24 Usable Hours » ROI

The Epic Facebook Experiment

Devon Smith — Wed, 13 Apr 2011 16:05:58 +0000

For months now I’ve been wanting to do an experiment on Facebook but haven’t had the time. The Daily Beast finally beat me to it with their Cracking Facebook’s Code article yesterday, which shows essentially:

If you don’t have a lot of friends, you won’t show up in the news feed of your friends
If your status updates don’t have a lot of likes or comments, you won’t show up in the news feed of your friends
The type of status update you have (photo v link v video etc) matters in getting into the Top News feed of your friends
Even the Most Recent News feed has an algorithm (isn’t purely most recent)
Stalking your friends won’t get you noticed (visiting a friend’s page won’t make you show up in their feed)
Having friends who stalk you will get you noticed (if other people visit your profile a lot, you’ll show up in their friend’s news feed
Links show up in the news feed more than plain text
Photos and videos show up in the news feed more than links
Status updates with more comments or likes are more likely to show up in the news feed of your friends
Even if you do all of the above, your status updates still might not make it into the news feed of your friends (aka, there’s more to the algorithm than just these 9 factors)

Their experiment had to do with profiles, which I think follow a different but similar algorithm (aka EdgeRank) than pages, but still immensely relevant, and an interesting model for an experiment.

Then today, MuseumNext tweeted

And suddenly I had a revelation. It took me 5 minutes to figure out what Pages I’m a fan of. It turns out, about 50 different pages, only 10 or so of which I could have recalled off hand. Here’s why: they never show up in my news feed. Maybe not never, but VERY rarely. As opposed to Twitter, where I KNOW what brands I follow because they show up in my stream much more regularly. One big difference is I have a thousand friends on Facebook, but only follow 250 on Twitter. But my ratio of brands:people is much higher on Twitter. And if I follow you, and you tweet daily, I’m going to see your tweet. Not the case with Facebook. Of course, this is a double-edged sword since it’s much easier to “unfollow” someone who’s tweeting too much, than it is to “unlike” a page who shows up in your news feed too often. You’re probably more likely to push the “hide” button on your news feed, rather than take the trouble to go to that page, find the unlike link, and click it. So here’s where we end up:

On Facebook, I’ve got 50 pages that I “like,” which is about 5% of my Facebook network, but for all intents and purposes they’re invisible because they never show up in my news feed. And they’ll stay invisible forever, because I won’t go through the trouble of un-liking them. Their only residual value to me seems to be when I click on a page that I’m considering “liking,” I am always curious to see how many of my friends already like that page, and this may sway me into pushing that like button.

On Twitter, I follow about the same number of brands (but how to account for the people I follow because they work at institutions I’m interested in hearing about?), but it’s a much higher percentage of my network–20%–and I see them all the time. My churn rate on Twitter is likely much higher though. I’ll try out following you for a bit, if I don’t like your tweets, how often you’re tweeting, etc, I’ll unfollow pretty quickly. And that same “residual value” exists in Twitter’s “you both follow” and “also followed by.”

There are some interesting implications here in the difference between Facebook fans & Twitter followers, in how we interact with these audiences. It turns out fans with small networks are really valuable to me on Facebook (I’m more likely to show up in their news feed, all else being equal they like or comment on my content, that increases my probability of showing up in the news feed of other unrelated people). It turns out followers with large networks are more valuable to me on Twitter (if they only follow a few people and/or successfully utilize private lists). If you’re visual like me:

Now, there’s all sorts of caveats. My propensity of sharing links on Facebook versus Twitter. The likelihood of my friends versus followers being interested in the link that I’m sharing. The number of “spam” accounts following me on Twitter versus the number of Facebook friends who no longer use Facebook. The network size of the “average active user” (for Facebook the long touted number is 130 average friends, on Twitter it’s around 30 followers). How often I’m on Facebook versus Twitter. I could go on. But we’re finally starting to see a way to mathematically model the (social) value of a Facebook fan versus a Twitter follower. This is heady and exciting stuff for a data nerd like me.

And here’s where you come in! I want your help. I think it will be fairly simple, relatively painless, and really really really useful once we get the results. Here’s the idea:

On May 2, we begin a 30 day (or maybe less…thoughts?) experiment. I select 5 pages I already follow on Facebook, I write down a bit of information about them. I select 5 new pages I’ve never been a fan of (I have some sense Facebook prioritizes this, and I want to test the hypothesis), and “like” them. Then I simply go about my usual routine on Facebook, counting the number of times I see those pages in my news feed. If I’m compelled to click on their content, great, if not, that’s okay too. I don’t want to change my behavior (Heisenberg be damned). I just want to record what I do, and how that changes what I see (if at all). And I want you to do this too!

