We have just launched the next generation case portal. Please visit our blog to find out what changes this brings. If you are experiencing any issues accessing the case portal please reach out to: Support.

We are also aware that supportcases.lithium.com is displaying a certificate error and are working to resolve the issue.
Data Science

The Fast Influencer Myth

727i2A698852917EF381Michael Wu, Ph.D. is Lithium's Principal Scientist of Analytics, digging into the complex dynamics of social interaction and online communities.

He's a regular blogger on the Lithosphere and previously wrote in the Analytic Science blog. You can follow him on Twitter at mich8elwu.


I was wondering what I was going to write about this week. Then I came across the Influence Project that was launched by Fast Company yesterday. I accidentally clicked on one of their links from my tweet streams, but quickly figured out that it was a link-bait with no actual substance. Moreover, it was designed solely for Fast Company to collect lots of emails and get lots of traffic for free. Subsequently, it has received a lot of negative coverage by bloggers, analysts and news sources. I’ll list a few links here, but I won’t bother to repeating the story, so read them if you like.

However, I must say that this influence project really has nothing to do with influence. According to Amber Naslund, FastCompany promotes this game by saying that “(1) You can use any means to spread your unique link to your online network… (2) Your goal is to influence as many people to click on it as possible.” Unfortunately, this pyramid scheme is not influence, it is not even popularity! What if someone tricked, manipulated, or even spammed/frustrated you into clicking the link? What you get at the end is, at best the biggest link-spammer, at worse the biggest trickster and liar, on the November issue of their magazine; definitely not an influencer.


influencer_sheep_wolf.pngThis led me to think why are there so many deceptions around the search for the elusive influencers? Part of it may be companies are taking advantage of the public’s ignorance. But I believe that part of the confusion is that the term “influencer” is so misunderstood that it has almost become a cliche. It means so many different things to different people that we are all confusing ourselves. And part of this is our fault for not clarifying what we meant when we use the word “influencer.” I want to start to correct that here.


The Marketing Influencers

When Esteban Kolsky, Amber Naslund, Maria Ogneva, etc. speak of “influencers,” I believe they mean someone who can influence another’s decision process. For most businesses, the decision of interest is usually the decision to adopt/buy their product or service. This decision process can be anywhere along the purchase funnel (i.e. it can be a real purchase decision late in the funnel, or merely sentiment about the product or service early on).


If you are in marketing, you should be very interested in this type of influencer for obvious reasons. You can market to this select group (which is easier and cheaper, because there are usually fewer of them), and then they will do the rest of the work for you. Their influence will essentially drive the message to their friends, families and followers on your behalf. Since this type of influencer has the potential to drive sales through their word-of-mouth (WOM), I will call them marketing influencers. They are the type of influencer that I studied earlier this year. To identify marketing influencers, you need the six factors of social media influence.

  1. Domain credibility
  2. High bandwidth
  3. Content relevance: This is the context that Maria mentioned in her tweets.
  4. Timing (temporal) relevance
  5. Channel alignment
  6. Target confidence: This is the trust that Amber mentioned in her blog post.

I must emphasize again that each of the six factors necessary for influence, but none of them are sufficient without the others. All six factors must be met in order to find the true marketing influencers.


The PR Influencers

But wait, there's more! What about influencers like Dave Carroll (in United Breaks Guitar) and Kevin Smith (who was allegedly Too Fat to Fly Southwest)? They certainly do not fit all the six criteria above (e.g. Neither Dave nor Kevin is an expert in the airline services industry). But they certainly have demonstrated their influence in this area. Are they not influencers?


Strictly speaking, I wouldn’t call them influencers, they are simply celebrities. But if you insist on calling them influencers, then they are not the marketing influencers we talked about earlier. I would call this type of influencer, the PR influencers, because they have the ability to spread your bad PR very rapidly among their numerous followers.


You would not use a PR influencer to drive sales: just as you would not use a PR expert in your company to do a marketer’s job. If you do, what you get is a pay to play advertising, not influence. This may work for a while, but people will quickly question the authenticity of these pay per tweet ads and ignore them altogether. Moreover, this will tarnish the credibility of the tweeter and the company they represent.


So how could you best leverage these PR influencers? I’m afraid to say that besides treating them with the respect they deserve, there is not much that you can do with this type of PR influencers. The most important thing with these PR influencers is don’t aggravate them, and the best thing they could do for you is to avoid a potential bad PR that could cost $180 million as in the case with United Breaks Guitar.



The search for the influencers is on, and they’ve certainly become highly desirable commodities. Many companies are claiming that they can identify influencers. This is good news and bad news. It is bad news because influencer identification is complex. Not only are there different types of influencers (in this post we talk about only two types: marketing type and PR type), using the wrong type of influencer for the job can be counterproductive. The good news is that there is a science that goes into influencer identification, and it can be done right. So don’t get fooled by the word “influencers.” If you are not careful, it could lead you to find exactly the person you weren’t looking for, the biggest trickster.


Now that I get my thoughts out, I can go back to my research, and analyze the business implication from the distinct role of communities vs social network. Next week we will revisit that topic. In the mean time, I welcome any questions and comments as always. I would like to hear your thought on this controversial issue.



Are there ways other than the obvious (hey -- that's Kevin Smith!) to identify people who have a propensity to be PR influencers?

Data Science

Hello PhilS,


Identifying the PR influencers, is much easier than identifying the marketing influencers. Usually high bandwidth is all you need. This is only 1 of the 6 criteria needed to find the marketing influencers. So identifying them and notifying them is usually not a problem. For such PR type of influencers, I think that is the best you can do in most situation.


Hypothetically, if they happen to get stuck somewhere and need help (but this is not in your control) maybe you can be extra nice in helping them, or go the extra mile to get them what they want. Then maybe they will say a good word about you.


But, remember PR influencers are really celebrities. Being such, there are usually some sense of entitlement to nicer treatment. So you would have to go beyond extra nice (which may be costly) to get them to spread a positive WOM for you. And even if you did went the extra mile and have been extra nice, there is no guarantee that they will say anything, because they may feel that they are entitled to it for being celebrities.



Hi Michael,


Thanks for the shoutout in your blogpost. I love love love your 6 points of influence, I think they are right on point. However, I haven't seen anyone articulate it quite this way. This is a great guideline. To me, context is united under temporal and content relevance, but it's helpful to break it out in two. I agree with Amber, that you can't and shouldn't measure all of influence. So much of it is "invisible", happens via backchannels, and a composite of so many factors. But... if you do measure it, you should look at something like Klout, which looks at different dimensions of influence - it's not just reach, but also measures of engagement and "the company you keep". I like that they allow you to view influencers by topic (as well as provide an overall measure of influence), which helps contextualize.


