Making Activity Streams More Manageable

February 3 2012 02:00:00 PM Comments Disabled
Activity Streams - Jambalaya or Steak and Potatoes?

In my last post Don't Cross the Streams, I challenged the idea that integrating multiple sources of information into activity streams is a good thing. This struck a nerve with some people (mainly vendors) while others completely agreed with me. A friend of mine an fellow industry analyst suggested that I follow up by posting possible solutions, so below are the few of the areas I think can help:

1. Don't put stuff in there that doesn't belong
2. Display multiple streams
3. Allow people to organize lists
4. Provide several notification options
5. Manual filtering
6. Crowd (friend) sourced curation
7. Automated filtering via analytics


Silos Are Not Necessarily A Bad Thing


The first thing that will help reduce the noise in activity streams is to stop pushing everything and anything into them! When Facebook introduced frictionless sharing into the Ticker, 800M people revolted. Hopefully enterprise software will learn from this and vendors will realize that the benefits of open sharing can be quickly overshadowed by the annoyance of irrelevant information and the hindrance of context switching. Not everyone wants to see status updates in the same place they see support tickets or new sales opportunities.

Mission Control


If we are going to continue down the path of taking dozens of different pieces of information and cramming them into one place, then a single stream is not the way to go. Most Twitter clients have helped reduce the issue by enabling people to view multi-columns of streams at the same time. (popular examples: TweetDeck, Hootsuite and Seesmic) Columns are set up to display information from specific people or groups, keywords or search terms. I've like to see the enterprise software vendors offer multiple stream UIs.

Groups Are Not Lists


Google+, Facebook and Twitter also have features where you create subsets of people (named circles or lists) and let you switch between streams which only display posts from those. Most enterprise products like Yammer, SocialCast, Jive and Socialtext enable the creation of groups, but these are not the same as lists. A group is public (at least to it's members) and controlled by an admin. They are not ways of personally organizing people into buckets that mean something specific to you. For example, there is a big difference between a company's "Marketing Team" group and an individual's "Folks That Inspire Me" list. Google+ even allows people to share their Circles. I'd like to see enterprise software vendors offer personal lists.

Notification


While streams are great (or perhaps aren't) most employees don't sit and stare at them all day. In order to make sure they don't miss important information there should be options for being notified via email and/or SMS, as well as a dedicated feature in the stream UI that shows notifications. There should be notification options for things such as posts you're mentioned in, when people respond to your posts and when topics you're interested in are being discussed. Notification frequency can range from immediate to daily digests. Yes, I know these notifications just create more email, but for most people that is still a very necessary evil.

Manual Filtering


It is true that a majority of people will always use the default settings for any software product. However, it is important to provide options for those that want to change what information they see in streams.

TweetDeck does a great job enabling filtering at the global level (ex: I don't ever want to see Tweets with the #SXSW or #SuperBowl hashtag) and for each column. The filter can be either include (+) or exclude(-) which I find really simple but at the same time affective.


TweetDeck Filters


Enterprise software vendor Tibbr has an excellent filtering system, where people can easily set up their own streams based on a combination of parameters such as subject, sender or time. You can give each stream a name, set up the rules and viola, they are available for a list on the left side of the screen.


Manage Filters In Tibbr


Bottlenose provides a similar way of creating your own custom streams, but takes it step further by offering "guides" to help you set them up. Bottlenose's filters can also be extremely advanced, offering metrics-based parameters such as number of posts or relevance. These folks are way ahead of the game and I hope that enterprise software vendors start to offer similar analytics-based filters.


Bottlenose


Socialtext provides a nice balance of powerful vs easy to use filters, enabling people to select from a set of choices at the top of the stream. I particularly like the "My Conversations" filter which displays just the threads started by you and the ones you've responded to.


Socialtext Activity Stream Filter


What's Popular?


One of the main issues with streams is that information often flows past the bottom of the screen long before people see it. To help with this streams need to offer sorting options other than just chronological.

Facebook does this by providing a choice of Most Recent or Top Stories. The problem is perhaps 3 people on the planet have any idea what logic goes into actually making something a "top story". According to Facebook help "We determine whether something is a top story based on lots of factors, including your relationship to the person who posted the story, how many comments and likes it got, what type of story it is, etc." I really like this in theory, but often question the results.

NewGator has a nice feature where you can display the activity stream as a heatmap. Colour coding is used to display the most active conversations, which you can then click on to drill down into more detail. The heatmap view can also be refined with standard filters such as author, hashtag, etc. The combination of the two methods makes this a very powerful and also fun way to view the stream.


NewsGator Heatmap


Another options is to sort streams based on the number of Likes that posts have.

Automated Filtering Via Analytics


Some enterprise software vendors are already implementing "smart filtering" options similar to Facebook's Top Stories mentioned above. For example, Jive Software has a "What Matters" stream which their help describes as "an engine that collects and analyzes user activity in the community to recommend useful content, people, and places to individual users, and reports trending content and people. The Recommender looks at business relationships, user expertise, and areas of interest based on a user's behavior in the community to suggest relevant content that Jive knows a user has not yet seen." IBM has told me they are working on similar smart filters for IBM Connections which will leverage the power of their Cognos analytics tools.

I do think automated smart filtering has potential to help people navigate huge amounts of content. However, I remain skeptical about an algorithm's ability to decide what I should, or more importantly what I want to see, especially at a specific time or context. If algorithms are going to be used (which at this point we know they will) then I hope ("fingers crossed") that vendors at least allow power-users to tweak the parameters that are being used. This does not have to be rocket science. The reason I use the Zite newsreader on my iPad instead of Flipboard is that Zite provides an incredibly simple way for me to help determine which stories to display.  On the right side of each story I can say whether I like the story or not, and if I want more content from that source, that author or various tags. I'd love to see enterprise software do this as well.





Summary


I don't want people to misinterpret my latest posts as "Alan hates streams." I don't. I'm just trying to paint a realistic picture of some of the challenges the present and hopefully provide some guidance as we move forward.  Remember, there is no one right answer. Sometimes things are better all jumbled up and sometimes they are best when consumed on their own.


Jambalaya Man gönnt sich ja sonst nichts [TM]