How Twitter Lists Influence… Influence
Photo by Globetoppers
As you may know, Viralogy (the company behind this blog) is highly involved in the analytic and measurement side of social media. We’re greatly invested in the measurement and ranking of a user’s online influence. I am one of the people working to refine that metric, and I truly believe that think Lists has the opportunity to throw the Twitter influence measurement system entirely on its head. Let me explain why.
For starters, Twitter influence has been measured over and over by a multitude of companies, none of whom have gotten it right yet. Twitalyzer, Twitter Grader, Twitter Friends – these are all sites that have made attempts to measure a user’s Twitter impact. However, many of them still focus far too much on number of followers, number of following and update frequency. These numbers, though important and interesting, do not tell nearly the whole story. Lists has a prime opportunity to help complete the picture for us.
How Twitter Lists Influences… Influence
Here’s something that may or may not be surprising to you: if you took a look at any category of Twitter users (tech, marketers, actors, etc) and ranked them by followers, and then right afterward ranked them by the number of lists they’re on, they would be ordered in a completely different way. I figured this would be the case, but wanted to do a sample test. I could not have been more shocked by the variation in results.
The above chart shows 3 things: the top 25 users (# of followers) tagged as “socialmedia” on Wefollow, the total number of lists they appear on, and their Follower:List ratio. This metric shows the direct ratio between the number of people who have followed them to the number of people who have added them to a specific list. What you see are vastly varying numbers of followers and lists, with the highest F:L ratios showing a user who has been added to a disproportionately low number of lists.
What does this tell us? For starters, a List add shows engagement with a Twitter account. Only if a user genuinely cares about and engages with another Twitter account are they likely to add them to their List. A List add is an unaided selection of a user based on certain criteria. Am I likely to add a spammer or someone whose posts I genuinely don’t care about? Probably not. It takes an additional effort.
By that same logic, users who are not added to Lists are not being engaged. Rock band Hypnogaja may have nearly a million followers, but is only listed on 80 Lists. That’s less than one list for every 10,000 users following them. What accounts for this? Perhaps the fact that they are a Twitter Suggested User. It’s extremely easy (and common) for a user to bulk add a list of “suggested users” upon joining Twitter. Hypnogaja happens to be one of these users. But they’re not the only one. However, the difference is that seemingly less people care enough about them to go further and add them to their Lists. For comparison’s sake, iJustine has >100K fewer followers but has been added to over 31X more Twitter Lists. Also, if you look right at the top of that list, Ashton Kutcher has over twice the followers as Mashable, but fewer List adds.
What else does this mean? Every analytics tool you use to measure your Twitter influence is outdated. Twitalyzer? All wrong. Wefollow? Incomplete. Twitter Grader? Give me a break. Without incorporating Lists into their metrics, they are now missing an essential piece of the puzzle. Listorious is an interesting application, showing the most popular Lists, and is a great application for the Lists addition, but is looking at Lists in a completely different way.
Though Twitter Lists are still in their infancy and are still able to be manipulated by malicious spammers and marketers, they present an interesting alternative to looking solely at followers as an influence measurement. The number of lists could eventually be more influential than the number of followers as a “popularity” metric. The advantage here is that fewer people will “follow-spam” as it will be largely unsuccessful. The disadvantage is that malicious users could easily game the system and add their own accounts to hundreds (nay, thousands!) of accounts. Unfortunately, with every step forward, we often appear to take a step backward.
The way to beat those malicious users is to not only measure number of lists, but number of followers of those lists (are you confused yet?). If we can eventually measure how many users are following @mashable or @aplusk across all lists, then we’ll have an opportunity to measure a user’s true influence. Until then, Lists are still but a piece of the Twitter influence puzzle, one that Viralogy is studying intently.
How do you think Lists will influence Twitter analytic measurement? In its current state, do you think it’s easier or harder to manipulate than number of followers?