I agree with your assessment that Kevin Smith is not an influencer but a celebrity. So he comes with a naturally built-in audience, but I bet things that people seek his advice on are mostly around comedy and filmmaking. A big audience also doesn't mean an engaged audience.


Finally, I agree with your assessment that most feedback was negative. I had a feeling that it was the case, just eyeballing tweets and blogs, but I had to test it out and measure.


Cheers, and keep influencing!





Data Science

Hello Maria,


Thank you for commenting on my blog.


If you like the 6 factors of social media influence, then this mini-series that Esteban mentioned in his breaking rant post would be perfect for you. It is all about influencers.

  1. The 6 Factors of Social Media Influence
  2. Finding the Influencers
  3. The Right Content at the Right Time
  4. Hitting Your Targets

Yeah, I agree that context would include both content relevance as well as temporal relevance. Timing is separated out because it is a factor that is often overlooked, whether it is in data analysis or algorithm development. This post (Community Influencers Step by Step) shows you step by step how we measure influence in our community platform.


I really like your sentiment summary on this influence project. I was expecting a greater polarity between positive and negative sentiment (obviously with more negative). But I guess it is still early in the game and people still haven't realized what's going on. It would be interesting to repeat this measurement over time and see how people's sentiment change as the learn and figured out what is going on. BTW, very nice charting widget for the graphs and easy sharing interface. Is that a product feature of Attensity heritage or Biz360 heritage?


Thanks again, and see you here next time.





Insightful post, I appreciate your thoughts, as the FastCompany 'fiasco' as I have personally referred to it, is this weeks poster child that has people thinking and talking. I am not an expert in marketing, nor influence, your 6 factors help me to get a better understanding of what to think about. They are mostly logical, which is good. The one which I believe will become increasingly relevant is the content relevance (along with context).


The small city near where I live (Burlington, VT) celebrated Social Media day last week. There was a bit of a vote for the King and Queen of Social. A friend wrote a post about the event. Rich's post started to explore the relevance of the online tools used to score influence on Twitter. The quick summary is that they were close to the votes that took place, but not quite there. This led to a brief conversation about geospecific issues.


In future the 6 factors might need to be weighted for an actual algorithm to exist. I agree with your statement that they all need to exist, but the weighting will depend. For example, influence for local may or may not have a large temporal component (a restaurant review or hotel review has duration). I may only have limited bandwidth, but when I say something people listen. Just examples - but I am sure you think about these things more than I do.


Thanks for the post - Mitch

If you simply ask your online followers to help you out with this Fast Company Influence Project, and don't lie or manipulate them to click on the link, isn't it ethical?


I agree using lies like "you will see an awesome picture" or "click here for a free iPad are unethical.  However, if you let your readership know the truth behind the project, I don't see anything unethical.


Hi Michael -



 Can you unpack this sentence for me a little bit more, "What you get at the end is, at best the biggest link-spammer, at worse the biggest trickster and liar, on the November issue of their magazine; definitely not an influencer." Feel free to get technical/mathematical with your language. I didn't finish the PhD Physics program at Gonville & Caius but I think I can probably keep up with your answer.



 I saw the Influencer Project as more of an influencer experiment. While not all experiments are equal (and I think this one is not elegant at all) I am interested to see the results and web map that gets produced. As a scientist, do you think it's better to wait and see how the experiment plays out before passing judgment or are some experiments so bad that they should be stopped before they are concluded?



 Not to be a nudge but how do you know Kevin Smith is an Influencer? By your definition wouldn't that Influencer have to effect the Purchase Funnel? Do you have data that shows that Kevin Smith was anything more than hilarious, fat and that people paid attention to him for a couple of days? Did he change or skew SWA's numbers or results?



 The team here at Fizz, my company, has been working exclusively with Influencers since 2000. The comeback of PBR, the success of The Closer and the growth of chocolate milk as a recovery beverage are some of our more well known case studies. Of course I know if your work with Lithium so I look forward to having a dialogue with you.





Hello Hello Dr Wu


Thanks for posting the thoughtful examination. I have to take exception to your comment:


"Moreover, it was designed solely for Fast Company to collect lots of emails and get lots of traffic for free."


That's just not true. I wrote a blog yesterday that gives a fuller explanation of The Influence Project (http://www.fastcompany.com/1667964/popularity-ego-and-influence-what-is-the-influence-project), but I'd also like to point out that in my May profile of the production/ad shop Mekanism, I was completely open, clear and transparent about Fast Company's intentions (http://www.fastcompany.com/magazine/145/repeat-offenders.html).
Regardless, I appreciate your contribution to this conversation and hope we can have a deeper discussion as I report and write my larger story about influence and influencers.


Data Science

Hello Mitch,


Thank you for the comment.


I like your quote about "By asking people to ‘Tweet’ about influence in order to measure influence is not pushing things to simply an unbalanced state, but an unnatural state." That is totally the case.


I'm glad you like the 6 factors of social media influence. These factors are not factors in an equation. Some of them are not even necessarily computed. So there is really no weighting per se. The 6 factors just need to be accounted for when analyzing the data or developing an algorithm for influencer identification. This post (Community Influencers Step by Step) shows how we do it on our community sCRM platform. That being said, we can certainly put more emphasis on relevance and timeing (both of which can be thought of as the context).


The geo-specific issue is more or less lumpped into the content relelvance factor because when you are online, the precise geo-location doesn't really matter anymore. If you are considering going to a restaurant in NY, you would search for that. And the person who writes the review for a NY restaurant, doesn't need to be living in (or near) NY to have influence on you. As long as what an influencer writes is relevant to what you are looking for (a restaurant in NY as opposed to a restaurant in LA), then you can potentially be influenced.


However, as mobile and real time web is becoming more mainstream. Location might be an extra factor that need to be considered separately.


Thanks for stopping by and commenting on my blog. Hope to see you again next time.



Data Science

Hello Eily,


Thank you for your question/comment.


First, I am no saying that this was an unethical project. Simply that it is nothing about influence. Besides, not only "you" have to be honest and don't lie, every single one of the participant has to be honest and not lie. Strictly speaking then, if only one person lies, then the experiment is invalidated. There is no mechanism that tracks who lie, or how they lie. The winner may be the liar, and we won't know. So there is no way to correct for that.


