , in November, you can tweet it once now, and then schedule a reminder tweet a week before Christmas.

While there are standalone tools that do this , I personally prefer a client that has this built-in, rather then using another tool. Currently the web-based Hootsuite and CoTweet has this. Tweetdeck  also recently added this feature.




BTW  Hootsuite and CoTweet are full blown enterprise tools that facilitate handling of Tweet accounts by multiple team members.  For example one can schedule shifts, assign tweets to answer by team members,etc. Personally I think most libraries probably don't need this level of complexity.

I would also add that many of these clients such as TweetdeckSeesmic Desktop ,Hootsuite and CoTweet, Postling are designed to monitor and manually post to more than just Twitter.

 Tweetdeck for example does MySpace, Linkedin, Facebook, FourSquares and Google Buzz. Services like hellotext and Ping.fm are even more capable allowing you to update dozens of services at the same time, but are purely update tools as you don't receive updates.


Of course, the problem with most of these clients is that library staff will have to learn another new interface to use (and some like Tweetdeck, or Hootsuite (see below) have pretty complicated interfaces), is there anyway to avoid this?



Of course, by default, direct messages are sent to the email account your registered your account with. So if all you want to do is to be alerted of direct mesages you don't need to do anything.

There are services that do the same for replies and alerts (such as Tweetbymail and Tweetymail), where an email will be sent when a reply is directed to your account, but personally I'm not sure how reliable they are. Besides even if you are alerted to a reply sent to your twitter account through email, you still need to be able to reply to it via Twitter.

Perhaps a better solution would be SpringShare's LibAnswers FAQ system which allows you to receive and answer tweets directed to your twitter account.


For those unfamiliar with this, it is a FAQ system designed for libraries. Users who query the faq system, will be matched to the best possible item, if no item is available, the user is invited to submit his question via an electronic form. The librarian will be informed via email or instant messenging, enters the answer in the LibAnswers system (web based) and in addition to emailing the user with the answer can also post the item as a FAQ.

On top of this, LibAnswers  also supports Twitter.

When someone tweets to your library account, it shows up in the "unanswered" tab and an email will be sent to an account you choose alerting you that there is an unanswered question, exactly the same as if a question was posed via the LibAnswers system.



The nice thing about this is that the librarian can choose to answer the question exactly as if it's a normal question sent via the electronic form without knowing anything about how Twitter works. When the librarian clicks submit, the user will be tweeted a url to the answer submitted by the librarian.



While this allows answers longer than 140 characters, I personally feel this can be quite frustrating if the user is limited to SMS, and especially if the question is a short one, e.g "Is the library open tomorrow?"

Also note that tweets that are direct messages to the account will not be picked up! So this isn't a complete solution yet.

Another way to avoid learning a new interface would be to use TwInbox 

TwInbox  is a addon for Microsoft exchange, since many libraries already use Outlook for email, using TwInbox , allows you to reduce the clutter in terms of additional programs and you can happily receive and tweet from your email client, so this is definitely a option.


So you can assign someone who is already manning the email box to handle twitter queries at the same time.


In the same vein, many librarians are fielding queries via Instant Messenging anyway, so one possibility would be to use DigsbyPidgin , Trillian , LibraryH3lp which can receive and update Twitter as well! The picture belows show how Digsby handles Twitter. 





Sadly, Meebo doesn't do this yet, while Textalibrarian allows you to post a question/answer via Twitter only, though they seem to be working on http://www.tweetalibrarian.com/ , not quite sure about OCLC's QuestionPoint whether it supports Twitter yet.

Incidentally, regardless of the client you use to tweet answers, you probably need to have a policy on the expected response time. Is a tweet more like an asynchronous channel (e.g email) or is it synchronous (e.g Instant Messenging) ?


5. How do you track click-through on links tweeted? 

You probably want to measure how  successful or impactful your twitter account is. Besides looking at number of followers and a slew of other twitter influence stats, you probably need to measure how often your links are being clicked. If the endpoint is a library page, you can probably catch it in the logs at the other end, but a surer way would be to do your own tracking of links on your end.

Probably the easiest way is to use bit.ly or other url shorteners with built-in analytics, so you can see who is clicking on the links you tweet. As I write this, bit.ly  is the current industry leader, and you can use it in almost every major twitter client including Tweetdeck. Do remember to register your own bit.ly account and add the API key to the twitter client you are using, so tweets are tracked using that account.  Twitterfeed supports it too.

Depending on the method you use to autopush tweets, they may use their own url-shorter.

Things get a bit tricky if you use  one way to auto updates and combine it with manual updates/replies, since they  may use different url shorteners or ways to track clicks. For instance, you can use Feedburner to push your blog posts (socialize service) , but use Tweetdeck + bit.ly for manual updates, in which case, you will have click-thru statistics in two places.

If you are using lifestreaming tools like Friendfeed which does not track clicks, you can first burn a RSS feed with Feedburner (which tracks clicks) before passing it through Friendfeed.



6.  Preservation

Want to keep track of all the tweets you have made? See a nice compliment you want to keep? Want to store all the questions you get via Twitter? If you think it's all there in Twitter and searchable any time you want them, you will be sadly mistaken, since only the most recent tweets is searchable at any times. While there are moves to archive this ( Library of Congress) ,  and Google provides you with a way to find older archives, the former is open only to researchers and the later isn't fully reliable yet.

