Data Science

What is Influence, Really? – No Carrot, No Stick, No Annoyance, No Trick

From the lack of response and low kudos on my last post, I get the impression that you probably had enough big data for the time being. Since I’ve decided to rotate more frequently between different projects that I’m working on, I feel this is a good time to revisit the topic of influence and pick up where I left off on this fascinating subject.


It’s been more than two years since I wrote about influence. You can find my chapter plus miscellaneous articles on influence via the label under “influencers”. I didn’t lose interest in this subject – the simple fact is empirical research takes a lot of time, much more than qualitative research. I also conduct research in many areas besides digital influence. If you’d like a quick recap of my earlier work, this video interview by covers the essence of what I did.



Today, I’d like to continue the discussion about influence and look deeper into what influence really means.


What is Influence? The Simplistic View

Despite the amount of discussion around influence on the web, I find that people have a general notion of what influence is, but can’t define it precisely. So let me start with two relevant definitions from the Webster’s dictionary. Influence is:

  • The act or power of producing an effect without apparent exertion of force or direct exercise of command
  • The power or capacity of causing an effect in indirect or intangible ways : sway

No Carrot_w.png


From these definitions, it is clear that influence is something that must have some effects on something else. The question is what those “effects” are, and what that “something else” is. For business and social media, that “something” is almost always going to be a person, or a group of people. We will refer to these people as the “influencees”.


The effects, on the other hand, can be a change in thought (i.e. sentiment, opinion, attitude, etc.) or it could be an action (i.e. purchase, referral, behavior change, such as loyalty, etc.). I must draw your attention to the disjunctive “or,” because contrary to popular belief, influence does NOT always have to produce an action. If you’ve convinced me (an Apple fan) that an Android phone is better than an iPhone, you have successfully influenced me, because you have changed my mind. Even though I haven’t taken any action yet, influence still occurred. Although that is not the most valuable form of influence, it is influence none the less. As far as influence is concern, it is sufficient to simply cause a change in someone’s thoughts.


So it seems that influence is just the ability to cause a change in thought OR action. Simple! But this definition is insufficient and incomplete.


The End Does NOT Justify the Means

The reason that the simplistic definition is incomplete is because it does matter how you cause those changes, whether it’s in thought or action. The Webster definitions also state there should be no force or exercise of command, and the effects (i.e. the changes in thought or action) must be caused through “indirect or intangible ways.” What does this mean? Simply put, when it comes to influence, the end does not justify the means.


No Stick_w.pngIt is actually relatively easy to cause a change in someone’s thought and/or behavior, because there are many different ways one can do this. The following is not an exhaustive list, but it covers some of the most common categories of method to change someone’s mind and/or actions:

  1. Carrot: paying the influencees, bribery, quid pro quo, etc.
  2. Stick: force, coercion, command, control, fear, etc.
  3. Annoyance: spamming, link bating, frustration etc. (e.g. Social Fame, Social Shame, and the Accounting of a Game)
  4. Tricks: deception, hidden agenda, any means that are not completely transparent, etc.


However, none of these constitute real influence. Certainly, if you threaten someone at gunpoint, you can easily change his mind and behavior, but that’s not influence. You can also pay someone to do something; that will certainly change his actions, but that’s not influence either.


As you can see, it is much harder to truly influence someone. Because influence really is “the ability to cause a change in thought OR behavior through non-coercive and transparent means where the influencees voluntarily want the changes even without monetary compensation.”


Applying the Definition

Now we have a definition for influence, let’s see if we can apply it to clarify some confusion. If you ever work on any influence model or influence scoring algorithms, you probably have thought about this question: should the number of follower on twitter contribute to one’s influence? Some of you may say nope, absolutely not, yet some of you may challenge why not.

No Annoyance_w.png 

The answer is it really depends. Again, the key is HOW you get people to follow you. If you pay people to follow you, that is definitely not influence. That is called a bribe. In fact, if you use carrot, stick, annoyance, or tricks to get followers, then it is by definition not influence. However, if you acquire your follower by produce great content and tweeting informative news, then that would be a valid form of influence.