If there is one thing that I learn about social systems, there will always be some people who will break the rule and game the system. Especially when the object of pursue is desirable (i.e. money, fame, reputation, or in this case, recognition of being the most influential person).


Whether it is ethical or unethical is a completely different question. I will leave that for people to judge. I am only saying that the experiement is so poorly designed in that its result is most likely in invalid. And I also find promoting it as an experiment to measure influence very misleading, because it doesn't.



Data Science

Hello Ted,


Thank you for your point of view, and I apologize if my writing have not been completely clear. I wanted to get this post out quickly, so maybe I've written it too hastily.


As I mentioned in my reply to Eily, what I want to get across is that the experiment is poorly designed and the results deviate significantly from the claim. The claim was that this is an experiment about influence “designed to answer the question: Who are the most influential people online right now?” And at the end of this experiment they claim they will identify the most influential person, who will be on the “November issue of Fast Company magazine.” They also explain that “Real influence is about being able to affect the behavior of those you interact with.”


Here is where I would disagree. To me, it matters HOW you affect other’s behavior. If someone frustrate you into clicking the link by spamming you, that is not influence. If someone did that and end up on the November issue of the magazine, then that person is not the most influential, he is simply the biggest spammer. And if someone lied and tricked you into clicking the link, that is not influence either. If someone did that and end up on the November issue, then that person is (I hate to say) the biggest trickster and liar. There are probably more examples that I can think of. In either case, the person that ends up in the magazine is not likely an influencer, because they explicitly encourage you to “use any means to spread your unique link.”


I hope the above paragraph offers some explanation to the longish sentence that you want me to unpack. There is no need to go technical with this discussion.


As a scientist, I encourage anyone to do experiment. But if the experimental method is fraud to begin with, then there is no point in doing the experiment. That is call wasting time, and I would not encourage people to waste their valuable time. However, some people may choose to do that for entertainment or whatever reason they like, and that is perfectly fine. I am perfectly happy calling this “Popularity Project,” “Super Clicker Project” rather than “Influence Project.” This aggrandizing title and misleading claim is what encouraged me to write this blog post. If they call it a popularity contest, maybe I’d be more interested to find out the result. However, knowing the resulting influencer is totally confounded by poor experimental design, I see little value other than pure entertainment/fun.


I believe I’ve said that I would NOT call Kevin Smith nor Dave Carroll influencers. But others in the social media community tell story about them. And they do often time refer to Kevin (as well as Dave) as influencers due to the PRs they’ve generated. I heard them on panels, talks, etc. This is why I specially called out that part of this is our (social media enthusiasts/practitioners/researchers) fault. We definitely need to be more careful with our language and define what we meant. That is why I want to clarify that there are different types of influencers. That was part of the original intention of this post.


I’m very curious about your case studies. If you don’t mind, would you point me to their location. I’m happy to talk to you anytime to explore this area at a deeper and more technical level.


Thanks again for dropping by and posting your questions and challenges here. Hope to see you around more often with engaging conversations.



Data Science

Hello Mark,


Thank you for commenting with your clarification here.


If I’ve misinterpret the intention of this “Influence Project” I apologize. However, it was indeed how I felt when I wrote this post. When I first encountered that link on my tweet stream, clicked it, and expecting to see some cool influencer project, but found it was a link bait with no substance, I’ve lost my confidence in this project.


"Not only are there no real content, and I have to register for that… No way am I going to give you my email for some pointless clicking contest." That was how it made me feel.


Regardless, I’m glad you posted your clarification. And I’d be happy to talk to you anytime. Designing a social experiment with the proper control is no easy task. But I’m happy to discuss and share my thoughts with you.


Thank you for leaving me a comment. Hope you will come by Lithosphere sometimes in the future too.


My organizational professor defined Power as the ability to make others behave in a way you would want them to behave. Willingly.


Whereas, influence is not really about Power - it refers to changing perceptions of people.


It is linked to one of the sources of Power - expertise power.


Online influence is actually a function virtual reputation.


Dr. Wu -




Thanks for your interesting comments. I agree that, as an experiment, it was not elegantly designed. I myself am not the best at experimental design so I have no suggestions on how the experiment can be tweaked to reach it's lofty goals. I would also agree that the experiment's claim to be able to identify the "most influential people online right now" is per se unobtainable because the universe of "the web" and all it's externalities are not taken into account. Like you, I wonder if the real problem here is that the rhetoric behind the experiment overpromised on the possible results of the experiment.




As someone deeply involved with this field I hope that the vitriolic reaction by many people, some of which you linked to, will not deter others from concepting and putting to the test further experiments. Years ago I remember sitting down with Myron Scholes and Bob Merton to talk about the years of work, trial and error that it took to be able to correctly price options (another amorphous task, like identifying Influencers) in such a way that the data was reliable. They, along with Fischer Black,  were eventually successful and  trillions of dollars of value was unlocked in our global economic system. Perhaps the same holds true for identifying on-line Influencers? I suggest that we acknowledge Mark Borden for his hackneyed attempt then move on with better experiments and results. Who knows, maybe Amber, Alexia and Esteban will even turn their energies from snarky writing to making positive contributions to solving this question.








Data Science

Hello Gautam,


You got some well articulated comments here.


I agree totally, and I think your comments are so right-to-the-point that I have nothing more to add. You should comment more... Smiley Happy


Thank you again, and hope to see you next time.


Dr. Wu,


First, thank you for continuing to share your thoughts and research. I enjoy reading each post and appreciate your passion for this area of business.


You mention in your post that there is a science that goes into influencer identification. Where can I go to read more about that. It has peaked my interest.


Data Science

Hello Ted,


Thank you for coming back again to carry on the conversation.


I’m glad you find my comments interesting. I do agree that the rhetoric behind the experiment was over-promising on the possible result of the Influence Project. And once I found out I was tricked into clicking the link, I lost my trust.


However, I must emphasize that it was not my intention to criticize this experiment, or deter people from participating. Everyone is entitled to pursue something fun and a chance to be feature in a magazine if he/she desires.


The main goal of my post is to bring to attention that the term “influencer” can actually be very misleading if not defined explicitly. Since it is a “hot” term now, many people want to use it as in the case with Fast Company’s Influence Project. But if you simply take for granted what an influencer is, and not look at how they were defined, what data were used to find them, and how they were found, then you can get someone totally different. Even in the social media community, people speak of influencer casually, when there are really many types of influencers (I only talk about 2 types in this post).