Some of the Twitter clients above such as TwInbox and probably Hootsuite etc will store all your tweets but some like Tweetdeck only show the most recent tweets. 

Anyway as good librarians we should know the value of archiving right?

I'm still researching this part, and as usual with anything associated with Twitter there are lots of options, see the list here "10 ways to archive your tweets"

Simplest seems to be just plugging in the Twitter RSS feed into Google reader, though I'm not sure if it grabs private messages.


 7.  Conclusion

As you can see, there are many ways to

a) be updated of a tweet
b) respond to a tweet
c) auto-push tweets
d) archive tweets


I've being experimenting with different methods, one possible method would be to use

Twitterfeed + Tweetdeck + some way to preserve tweets

The advantage is that both of them, allow you to use bit.ly for analytics.  Tweetdeck  is also probably the most capable Twitter client, allowing you to watch real time alerts. Until recently, it lacked the ability to schedule tweets but in the latest version this was remedied which makes it a almost perfect Twitter desktop client in my view.

That said, Tweetdeck doesn't store all tweets, so some other method to archive tweets will need to be used. 


I suppose if you have a full blown social media strategy for your library, a good choice would be something like CoTweetHootsuitePostling which are seriously powerful tools for big corporations that need to have and manage a heavy social media presence.

A more traditional all-round library approach would be to use LibAnswers , or LibraryH3lp which allows you to treat tweets like queries sent by other channels such as email, IM from users


Now it's your turn, please describe briefly in the comments what method you use to be manage your library's Twitter account.  Would love to know what you are doing in terms of these issues, so we can compare notes. 



Thursday, May 13, 2010

Environment scanning for libraries - Facebook

In an earlier post ,  I talked about using Twitter, Google and other tools to scan for mentions by users of libraries online. I noted that Twitter in particular was effective due to its real time nature, and that unlike Google alerts, Twitter hits tend to be far more likely to be relevant, as the former often contained false hits, compared to a 140 character tweet which are more focused.

3 Techniques were suggested to find relevant tweets

1) Keyword match -e.g. NUS Library
2) Geolocation - e.g. Library near:National University of Singapore
3) Filtering based on person Tweeting - e.g. If user is following you and tweets about library, it's probably about you.

See earlier post for more details

Since then, I have refined my search terms to further improve accuracy (recall and precision in librarian speak), but in general that still remains my latest thoughts on the subject.

I've improved my technique slightly though for Facebook scans. As Facebook pushes their agenda to make status updates more public, more and more public data is available. The method I suggested in the earlier post I think can be improved, particularly after the launch of  Facebook Open Graph and a slew of other changes including Facebook’s connections and launching of community pages  (See post on implication for libraries)

 For one thing, Facebook's open graph API means a direct search of public Facebook content becomes a lot easier. Why rely on third party indexing by Google, Bing, Social Mention etc when you can directly query Facebook?

 Booshaka for instance is really effective. 




There's no RSS feed for this though which is a disadvantage as you can get all your other searches (Twitter & Google alerts) in RSS.  

 Open FaceBook search is similar , though they don't provide RSS feeds, they allow you to embed widgets for each search.

So if you use something like NetVibes, iGoogle, or even LibGuides for monitoring RSS feeds, you can use this as a widget (Use html widget).


Example below shows it in a unpublished LibGuides 


I suspect NetVibes, iGoogle is probably more suitable actually but I'm just testing. Besides this tab. I have one tab for RSS feeds of Twitter searches, one tab for RSS feeds of Google searches, one tab for RSS feeds of official library web 2.0 accounts (Blogs, Youtube, etc).

Essentially creating a information dashboard of the library's presence on the web.

At this point, I'm still undecided if this dashboard approach (using dynamic startup pages like Netvibes, iGoogles, even LibGuides) is better compared to putting everything in a straight RSS feed reader, particularly Google Reader. The former is better for viewing, but the later is better for long term perservation I think. But in this situation, as there is no RSS feed, you can't use Google reader.

I just noticed that Guus van den Brekel, Coordinator of Electronic Services, Central Medical Library (CMB) , Netherlands, posted something very similar ,he uses Netvibes.

He is using the real-time search engine Socialmentions to scan , which was what I was doing as mentioned in an earlier post (I also sugguested Bing which supposedly had more access to Facebook compared to Google). But I find Social Mention misses a lot in Facebook (evenh when you restrict SocialMentions to searching only Facebook) compared to the two Open Facebook Search or Booshaka  above.

I suspect SocialMention, Google and even Bing (despite what has being stated) is searching only fan pages and not all public status updates.

Compare say this search on Socialmention versus this on Facebook Open search.

Anyway, I think Facebook is important enough that you should have a dedicated scan for it.


 
Lastly, with the launch of the Facebook Like button, you can now see how many users have liked content on your webpage. My library doesn't implement the like button or even have a Facebook fan page, despite that you can still see that users have liked some content on our domain.


Again there is no RSS feed, but you can put it into Netvibes, LibGuides etc, by following instructions here.



That's all for now. Hope you found it useful.

Share this!

Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License.
Related Posts Plugin for WordPress, Blogger...