Although twitter users have become savvier today, most users will still follow you back blindly as a quid pro quo to your initial following. As we discussed earlier, this quid pro quo behavior is not influence, because people can easily game the system by simply following random strangers all day long. They would accumulate followers due to this quid pro quo following behavior, even though they haven’t done anything worthy of any influence. Therefore follower count shouldn’t be used in any way to measure someone’s influence, at least not now!



So influence is NOT just the ability to cause a change in thought or action, because it matters how you cause the change. We must extend this simplistic definition to include the valid means of causing a change.


No Trick_w.pngIn short, my definition of influence is: the ability to cause a change in thought OR behavior under the following four conditions:

  1. No carrot – means that do not involve monetary compensation
  2. No stick – means that are non-coercive
  3. No annoyance – means that are voluntary
  4. No tricks – means that are completely transparent


Alright, now that we have a definition, I hope everyone can be on the same page when we talk about influence and what it entails.


Meanwhile, let me know if you like me to continue writing about influence by either commenting or giving me a kudo. In the future, I will use the number of comments and kudos I received to gauge how frequently I should rotate between various subjects that I’m researching. Stay tuned and see you next time...



Michael Wu, Ph.D.mwu_whiteKangolHat_blog.jpg is 927iC9C1FD6224627807Lithium's Chief Scientist. His research includes: deriving insights from big data, understanding the behavioral economics of gamification, engaging + finding true social media influencers, developing predictive + actionable social analytics algorithms, social CRM, and using cyber anthropology + social network analysis to unravel the collective dynamics of communities + social networks.


Michael was voted a 2010 Influential Leader by CRM Magazine for his work on predictive social analytics + its application to Social CRM. He's a blogger on Lithosphere, and you can follow him @mich8elwu or Google+.



Excellent post that lays the groundwork for a lot of discussion around both social and gamification. In fact, it's interesting to watch those fields evolve, since it seems much of the structure was hastily built before the foundational thinking was in place. I'm probably as guilty of that as anyone. 


When you apply the definition to Twitter, it makes me wonder. Agree that sheer follower count is a bad proxy for influence. What about in combination with the user's ratio of followed to followers? Only if that person hasn't used shady tactics to acquire followers, I suppose.

Data Science

Hello Silas,


Thank you for taking the time to comment here.


You are right on and thinking ahead... I will eventually relate all these with gamification. But that will be a much later post. There are lots of interesting topics that are more fundamental. And we should cover them before I discuss the relationship between influence and human behavior, which is what gamification is all about.


What you use really doesn't matter. You can game the ratio of followed to followers too. Simply follow a bunch of people and unfollow them. In this case, the person who follow back is thinking of returning a favor (i.e. quid pro quo). But later he will find that he's been tricked into following, b/c he's been unfollowed. This is worse than just carrot. It's carrot and trick. So it is definitely not influence. But again, if no one does this then it is OK to use this ratio in the estimation of one's influence score. It really depend on how you get that ratio...


It is HOW, now WHAT (whether it is followers, or followed to follower ratio, or anything else) that we should focuse on.


OK, thank you for the comment. Hope this helps.

See you again next time.


Excellent post!


I really appreciate you laying out a "definition" to an Inlfluencer.


What are your thoughts about adding emotion into the definition or algos of influencers? Basically what I'm talking about is an influencer who can get people to respond with extreme emotions such as 'excitment', 'frustration', 'infuriation' etc. based on a tweet, fb post, blog etc that the influencer has posted online. If you could measure the intensity, type and frequency of emotion an influencer was able to get others to exude it would provide a much better understanding of the degree of Influence someone may have.


Emotion could play a much greater role in understanding the effect an influencer has on different topics and different demographics. It also could be used to help identify influencers much more than it does today. 


Data Science

Hello MT,


Thank you for the comment.


You raise an interesting point about emotion. But I want to clarify the difference between influence and influencer. They are different. I only defined what influence is, not the criteria to be an influencer. There is also the difference between definition and correlated variables. Definition is what defines the concept (i.e. influence in this case), but there are variables that might be correlated or even predictive of the concept. 


For example, a bird is feathered, winged, bipedal, endothermic (warm-blooded), egg-laying, vertebrate animals. That is a definition, every bird must have feather, wing, 2 leg, warm blooded, and lays eggs. But flight is a highly correlated, and predictive variable, because if an animal is a bird, it's most likely capable of flight. But that is not a definition, because there are some birds that doesn't flight (e.g. penguins, ostrich, weka, etc.). So the capacity to flight is not a definition. It is only a highly correlated variable.