Just curious, are you involved in the development of the Black-Scholes Option Pricing model?


Regardless, thank you again for the conversation. I do hope to talk to you someday and discuss/share our approaches and methodologies for finding influencers.



Data Science

Hello Kevin,


Thank you for your interest.


In my reply to Maria, you will find a list of blogs that I posted earlier this year on some of the research I've done on influencer identification. The model and theory can be found in this mini-series on influence analytics (below):


  1. The 6 Factors of Social Media Influence
  2. Finding the Influencers
  3. The Right Content at the Right Time
  4. Hitting Your Targets


Subsequently, I also posted a step by step example of how I apply this model to find influencers in our community platform. Following that, I also posted some interesting applications and insights I found using this analysis. And these can be found here (below):


  1. Community Influencers Step by Step
  2. Social Graphs: The Art and the Insights
  3. Social Network Insights from Unconventional Graph Metrics


I think these articles will get you started on this fascinating topic: the science and analytics behind influencer identification.


Thank you for the inquiry.



Hi Michael,


Thanks! I was actually expecting a greater polarity too, but you are right, it probably looks different now. I'll try to make another recap next week, if I see a big difference. I literally thumbed through a large sample of the blogposts, and the positive and neutral ones were in fact that. Some people do want to participate in them. However... the blogs with most reach, engagement and overall "street cred" were overwhelmingly negative. We actually have impact as a metric, so I can actually look at neg. vs. pos. based on impact (i.e. how many people were actually exposed to and somehow interacted with the negative posts vs. positive). I haven't looked yet, but I bet that it would be skewed much more heavily in the negative direction.


Re: charting widget - that's actually not part of the product, although you can easily share, email and export all of our reports and dashboards (Biz360 heritage) - but that happens within the product - what you see here are just embedded jpegs. The "click to share" is just part of the wibya toolbar below.





Hi Mike,

Very interesting. You are doing a great job of breaking it all down into the right sized elements.

You are a Scientist while several here and others that are reading aren't. But all it takes is plain common sense to understand that enticing people to click a URL to accomplish one's own selfish goals is not the right thing to do if you are building Trust. I would not dwell too much on whether that can be called influence or not. I would say, anybody that is trying to influence a group of people or persons can only do so in the long term if his 'Motives' are right. It wont take long for people to recognize.

Well, if I trust Mike, due to his credibility, bandwidth, and the content relevance that he has demonstrated in the past, I may be influenced to click on a URL because he recommends.

But, this is a sure shot way of losing all the trust that Mike has built over may be days, months or years if it leaves a bad taste. In other words, Mark Borden and Fast Company have definitely lost a large extent of the 'Influence' and trust that they ever exercised or built respectively in the online community. Anybody who makes the front page of their magazine as a result of this experiment may end up losing friends. I agree with SFWeekly's Alexis Tsotsis: It is like giving out your list of acquatainces and friends that you have carefully nurtured, with relationships built over the years to a network marketing company such as 'Amway'. I would feel terrible and so would my friends...I would lose their trust.

Trust OR Target Confidence as you mention is the foundation for Relationships. One's sphere of influence will soon diminish if the audience doubts his 'Motive'.

I would say this experiment from Fastcompany will end up measuring how quickly can one lose trust and influence.

Excellent blog post, Dr. Wu

I agree that there are influencers and then there are "noisemakers"

Both need to be dealt with, and the approach is dictated by the contextual and inter-presonal parameters you defined in the article.

Thanks for sharing this and for not witholding your opinion.


Data Science

Hello Maria,


Welcome back to the conversation.


What you observe matches my gut check too. I will also do some temporal analysis later to see how things change over time. Should be interesting to see this after a few weeks.


I am quite curious about your impact metric and how do you measure it? Is it just the page views of the article, or some proxy for page views?


Anywya, thx for the tip. I like that click-to-share feature from wibya toolbar.


Data Science

Hello Whatsay


Thank your for commenting on this blog post.


You said it very well, "all it takes is plain common sense to understand that enticing people to click a URL to accomplish one's own selfish goals is not the right thing to do if you are building Trust." And influence has a lot to do with trust and reputation as stated in Gautam's comment.


As in Maria's comment, there are people who want to play the game. But as you said, I think that people will eventually recognize the ulterior motive and and slow down or stop all together. This is why, I want to wait a few weeks before I do a more in depth temporal analysis on the voice of the social web and see if I can chart any changes over time.


You are definitely right: If you are not careful with "influence" you can end up on the very opposite end of the spectrum.


Thank you for the excellent comment.



Hello Michael,

I have been following your posts on Social Networks/Communities/Influencers but never did sit down to comment because of travel and some personal constraints. Finally got some time and so here are some quick thoughts on these topics.



First, great job with the last few series of posts. Informative and definitely thought provoking. I agree with most of what you say but wanted to share some of my own perspective:



* <b>Ties vs Relationship:</b> While most equate the two to be the same, I tend to see it is with a more complex dynamics in place. To make it really simple, I view "tie" as a  pre-defined relationship - one, where there is an understanding of a future mutual benefit. Or to pose it as a question -"Isn't it possible to have a strong relationship with a person but yet have a weak-tie with that person in terms of your social network?"



* <b>Weak Ties & Strong Ties:</b> Agree with a lot of what you said here.  I do think though that weak ties can be  formed intentionally and not necessarily because of  'forced by' limitations. I view "the type of ties" formation as very much defined by the goals and personalities of the individuals.  I might be the type of person who want to deeply get entrenched in a domain, in which case I would slowly but surely migrate towards forming strong ties with the other "nodes" in that domain. On the other hand, I might be a person who transcends specialities and want to skirt around different domains - in this case I would eventually move towards forming more weak-ties or bridging-ties (relatively speaking - they will have strong ties too). Obviously, as you said both are important and play different roles and with its own benefits - a topic of conversation unto itself.



* </b>Influencers</b>: Again, another very interesting topic that cannot be done justice to in one blog post or comment. Going back to Weak-ties/Strong-ties, folks having either can be  influencers (there is a misunderstanding among some that only strong-ties can be influencers). People with lots of weak-ties are the best expediters and play a critical role in the virality of a topic. People with strong-ties are are good influencers in percolating and diffusing the knowledge within a domain and expanding existing knowledge within a domain. The best influencers are those though who has a good collection of both strong and weak ties.