Jaremy Rich writes a marketing, technology and gaming blog called Techshots. He is working to refine Viralogy’s vScore and is sad that he’s only been Listed thirty times.
Great post! Lists can push for more relevant following on Twitter because it actually encourages us to assign some sort of category, or a binding characteristic among the people we choose to add. If this successfully causes more users to reflect on the relationships they have with those they follow, then maybe our experience of Twitter will be more meaningful.
Thanks Em! I completely agree. As these listings become more robust, it will help to truly tag/define each user. For example, I’m tagged as Seattle, tech, friends, social media, oberlin alums. As this expands, it could create a better map of Twitter users, and will create even more implications.
Good work, Twitter. Ya done well (:
Thanks for going beyond where I started in calculating Twitter Influence Ratios. Lists really are a new Twitter metric, aren’t they?
Of course, some users that do engage don’t really belong on many lists, or should be compared with other users of the same type. A band should be compared with other bands; a tech pundit with other tech pundits. I imagine eventually we will have a category ranking factor that can be used to equalize so that an account from one category can be fairly compared to an account from another category.
They absolutely are a new metric. I couldn’t agree more. What I would love to see is tag clouds created using Twitter Lists. There will be an issue with case sensitivity, but eventually seeing the most listed users in “socialmedia” and the most listed users in “restaurants” or “seattle” could be interesting.
There is an enormous amount of potential in Twitter List measurement and analysis.
Hello Jaremy,
I believe that bit.ly is the one who is dominating the viral component of twitter. Since they convert most the links and can track them. I would then assume they could measure who RTs from whom and therefore be able to identify the most influential people in the twitter space.
I don’t believe they’re really measuring them at the moment. Additionally, RTs are not necessarily the best measurement either. Users that have a higher post velocity are likely to get a higher volume of RTs - plus, some users’ content is meant more to be read, not shared. Just playing Devil’s Advocate.
But yes, if bit.ly decides to do a better job of aggregating and sharing their information, it will also help a great deal to refine Twitter influence.
I appreciate the other perspective
The reason that I believe that they are tracking links is partly due to this article:
http://bit.ly/qGT2j
I also have a hard time believing investors would put up millions of dollars, if they were not measuring web influence. I am sure they will figure out how to accurately measure the influence of RTs in relation to post velocity.
Interesting analysis.
Unfortunately, I’m still not sure what you - or anyone else - means by “Twitter influence”. I can envision a few different interpretations, including influence _within_ Twitter (somewhat limited, but perhaps more easily measured) and influence _through_ Twitter (which may well include dimensions of impact outside the Twitterverse, such as inciting people to demonstrate in the streets). Although this may be a subtle distinction, I think that the whole process of assessing “Twitter influence” would be greatly aided by more clarity on the definition of terms.
Thanks.
Thanks for the comment, Joe.
What I mean by influence is the “impact a Twitter user has on an audience”. There are a great number of factors involved here, but think of it measuring the amount of value a Twitter user has. So it’s certainly not just the number of his followers (if no one is listening, his influence and value isn’t useful), and it’s not even simply the number of @replies or RTs (some users carry on more conversations, hence more @replies, some users’ content is just not as retweetable). What we’re trying to distinguish when specifically measuring “Twitter Influence” is how a user’s audience engages and receives a Twitter user’s content.
Additionally, at Viralogy, we are working to measure an ENTIRE person’s influence - eventually working to incorporate blog, Facebook and a number of different social media resources. With that, we should be able to eventually show impact THROUGH Twitter. However, it’s slightly more difficult to measure non-electronic impact in a broad-stroke way, so our metrics will likely be limited to influence online.
Makes sense for the most part but I’ll just add some notes on this statement:
“For starters, a List add shows engagement with a Twitter account. Only if a user genuinely cares about and engages with another Twitter account are they likely to add them to their List.”