Likewise for emotion. It may be a highly correlated variable, and we have to look at the data to see how highly correlated it is. I don't actually have emotion data, so I can't say for sure. But it is definitely not a definition. Because there are certainly, and probably a lot of, very influential influencers who does not use strong emotion to influence.


OK, I hope this address your question.

See you next time.

Hello, Kudos on such insightful article.


It would seem that a measure of influence would somehow, have to include a correlation between the influencer, and influencee relative to demographics of influence. Data in terms of ratio of followers to followed may be a weak rationale for measurement given the emotive thrust of the criteria behind followership given a particular area of influnce.


These are just some of my thoughts.

Data Science

Hello Emeka,


Thanks for the comment.


You get the idea. You have to consider both the influencer and the influencees in order to quantify influence. You actually cannot measure influence no matter what you do. We will discuss what does it take to really measure influence. More on that coming on the next article, so be patient and stay tuned. But it's good that you are thinking ahead already.


Thanks for the thought. And hope you will continue this conversation on the next article.

See you next time.


Hello Michael,

Sorry but i read your post only now...

I understand that we should perhaprs analyze influencees to identify the influencer? Or shoppers, to look for someone who could have influenced them?

Other question: have you identified if there's people who are never influenced? What are their proportion?

many thanks


Data Science

Hello RaphaelleM


Thank you for the comment and question.


Before I answer your question, I must say that the answer is not as simple as the question you phrased. But the simple answer is that I don't believe there is anyone who is never influenced. Everyone is influenced on something by someone at point in time.


However, it is definitely possible that some people are never influenced on a specific topic (e.g. euthanasia, capital punishment, abortion) because they have a strong point of view, but they will be influenced on some other topics. For a brand's concern, it is definite possible for people to be never influenced on a particular product or service (e.g. time share, etc.), but these same people will be influenced on other products and services that they don't have a strong view. 


So the more relevant question would be are there people who are never influenced on a particular topic, product, subject, etc. And the answer to this question is much more complex, and it will depend on the audience and demographic, even cultural or geolocation. 


Anyway, I hope this helps clarify your question.

Hope to see you again next time.


It's not that the definition of influence was ever insufficient or incomplete, Michael. 


All of the others examples that you provided are indeed examples of influence. You've simply presented an ideal to which some people might wish to aspire, and you cherry-picked definitions from a single source to support it. There's nothing wrong with that. However, I wouldn't expect the other forms of influence to disappear any time soon....and like it or not, they do represent influence.


I measure influence by measuring outcomes--always keeping in mind that influence is contextual. Let's use your Twitter example. One's number of followers does tell us something about one's ability to influence people to follow him (whatever their reasons) as well as retweet his comments. However, it reveals little if anything about that individual's influence beyond Twitter follows and retweets. 

Data Science

Hello Joseph,


Thx for the comment. 


However, I must say that I disagree with you. I do not believe coercion, bribery, deception, etc are real influence. They are ways to affect behavior, and outcome. But there are hundreds of millions of ways to affect behaviors and outcome. Not all constitute real influence. That is why we have different verbs to describe them.


That said you are entitled to believe what you like and do what you like.


You said it yourself, "One's number of followers does tell us something about one's ability to influence people..." It is the ability to influence, NOT influence. Capacity to influence and influence are not the same thing. If you don't understand the difference, I suggest you read this WIRED article on why brands still don't understand digital influence, or Brian Solis' White paper.


Thank you for the comment, nonetheless.

And see you around.

hi, Michael


how do u measure a person's power of influence in social media? it's easy to make up fake followers. 

Data Science

Hello Lian,


Thx for the comment.


As you mentioned, it is now pretty easy to get fake followers, so followers shouldn't be a factor in measuring someone's power to influence. To further answer your question, I suggest you read the following articles I've writte on this subject.


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

Finally, Here is a step by step guide to the computation in a community setting:
Community Influencers Step by Step.

OK, hope this addresses your question.
See you next time.

Smiley Very Happy good


Lithy Awards 2017

The winners in digital CX have been crowned!

See the winners!!