Going back to basics - more than power, authority, or expertise, I see an influencer as <i>someone who can 'extract' an action out of another person voluntarily</i>. So from this perspective, celebreties can be  influencers - as millions around the world "do" things because of their influence. And on this one Michael, I would not differentiate them as "PR Influencers" as they can be marketing influencers too. As an example, there are millions of girls who use Brand X of something just because Hanah Montana or Angelina Jolie uses it. And this is one of the reasons I disagree with the assumptions made by some that "expertise" is a must to be an influencer.



And lastly, one cannot talk about influencers unless the context & constraints are spelt out. Social influence is different from Academic influence for example. Similarly, social media influence is different from search optimization (SEO) influence. For example, when I advice folks on SEO one of the first things I tell them is that they should separate their strategy to influence search from their strategy to influence their customers as the "influencer list" for both can be quite different.



All in all, the last few posts were a great read. Enjoyed it.



Data Science

Hello KevinC


Haven't seen you for a while. Thanks for coming back and commenting.


There are definitely influencers and "noise-makers." But personally, I don't think Fast Company's Influence Project is able to distinguish the two. Maybe I should write a blog on how they could design this experiment so that it is less prone to gaming and abuse.


Always glad to share my thoughts, and I'm glad you find it useful.


Data Science

Hello Ned,


Welcome back, and thanks again for the elaborate comments on this post and earlier post.


I must say that I’m confused by your notion of ties vs relationships. According to you, if a tie is a pre-defined relationship, then any tie formation must happen before you establish a relationship, but not all relationship has to start with a tie, they can simply start out as a weak relationship (since you made the distinction between tie and relationship). So you either go from weak ties -> weak relationship -> strong relationship or weak -> strong relationship. If that is the case, by your own argument, I don’t see how you can have a strong relationship yet a weak tie with a person. Maybe I’ll need you to rigorously define what you mean by weak tie, strong tie, weak relationship, vs. strong relationships before commenting further.


I agree that many ties are formed intentionally. This is a topic that I briefly introduced in “How Do People Become Connected?” under the section title: The Desire to Connect is a Basic Requirement. As I mention there, we can use Game Theory to analyze this situation, and view tie formation as a result of cost-benefit analysis of both parties. It is a topic that I will write about at a deeper level later.


Concerning influence, weak ties can definitely influence too. In fact, according to Prof Mark Granovetter (in his seminal paper The Strength of Weak Ties), weak ties are more important for the global WOM propagation in a social network. Information spread further via weak ties. Moreover, because any individual can have many more weak ties than strong ties, the influence of weak ties may sum to be stronger than the influence of the stronger ties. However, if we look at person-to-person influence, strong ties (provide they meet the six criteria) will have a greater influence than the weak ties.


I cannot agree that an influencer is JUST “someone who can extract an action out of another person voluntarily.” It matters a great deal HOW he/she do it. Please see my reply to Ted Wright on this. In fact the voluntary clause is probably irrelevant in the online world, because pretty much everything is voluntary online. Even if people spam you, or tricked you, you are the one who decide to click the link at the end. The key is how they extract that action out of you. If they spam, lie, or tricked you, that is not influence to me.


If the choice was up to me, I would not call a celebrity influencer. They are simply celebrities. And no doubt that they can certainly exert some influence. The reason that people will follow and use brands that are used by Hanah Montana and Angelina Jolie is because they considered by the public (those girls) as “experts” in fashion, style, designed etc. But will companies start using Lithium if Hanah Montana says Lithium is cool? Celebrities can definitely influence the market, so can everyone; they just wouldn’t be as effective as the marketing influencers. If there is any influence, they will probably not last very long as in the case of Kim Kardashian, when people start to question her authenticity.


I simply want to point out that there are just many different types of influencers out there. So when people see the term “influencer” they should think about what that really means. Whether I call them celebrity, PR influencer, marketing influencers etc., doesn’t really matter. They are just different.


Finally, I agree that there are many different types of influence, and getting the right context is crucial. The six factors that I outlined is for social media influence. It may or may not apply in other context, such as academic influence or SEO influence. I would have to re-examine that.


Whew, looks like our discussions can almost be a blog post all by themselves. Thank you for spending the time to voice your thoughts. I’m sure this will benefit many others in the community.


Thanks for the great response.



I am going to flip my counter and talk about the influencer part first as I feel more strongly about it.



On the celebrities vs influencers - I do feel celebrities can be classified as influencers depending on the context (and yes, in some cases they are not influencers but just doing a PR).  Corporations use celebrities to  promote their products/service through ads, displays, and promotions precisely because they can influence the marketing & sales impact of their products.



Also, I am still not convinced on the importance of  "HOW they did it"  on influence -- personally, I would say that the "HOW" matters less on classifying someone as an influencer - it might matter in classifying them into the "type" of influencer they are. I did read your response to Ted, and while I understand and hear  where you are coming from I would say that the onus is on each individual whether to get influenced on not -- even clicking a link online. If one were to start talking about integrity and honesty, then I would say that most folks who are currently known as influencers will fall in the non-influencer category (This is especially true in the Advertising world where most corporations use models and well-known folks for the visual and psychological impact even though the person in reality might have no interest in the product they are advertising -- and so in essence is "tricking" the audience). My personal take is that an influencer should be judged by the end result and not by the means.


In terms of expertise and its role, again I understand and respect your point of view. But I feel that expertise is secondary - and the more important one is trust and in some instances a need to have a vicarious equality with the influencer.  So in Hannah's case, I think the girls "want to be like" Hannah and that is the influencing part - more than them considering Hannah as an expert. Another example would be me influencing my kids. My kids know I am not an expert at "certain things" - however, I can influence them in those areas because they inherently trust me in making decisions that is best for them in the long run (given our family context - yes, context is critical to influence).

You made a statement in your comment, "But will companies start using Lithium if Hanah Montana says Lithium is cool?" The answer to this one question itself is very deep and can be a blog post in itself. Here is my short answer.  Influence and being an influencer is very much dependent on who is saying it and in what context/domain and who is listening (the audience). Even the most influential of influencers is 'effective' only under certain contexts and for certain audiences.