I think that mostly holds true for users with a large following. Take me for example, I’m following about 370 users and am just about done going through all of them and putting everyone on a list. Some of these people I’ve never engaged with or whose content I don’t really care about but every now and then I get a little morsel of coolness that its worthwhile to keep following them. I just like having my followings organized, plus it has helped me re-discover many who only post once or twice a day, whose tweets were hidden by those that tweet 200 times a day. Plus I’m sure there are those that get put on the “dontFollowThisGuy” or “jerkAlert” list. There are negative lists (aren’t there?)
I’m also with the last commenter, what is “Twitter Influence” anyway and how is it even measurable? (other than counting link clicks?)
-Chaalz
Absolutely true. Hopefully in the future, some of those “negative” lists can be measured. However, in terms of adding people to lists “just to do it” vs. “because you care about their content”, it’s always a directional. Sure some people will be added “just because”, but users with extremely prescient and valued content will be added to a far greater list. The big issue here would be measuring those with a smaller number of list adds. It’s easy to tell the difference between 3,000 adds and 2,000 adds. But for those users with only 4, 5, or 10 lists added, what does the difference actually MEAN?
Answered your last question in response to the commenter above. However, here are some ways to measure Twitter influence:
-Link clicks
-Followers
-Retweets
-Mentions (@replies)
-Lists added
Currently we have incorporated bits of the first four metrics into Viralogy. They each have varying degrees of importance (that differ based on what you’re looking for/at).
Also curious, what would you say the following implies:
1. You following someone but don’t put them on a list
2. You put someone on a list but don’t follow them
Maybe it doesn’t say anything at all.
Hard to generalize for ALL users, but what I’d take away from these:
1) Either not as important to you (as those you have put on lists), or you just don’t often add people to lists - which is why it’s more important to look at the entire sample (ALL Twitter users) as opposed to an individual user or handful of users.
2) Not sure. But that may happen more and more often if Twitter changes gears from adding followers to adding Lists.
Great post and commentary…and very cool chart!
I think we all agree that the Lists feature is a good thing and the current metrics for rating Twitter influence sucks. I’m not sure that there is a way to accurately gauge influence since by it’s very nature - the openness of Twitter - it can and often is gamed.
Lists certainly enhances the quality of measurement though I do agree with Chaalz - people use lists in different ways and for different reasons, sometimes simply to categorize the users they follow.
Unfortunately numbers will always be a significant indicator of status - especially with the growing demand for a social media ROI. So for many, Twitter will remain a popularity contest.
But keep plugging away - I love what you guys are doing at Viralogy and though I maintain that an accurate measuring standard is elusive, I’m rooting for you to prove me wrong
Tim @uMCLE
Thanks for rooting us on
No matter what, any sort of broad-stroke measurement will have its slight flaws here and there, but our goal is to be able to give a directional measurement of a user’s influence. Sure, the difference between a vScore of 1,000 and 1,050 is pretty negligible, but in the end it’s also a directional - compare it instead to a user with a vScore of 500, or 5,000.
It is really important to have ways to measure social media ROI, as it is so new. Right now there are two camps of people: those who get into social media because “everyone is doing it, and we have to!” and those who truly believe its possibility and importance. For the second group, it’s about proving how worthwhile social media campaigns are, and for the first group, once the shiny new appeal wears off, it’s important to have statistical measurements and analysis to fall back on.
Looks like we both agree that Lists are (potentially) a far better unit of measurement than the current metrics. However, only time will tell how important they truly are.
Jaremy,
I want to clarify some possible misconceptions you presented about Twitalyzer. You stated that many of them still focus far too much on number of followers, number of following and update frequency. We intentionally do not focus specifically on the number of Followers or number of updates. We do take Followers and Update Frequency into account as part of our overall Influence score, however there are other components of this weighted metric, such as how often you share information (via retweets, links and hashtags) and how often others reference you.
OK, now more about your post. You wrote that a List add shows engagement with a Twitter account. Only if a user genuinely cares about and engages with another Twitter account are they likely to add them to their List. I would submit that Following another twitter account also displays a similar level of engagement and logic as Lists.