It would be awesome if they were nonprofit pages of different shapes, sizes, and colors if you will. It would be awesome if all of you were people with different Facebook network sizes and habits. It would be even awesome-er if you took notes in a form that looked something like this, and sent them to me so I could compile, analyze, and report back to you on in aggregate:

Are you interested in taking part of this experiment? Do you have suggestions for how I could change these parameters to make it more useful? Other ideas? You think I’m crazy?

Possibly related posts:

A Rant Against Facebook

A Social Media Measurement Plan

Devon Smith — Wed, 28 Jul 2010 02:42:37 +0000

Measuring social media is hard. It’s easy to be overwhelmed by minutia, there’s yet to be a measurement site that even comes close to adequately tracking people & content across social media platforms, and there are no standardized benchmarks to know if 43 (fans, comments, @mentions, whatever) is good, bad, or average. And the truth is, social media is more art than science. Not to mention we’ve become obsessive about measuring this just because we can (or at least, we think we can). Besides, when’s the last time you checked the ROI of that poster you put up, the press release you sent out, or the postcard you mailed?

Let’s talk about a new way.

Putting people front and center in your measurement plan.

What if every day, or week, or whenever, there was a list of people delivered to your inbox. These are the people who commented on your blog, or your facebook comment, or began following you on twitter, or favorited your slideshare presentation. The list would include the actual text of their comment. In v2.0, you could give values to each of these activities. As in, 5 points for an RSS subscriber, 1 point for a facebook like, 2 points for a delicious tag, etc. And the measurement tool would know the network size of each of these people. Total score is a function of the weight of the user’s action, and the size of the user’s influence base. The list delivered to your email inbox would be ordered by descending total score. In v3.0 this is not only an email, but an entire database. Search by comment keyword, by “most influential/engaged facebook fan” or how a user’s score has changed over time. This is as much a customer relationship management tool as a social media measurement tool.

But wait, that’s not all.

How is your content doing online?

You want to care about this for a few reasons:

If a piece of content is suddenly doing well, maybe there’s something you could do (like responding to a tweet, taking a photo, adding tags to a YouTube video) to increase the momentum.
If you can discern a pattern to the characteristics of the content that is doing well (or for that matter, is doing poorly), maybe you could build on the thing that’s working, or fix the thing that seems to be broken.

This list of content items would be emailed to you. You’re probably going to want to segregate new content items (recently posted) from recurring content items that continue to get pageviews or comments long after you’ve first posted them. In v2.0, you’d be able to tie the size of your network to each content item, so a blog post that you also tweet about to your 10,000 followers would have a higher score than the comment on your flickr photo that gets 4 views a day. In v3.0, if a recurring content item jumps some x% over last week, month, etc, then an emergency email would be sent to you, with the assumption that there’s probably something urgent that needs your attention. Clearly, these emailed lists would also benefit from a database that was searchable, and showed trends over time.

But wait, there’s more.

How did people arrive at your content?

The two big categories here are links and search terms. Within links, you’ve got to consider referrals, direct links, and embedded content you’ve got elsewhere. Within search terms, you’re going to want to know the search terms people used to find your content, the search terms they’re asking for on your site, and the hashtag & list terms they’re associating with you on twitter.

So this is your 3rd email of the day, week, month, etc. At this point, you’re probably asking yourself not how people arrived at your content, but why you should care. For links: having the opportunity to thank, and more generally connect with, the folks that are sending traffic your way. For search terms: to both guide future content creation decisions, and refine any SEO/SEM that you might be doing. And again, with the descending scores, and the searchable database. In v2.0, it would be nice to group similar links, and similar search terms together. In other words, I know that search phrases including the word foursquare drive quite a bit of traffic to my site, as do many different links from Beth’s blog and Ian’s blog. There are some nifty tricks in Google Analytics to combine each of these sets of links & terms, but I want a simpler solution.

Finally we arrive at the place where most folks seem to start.

What’s the size of your network?

You’ve got your people, your engagement, and your traffic. These are big aggregate numbers that are best tracked over time, and hopefully compared to your competitors.

It’s probably not the case that a subscriber to your slideshare account is ‘worth’ the same as a fan of your delicious tags. So again, it would be nice to be able to give a weight to each of these types of people, engagements, and traffic. And you’re probably going to want to see that “people” score as a total of all platforms, or by several platforms at a time. Like in Google Analytics, it would be nice to attach annotations to these aggregate trends.

It’s impossible to do all of this measurement though right? A total pipe dream. An analytics package that would be impossible to implement? A fools errand to try to connect so many platforms, so many third party tools, with so many divergent goals?

Probably.

But here’s the tools you would need to do it.

True, some of these cost money, and less than half the platforms have open APIs, but let’s at least take a quick gander at each of them. These are, of course, in addition to the many one-off possibilities I’ve mentioned previously.

WordPress Stats Plugin

*You’re going to need to install the plugin to get these stats.

Google Analytics

*GA has a TON of data to dig into. It’s easy to fall down the rabbit hole. I’ve been promising a more in depth post on this for months. It’s still in the works.