The second thing is that in many cases the effect of the influencers are heavily skewed towards 'acquisition' and decreases exponentially when it comes to 'retention'. So hypothetically, if Hanah says Lithium is cool then you are right that companies are not going to start flocking around Lithium (they are not the target audience watching Hanah or influenced by Hanah) but I would not be surprised at all if you get a spike in the #of visits and visitors and if you were to segment/profile those, most likely the spike would have been caused by the teenage girl demographic watching Hanah. Now obviously, I said 'spike' because these visitors will soon figure out Lithium is not for them and so will bounce immediately. There are many more nuances to this particular discussion ....

And lastly, I think influence like engagement can be positive or negative. An extreme example of negative influence would be the 100s of kids who are "influenced" to try out drugs. Again, the topic of negative influence can be a discussion item in itself. The methods and means a negative influencer uses can be drastically different than those used by a positive influencer. In addition, there can also be classifications based on domain - influencers in business/marketing has a totally different profile than influencers in academic (expertise definitely plays a huge role in the latter).

And now to the tie vs relationship - mea culpa on being fuzzy. I really need to put some thought into this and so will expand on this on a later comment or post. However since I mentioned it, here was my train of thinking.

I think of a "tie"  as a link between two nodes (and in this similar to a relationship). However, I also think a tie has many more quantifiable aspects to it than a relationship -- things like amount of time spent in common, reciprocity, contextual factors like sharing of a social circle, the breadth of involvment etc. As an example of this train of thought - I have a strong relationship with my grandmother  but she is really a weak-tie in my social network. Of course, if the constraint is that we are only looking at online relationships while defining ties then obviously the two converge in most cases to mean the same thing. Anyway, at this point I am going to take this off the discussion block -- and will write a post on this at a later date.

Sorry for the lengthy comment - I agree our exchanges are getting to be a post :-).  I have great regard for your views and writings and so feel this interchange of thoughts would better my understanding (or correct any erroneous assumptions I might be making).

Thanks again,

Data Science

Hello Ned,


Thanks for the engaging conversation.


What you said is precisely why I would classified celebrities as PR influencers, even though I really think they should just be call celebrities. Smiley Happy


On to the HOWs. Let’s take thing to the extreme a bit. If all that matters is the result, then would you consider someone who gunpoint me and coerce me into doing something an influencer, even though I am the one taking the action at the end? I would not. But I think at this point, the details are in one’s definition of influencer. If I were to go with your definition of influencer, then I would qualify this type of influencer as coercive influencers where they influence other’s behavior by coercion. And if someone influence by trickery, then under your definition, I would call them deceptive influencer, or under my definition, I would simply call them a deceiver. And likewise for other means of influence. I am fine with using your definition, but then everyone would be influencers then, because everyone can exert some influence and altering another person’s behavior. Even a baby crying, who made me hold him (a change in my behavior) would be consider an influencer then. They are all just different types of influencers.


There is no doubt that expertise is secondary to trust. In my earlier posts on the topic of social media influence, the last factor needed to achieve social influence is consumer confidence, or what you call trust here. In a later post, I call it the last mile of social media influence. If all other five factors are met, but the target just doesn’t trust the influencer (even though he is a real expert in the relevant topic with respect to the target’s need), then there is no influence. However, expertise or reputation in a domain can often affect the target’s confidence, so they are definitely related. But in my social media influence model, the influence process involves 2 parties: the influencer and the target. Expertise is an attribute of the influencer, where as confidence is the attribute of the target. Breaking down the notion of trust this way simply facilitates designing algorithms for influencer identification, because you can then focus on the influencers without confounds introduced by the targets.


I will take a detour back to the importance of how an influencer affect other’s behavior again. If I have a rebellious kid who doesn’t trust me, he may still do what I told him to do. So there is an action that I extract out of him but there is no trust. Am I still the influencer? Would you still say that I’ve influenced him? I guess you will argue that this is a different context. And that is probably fair enough.


I agree with the point you made about having influence and being an influencer is very different, and context is again very crucial. Also you made a very good point about the skew toward acquisition and bias against retension. So I should really say that an influencer is someone who can influence another’s decision process throughout the purchasing funnel and after purchase as well.


Yes, influence can definitely be positive and negative. In most business context, we just call the positive-influencers promoters; and the negative-influencers detractors.


On ties vs. relationship, I think all the quantifiable aspect you said about a tie translates to relationship too, even though the language may be a little different. Time spent in common definitely apply to relationship as well. Reciprocal ties are just mutual relationships. But I think I get what you are trying to say. If you use Granovetter's notion of tie strength (which is based on the overlaps of personal social network) or define tie strength purely based on the structural properties of the social network, then it is possible to have strong relationships that are weak ties. But in terms of interpersonal relationship, I think the two terms does converge.


No need to apologize. I also learned quite a bit and gain some different perspectives from our discussion, and I enjoy the discussion. What more can a blogger ask for when there are loyal, engaged and outspoken readers? Thank you for your continued interest and support.



I completely agree with you that if someone coerces you at gunpoint then it is not influencing (wanted to be clear on that :-) ). To be honest with you, I was not even looking at that particular "type" of influencing when I penned my thoughts -- the reason being I consider them to be outliers.



So I should preface my thoughts about with the basic assumption that I am not talking about forced, no-choice, coercive behavior. However, when it comes to  non-coercive action but caused by "deception", yes - now you are walking a fine line. I think they are influencers and one of the reasons being that there is an area where deception (negative) and persuasion (positive) overlap -- I know some of the super persuaders who semantically speaking use deception to guide their audience into a certain behavior.



Anyway, will stop before we start another discourse on what 'deception' means in Marketing :-). As always, throughly enjoyed the conversation.




Hey Michael,


I actually decided to go ahead and measure the share of impact, and negative does have a MUCH higher %, as expected - 38.3% and 17.4% is mixed. I wrote it up here and posted a chart.




- Maria


A very interesting and vital discussion.

While we are essentially discussing influence, there is an overlap between what we are trying to call influence vs power and understandably so. Both these terms are very much related. One can exercise Power to influence another. And vice-versa, one can influence others and derive power as a result.

Mike’s examples of coercive influence (both examples: Gun Point, rebellious kid) are more towards one exerting his power to influence. Similarly Ned, your example of the influence that you are able to exert on your kids is primarily due to Positional Power.

It is really fascinating on how this topic started in the first place. Fastcompany’s experiment has had a very fruitful side-effect in terms of a healthy discussion on Influencers: what constitutes them, how they influence, and the anatomy of influence.