I agree that Lists should and will have an impact either as a stand-alone metric and/or incorporated as part of a larger engagement metric, but to state Every analytics tool you use to measure your Twitter influence is outdated. Twitalyzer? All wrong. Wefollow? Incomplete. Twitter Grader? Give me a break. is a bit of hyperbole, no? I think trying to incorporate something that is truly a work in progress (it is a week or so old feature for a majority of Twitter accounts) in an the overall measurement strategy is a bit premature. Even you say Lists has a prime opportunity to help complete the picture for us.” The keyword here being has.
Speaking of tagging - did you know that you can tag yourself in Twitalyzer now? http://www.twitalyzer.com/twitalyzer/tag.asp Also, you are now on 15 lists
Hey Jeff, thanks for the comment! Let me first clarify: I did group you in with the others, but honestly I do believe that Twitalyzer does by far the best job of Twitter analytics of all measurement services. By. Far. I’ll admit I was taking a little bit of artistic license by sensationalizing ALL analytics programs together.
I know that number of followers is not specifically taken into account in any individual metric, and that’s one of the reasons I like Twitalyzer - I’ve always felt Twitalyzer scores were a better reflection of my impact than, for example, Twitter Grader. However, I was under the impression that “update frequency” is taken into account - hence, “Velocity”. I’m not saying that I think it shouldn’t exist - there are certainly reasons that it’s valuable - but I was pointing out that it is still a fairly poor metric.
I disagree that Twitter following is just as important in terms of engagement as Lists. With programs like Tweepular - which allow a user to follow hundreds to thousands of accounts in hopes for a follow-back, the “following” number is nearing irrelevancy, in my opinion. A little over 4 months back, I did a “Twitter Experiment” used to analyze two separate accounts I’d created (@jaremy and @techshots). It was an experiment with an obscenely small sample size (2, over the course of a week :P), but my hope was to eventually do something more large-scale (which I have not done yet). I used the @techshots account to build a larger following through methods like auto-following accounts, mass following, etc. Whereas with @jaremy, I built the account organically (I’ve never followed anyone I didn’t care about). What I found was that despite a much larger following, my impact was much smaller. (Which is one of the reasons why I really felt Twitalyzer does an accurate job of analysis - it was the only service that showed those same results. Overall I found that an account with 1/4 the followers still had more click-throughs to links and a higher level of engagement (through RTs, @mentions). That impressed me. You can take a look at the study here: http://www.techshots.net/2009/06/make-friends-not-followers-twitter-experiment-part-2/ Just remember that it was conducted as a test more than as any kind of quantitative study. At some point I’d like to do a much larger scale analysis. But that must wait for a later day.
Lists are certainly not refined at this point. However, I still believe it currently offers a better portrayal of true influence than pretty much all of the other metrics. That said, you guys currently do a great job at Twitalyzer of truly measuring influence, and I’d expect you to take your time when adding a new statistic into your database.
Great to see that you’ve added tagging. I’d love to hear an update on how that is going over time. Sounds like it could be interesting.
Hi Jaremy,
I think there are semantic differences that I are causing some misunderstandings between us. There is Influence and there is Engagement So, when I said I would submit that Following another twitter account also displays a similar level of engagement and logic as Lists. I was not referring to Following potentially having the same impact on Influence as Lists, but rather the act of intention (part of engagement) by someone to follow someone and the act of intention of adding someone to your Lists. Yes, currently it is a lot easier to add people to follow, as you point out, like Tweepular, but it is only a matter of a time before a “List-pular” comes out
Yes, velocity is accounted for in our Influence score, but it is part of a larger weighted computation.
We will have to agree to disagree about the relative significance of Lists as a measure of ones influence, especially as a stand-alone metric. It is simply not ready for prime-time.
Thanks for your comments about Twitalyzer.