Feedburner

*of particular note: check out the publicize tab to know who’s subscribed to your blog via email.

Facebook Page Insights

*Only page admins will get access to this data.

BackType

*of particular note: this is the only way I’ve been able to figure out the amount of Facebook buzz my content gets from folks who aren’t my friends/fans.

Twitter RSS Search Query

This one’s a bit of a hack. You can “subscribe” to an RSS feed for any search term on Twitter (like @yourname). This is a great record of all of the people who mention/retweet you. Google reader will then graph for you the number of “posts” by day of week, time of day, etc.

Bit.ly

*If you haven’t already, please give TweetDeck (or HootSuite, or CoTweet) your bit.ly API, and it will shorten and track all of your links automatically. For added bonus, connect bit.ly > TweetDeck > Packratius > Delicious. More on this soon. It’s magical. You can also stalk other people’s bit.ly link stats by adding a “+” sign to the end of the link.

TweetMeme

*Especially useful to track twitter links that might not mention your name, or anything specific to your brand.

YouTube Insights

*These stats get pretty confusing. Referrals go missing, the numbers don’t add up, totally unpredictable when or how videos will go viral. But work with what you’ve got.

Flickr Stats

*Sadly, only available to (paid) pro account users.

*Just learned this one last week. If you upgrade to a (paid) pro account, you can see the actual names of the folks viewing your profile.

SlideShare

*Pro (paid) accounts get additional analytics. But I couldn’t find any examples online, so it’s unclear how much more detailed the stats get.

Foursquare Venue Analytics

*You’ll have to claim your venue to get access to this data. Do it now.

Alright, quick review

Measurement is hard. Don’t get distracted by the numbers, and lose sight of the people, content, links, and search terms; in other words, text > numbers.

WHO: Focus on active relationships: people engaging direclty with your brand by commenting, liking, joining, etc.
WHAT: Keep track of content: especially new items, and recurring items that spike in traffic; search for patterns
WHY: Pay attention to passive relationships: people engaging more indirectly with your brand via linking to your content, or searching for your keyword terms.
HOW MUCH: Spend the least amount of time on the size of your network: but don’t ignore it completely. Compare your stats over time, and to your competitors.

There’s a plethora of tools. One day someone will integrate them all. Until then..if you’re interested in getting a copy of this prototype document, just send me an email (Google Docs didn’t handle the images & formatting very well, or I would just post online).

Possibly related posts:

Theatre & Social Media: TCG 2010

Devon Smith — Sun, 20 Jun 2010 16:56:32 +0000

The Theatre Communications Group 2010 National Conference wrapped up last night. I heard word it was the largest in the conference’s history, clocking in at just under 1,000 attendees. A sizable chunk of those folks live tweeted several breakout sessions, mostly under the #TCG2010 moniker. My presentation, Social Media Strategy: Why ROI Isn’t Enough went quite well. I’ve received so much great input from folks over the past four months about what I should talk about, and was overwhelmed with gratitude from the many good luck emails in the days leading up to my presentation. So without further ado, for those of you who couldn’t be with us in person, and may not have caught the original tweeted link, here goes nothing:

TCG Social Media Presentation Draft 1View more presentations from Devon Smith.

You can also check out the live tweeting commentary. A few folks requested my notes, since I rattled off a bunch more stats.

Shortly after my presentation, my interview podcast with Technology in the Arts also went live! It in, I discuss the most significant findings of my social media research for theatres.

Speaking of, now I’m off to gather my thoughts on speaking at Americans for the Arts next weekend in Baltimore about Branding Yourself Through Social Media with the estimable Ian David Moss.

Possibly related posts:

Let’s Talk Metrics

Devon Smith — Thu, 25 Feb 2010 16:30:35 +0000

Social Media is growing up. We have more platforms and more thought leadership than ever before, but already our friends are trusting us less. In the beginning, there was sticky. Now we’re connected. And we want our content to be slippery, easy to pass on to our friends, and their friends, and their friends. I think reach, across users and across platforms, is becoming as important as volume.

Fortunately, there’s a nearly endless array of free online tools that can help us measure who we’re reaching, what they’re saying, and how they found us. I recently needed to capture the Social Media Footprint of an organization (Vocalo–”the YouTube of Radio”). I tried to follow their digital tracks to figure out, via backwards induction, their Social Media Strategy. Here’s what I found, in pictures. Or skip to the end to see the final report.

OVERVIEW

WEBSITE

Google Insights for Search

Next page…Facebook and YouTube!

@mentions as measure of user engagement

Devon Smith — Thu, 05 Nov 2009 02:50:00 +0000

So here’s a theory I’m testing out: Having more followers, tweeting more often, and/or being followed by more lists doesn’t necessarily lead to more @mentions (the holy grail, at least for now). I’ve yet to determine what does generate @mentions, but I’m working on it…more soon. Link to more

Possibly related posts:

Tracking @mentions