To bring some perspective back to this discussion, Mike started with specifically looking at how influence works when it comes to the Social Media.

Several traditional concepts of power (Bases of Power) that is used to influence such as positional, reward, coercive etc. are not very relevant to Social Networks and media. The two types of power in my opinion that are still relevant are Referent (such as Celebrities, communities) and Expert (such as MikeJ) powers.

Being an expert in a field could result in Celebrity status within that community or domain too over a course of time (actively blogging and contributing on a specific subject over a period of time) or through an event (such as authoring a book on a subject, presenting a widely acclaimed paper). In other words that is an influence due to the referent power derived by being an expert. Alternatively, referent power can be derived through a totally unrelated field such as Show-biz that will have the power to influence in related though different fields such as Fashion. Can this happen through social media channels? I think Yes.

Let’s take a re-look at the factors that Mike detailed in the The 6 Factors of Social Media Influence

Factors or attributes related to the Influencer:

  1. Credibility
  2. Bandwidth



Credibility: Mike, you define it as ‘The influencer's expertise in a specific domain of knowledge’


I would recommend modifying it to say The influencer's expertise as perceived by the target in a specific domain at a given time’

‘Domain of Knowledge’ would in my opinion specifically apply to expert power. Perception plays a very important role when it comes to influence. A teen may buy an Iphone 4 having been influenced by let us say Hannah Montana sporting one. In case she finds that the calls are dropping off left and right, she may still give the benefit of doubt to Hannah. Even when Consumer Reports calls out the particular defect, the teen may still be forgiving of Hannah. Hannah’s influence in this case is due to the teen’s perception and highly emotional and less a factor of Hannah’s expertise. Expertise I would infer is required in varying degrees based on the domain or field.

How is this relevant to Organizations? Based on the domain that an organization operates in, they will have to choose their influencers in the social media. A Pharma company that manufactures drugs for say cancer would derive benefits through an influencer in a community who is on the drug. Hannah Montana or an expert scientist on cancer drugs will have far less influence.

Will send in more thoughts in a subsequent post.

Data Science

Hello Maria,


Thank you for coming back to the conversation.


It's so nice to see someone coming back with real data. Feels like my teaching days at UC Berkeley when students come back with completed homeworks.  Smiley Happy  Nostalgic...


Excellent analysis and data. I think the data looks much more believable now. It's not surprising though.


Finally, I must say that I really like your multi-factor approach in estimating impact via traffic, engagement, sharing, and comments. That gives a more accurate estimate and is probably more stable too.


Thank you for sharing this with us, and see you next time.


Thanks Professor Michael! Smiley Happy


I think both approaches (coverage and impact) have a role. If I just want to count the number of times someone talks about me vs. my competitors, coverage is great! But if I want to delineate the *quality* of that coverage, then impact would be  better. If blogger x and y wrote about the same thing, they'd both count as 1 unit of coverage. But if blogger x was more widely read, discussed and engaged with than blogger y, then blogger x could be worth 10 impact units, while blogger y is 1 unit (gross approximation here - but you get the point).


- Maria

Data Science

Hello Maria,


There is no doubt that coverage and impact are both important. They answer different kinds of questions and have their own uses.


Thank you again for coming back with more explanations to clarify potential confusions. You would make a good teacher/professor yourself.  Smiley Happy


Data Science

Hello Ned,


Thank you for returning and clarifying your thoughts.


I thought you would agree that involuntary action is not really influence. But if deception, psychological manipulation, etc, were used to influence a behavior and at the time the decision for action was made, there is not enough information for them to act otherwise, or stop them from acting, then I feel that this is not much different from involunary actions. If people would say "Here is blah blah blah, think about it, do some research before you click it." that would be better. But then probably no one would play the game. Smiley Happy


But you are right, deception could be a very touchy subject and another long conversation. Let's chat about that some other time. I think we will have a lot of good conversation if we meet face to face.


Thanks again for keeping the conversation going. See you again.


Data Science

Hello Whatsay,


Very glad to see you back, and glad to know that you've been following the comments and discussion.


Your comment on power vs. influence is excellent. I should incorporate some of the ideas based on power into my explanation for the 6 factors of social media influence. One of the benefits of discussion is that it will help evolve the conversation to meet the needs of the public. That is why community is such a great platform for exploring the possible use case and ideation.


I like to clarify why I define credibility the way I did. As I alluded at the end of the 2nd paragraph of my reply to Ned, I separated the attributes of influencers from the target to facilitate the design of influencer identification algorithm. Your suggestions for defining credibility mentioned the target: "The influencer's expertise as perceived by the target in a specific domain at a given time". There is nothing wrong with that. Target's perception of the influencer's credibility (his confidence or trust) is necessary factor for influence. In fact I can collapse all 6 factors together into one definition and call it influence. But that would make designing the influencer ID algorithm very difficult.


By breaking out target's perception as a separate factor among the 6 factors, the algorithm can go through each factor step by step systematically. First they will focus on identifying the "pre-influencer" (only the credible and high bandwidth user). And then check if they are relevant in content, time, and channel (and filter out those irrelevant influencers), and then find the final influencer among the pre-influencers that the target has the greatest confidence (See "Hitting Your Target" for more info on this.)


Essentially, by breaking up the influence process into 6 factors, it allows me to develop an influencer ID algorithm where each step of the algorithm is relatively easy. I'm not sure how many analytics algorithm developer will be reading this, so I'm not sure if I am just talking Latin. I hope I made this point clear. If not, please ping me again, and I will try to explain it another way.


Thank you again for coming back to voice your thought. They are great. And I hope to see you again next time.


Hi Dr. Wu,


Thanks for the interesting and substantive dialogue on influence and influencers. As the founder of the Influencer Project--which has no relationship to, but launched at almost exactly the same time and has been confused with Fast Company's Influence Project (ours has the "r")--I'm glad such an important topic is getting the treatment that it deserves. While I don't have the academic depth that you and many of your readers have, I do have some thoughts I'd like to share.


First, and as I mentioned to Mark Borden, the Fast Company Senior Editor who commented above and is spearheading the project, Fast Company deserves praise for sparking such an interesting debate. There's a saying that goes something like, "You never know how far you can go until you've gone too far." So, yes, maybe they've crossed a line--a philosophical one, I'd argue, more than a moral one--but they're brave enough to go there in the first place, so I applaud them. Plus, Mark has handled the heat with grace, so kudos to him.


In a thread between Chris Heuer, Mark, myself, and others, I wrote that I think that one of the primary issues here is a "contextual mismatch." If the context were, "Hey, click on my link because I want to be on the cover of Fast Company," it would be an explicitly "fame-based" context, and few would have such strong reactions to the campaign. Fame is a smaller, or narrower, context than "influence." Fame implies simply that people know about you; influence implies that you have the power effect important change in a particular, or multiple, domains.


So by making the campaign about influence, the good people at Fast Company conflated a larger, more philosophically important context (influence) with a smaller one (desire for fame). By saying they were going to find "The most influential person online," they effectively said, "We're going to engage in a very important, rigorous, and serious social experiment to find the one individual who can most move people to act." And that's where it becomes problematic: they're mapping a higher context onto a lower one, and calling them more or less equal.


A small context in and of itself isn't bad, per se, but it is when it claims to be a big one. The original brief from Mekanism--in which they called it The Cover, not Influence, Project--intended to explore "how things go viral and how ideas and info spread online," which "is right up Fast Company’s editorial alley." Indeed. Influence, though, is a much more complicated, complex, and emotionally charged topic than virality.


But then again, if it were only about things "going viral," we might not be having this conversation right now. I think it can all be salvaged and turned into a much bigger story and conversation that helps those of us who are interested to think more critically about the nature of influence. Again, I'm thankful that Mark's been so generous with the feedback and criticism that's come his way, and I look forward to expanding this dialogue in the months to come.



Data Science

Hello Sam,

Thank you for stopping by and thank you for the comment. I actually listened-in near the end of your Influencer Project conference via a colleague’s registration since I registered too late (after it has started). Very interesting format and approach, kudos to you!

Yes, I do agree that Mark Borden took the heat very well despite all the criticism about the project. And I also agree that there is a “contextual mismatch.” However, even if it were a popularity/fame contest, I would still argue that the methodology is somewhat flawed. I can still make the same argument that someone who uses spam, deception, manipulation, etc., to get you to click their link, is not truly popular or famous. Being a scientist, I can’t help but think about how they can make this experiment better, beside several people have asked me about it. Maybe I’ll write up a piece on that.

Even though my first few impressions on the Influence Project were quite negative, my original intent wasn’t to criticize them. I just want to point out that “Influencer” is really being over used due to the market hype. People really need to think critically when they hear the word “influencer.” They should ask how influence was measured (or inferred), and what data was used to do this inference, and how they collect these data. All these will affect the final result.

I must say that IF Fast Company only wanted some buzz out of this project, then they’ve achieve their goal. Well done! And some people get some fun out of playing this game. That is great. But I simply cannot trust the result, whether it is influence or mere popularity.

However, we may get some other insights on how viral information spread (which is a whole science itself). For example, how to driving a simple action (single click) around some content into a positive feedback loop? This is the key to any viral propagation. Or we may be able to estimate the viral coefficient of this particular experiment. This is important, because the greater the viral coefficient, the smaller number of initial participant is necessary to drive the process to critical mass, beyond which it takes off and self propagates. There are lots of interesting scientific questions that can be address with their data. Unfortunately, not on influence, not even on popularity.


Regardless, I appreciate your comment. It certainly gives a different perspective on this project.


Academic or not academic is not a problem here (I welcome everyone and all comments). The important thing is that you speak out. Social media is very new, so in a way, we are all learning. I always remember a comment by Jimmy Wales, the reason that Wikipedia become so great now, is because it is a product of countless arguments from many people with different perspectives. So thank you again for stopping by and hope to see you again.

Hello Mike/Whatsay,

Thanks for the great convo -- definitely got some new perspectives. As to the concept of power, the three of us should sometime talk about this over a cup of java -- should be  an engaging conversation :-). As both of you point out, there is an intricate relationship between power and influence.


Good day.




Data Science

Hello Ned, Whatsay, and everyone,


Definitely great conversation. Would love an opportunity to meet up sometimes. I am in SF Bay Area. Anytime you guys come to the SF area, please don't hesitate to drop me a line. We'll have coffee (or beer) and we can just let our minds wonder and explore the possibilities.


Thanks to all the participants in this thread. You all contributed a lot to this wonderful discussion with your unique perspectives.


Hello Mike,


I think it will be interesting to see if the biggest link spamming trickster is crowned the most influential.


Or will those that started with a larger & presumably well earned following maintain their initial lead?  I'd say these people have a lot to lose if they go the linkspamming route (something Mr. Borden pointed out "there may even be people involved in this project who exert tremendous influence in the time this pl...


Or will this project get even more interesting?


Social media is a bi-directional endeavour.  So far that bi-directional nature has occurred as commentary - critical or otherwise.  The debate (and thank you for your article) is interesting & enlightening.


However, what would happen if the bi-directional nature of social media took on the actual project?


What if the social media masses were able to influence The Influence Project?


Best regards,



Data Science

Hello Brent,


Thank you for the comment.


What you said is very true. It would be very interesting to see who actually came out on the top. But as it is now, I don't think there is a way to tell if they did it with spamming or by deception. Since that record is not tracked, the allegedly top influencer, can easily deny their misbehaviors. I've thought of a couple of single-point changes in the experiment that could potentially address the spamming and deception problem. For that, I invite you to read my next article in on the topic of "How To Fix Fast Company's Influence Project." I would love to hear your thoughts on that.


The idea of influencing the project in interesting. I've read your linked post above. Although some of the experiment you suggested would be pretty icky, but they would be interesting.


Thanks again for stopping by, and I hope to see you again on Lithosphere.


Hi Mike,


Thank you for the reply.


I agree.  As far as I can see, it will be hard to tell the methods used to get to the top.  Although maybe Fast Company is monitoring things.  It would be interesting to see who's links end up in email SPAM honey pots...

What did you see in particular that made your skin crawl?  The missing children idea?  I think people switching their pictures to that of missing children could help find the missing children.  I don't think there is much personal gain in doing this, so it seems to me like it is respectful.


Best regards,


Data Science

Hello Brent,


Glad we agree.


I was just thinking that maybe people would use missing children to gain people's sympathy and therefore click the link. It is yet another way that people can game the flawed experimental design of Fast Company's Influence Project. Rather than spamming and deception, I would call this psychological manipulation. In that case there would be a personal gain.


Anyway, thanks for coming back and carrying on the conversation. See you later.



Lithy Awards 2017

The winners in digital CX have been crowned!

See the winners!!