In this episode, Katie and Chris discuss the different algorithms for identifying influencers on social media services and advantages/disadvantages of each. They look at network graphing databases and how many influencer identification tools struggle to correctly identify influencers based on business goals.
Subscribe To This Show!
If you're not already subscribed to In-Ear Insights, get set up now!
- In-Ear Insights on Apple Podcasts
- In-Ear Insights on Google Podcasts
- In-Ear Insights on all other podcasting software
Advertisement: Data Science 101 for Marketers
Do you want to understand data science better as a marketer? Would you like to learn whether it’s the right choice for your career? Do you need to know how to manage data science employees and vendors? Take the Data Science 101 workshop from Trust Insights.
In this 90-minute on-demand workshop, learn what data science is, why it matters to marketers, and how to embark on your marketing data science journey. You’ll learn:
- How to build a KPI map
- How to analyze and explore Google Analytics data
- How to construct a valid hypothesis
- Basics of centrality, distribution, regression, and clustering
- Essential soft skills
- How to hire data science professionals or agencies
The course comes with the video, audio recording, PDF of the slides, automated transcript, example KPI map, and sample workbook with data.
Sponsor This Show!Are you struggling to reach the right audiences? Trust Insights offers sponsorships in our newsletters, podcasts, and media properties to help your brand be seen and heard by the right people. Our media properties reach almost 100,000 people every week, from the In Ear Insights podcast to the Almost Timely and In the Headlights newsletters. Reach out to us today to learn more.
Watch the video here:
Can’t see anything? Watch it on YouTube here.
Listen to the audio here:
- Need help with your company’s data and analytics? Let us know!
- Join our free Slack group for marketers interested in analytics!
What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for listening to the episode.
Christopher Penn 0:02
This is In-Ear Insights, the Trust Insights podcast.
In this week’s In-Ear Insights, we are talking influencer marketing network graphic databases and so much more.
So as we talked about on last week’s show, one of the things that we found in our Instagram analysis was that Instagram influencers, for good or ill are four times as effective as brands, in terms of just getting engagement on Instagram.
And so it actually brings up the question well, that how do you find these influencers? How do you know who’s the real deal and who’s just, you know, talking to themselves a lot.
We all know plenty of people who love to wax rhapsodic on, on the internet and, and pose their deep questions to the universe.
But sometimes the universe doesn’t really respond.
So Katie, when you think about identifying influencers, and who even constitutes an influencer? What do you think about and how is it different from, you know, most number of followers or largest number of engagement?
Katie Robbert 1:05
When I think about a good, solid, valuable internet influencer, I’m using the word internet, because it kind of covers the different social media channels, you know, so it could be Instagram, Twitter, whatever.
But basically, it’s someone who’s making those connections.
It’s someone who is a subject matter expert in their space, whether it be fashion, beauty, martec it, you know, whatever it is, and who can, you know, bring people into a conversation.
And it’s almost like, you just want that influencer to start the conversation, so that other people can continue it on and reference back to that person, but you don’t necessarily, I think the problem with influencers is that sometimes you’re looking for someone who’s just like, shouting at everybody, and just making noise and who’s the loudest, but I don’t think that that’s a really good valuable influence, you might get a lot of eyeballs on it.
But that doesn’t mean that people are going to engage and take action with whatever this influencer is saying.
So when I think about influencers, I think about someone who is, you know, that idea starter, that conversation starter, that seed planter who can say, Hey, guys, I’m going to drop a little nugget of information here, go run with it, and it sort of spreads far and wide, but always references back to that person who started the conversation.
Christopher Penn 2:31
Sounds like a good spirit, see theorists leader, like here, go do you?
Katie Robbert 2:35
Well, maybe, I mean, that’s a type of influencer.
Christopher Penn 2:41
It certainly is.
Yeah, you know, it’s funny A number of years ago, we wrote a book, it’s no longer in circulation, maybe we should dust it off and, and updated on identifying influencers.
And the three categories that we came up with at the time were, you know, sort of the thought leader, you know, the person who, for the people who are old like me, there were commercials back in the 70s, and 80s, for financial expert called EF Hutton.
And then the tagline to their commercials was when EF Hutton speaks, everybody listens and knows all these things.
Like, you know, he was on a train, he started to talk, and this everyone the train stopped and looked at him.
So you have that sort of that thought leader, and those are people who don’t have to be loud.
In fact, usually not in the marketing space, you know, one of the ones that that people reference the most is Seth Godin.
And he isn’t on social media, he really doesn’t participate.
But yet, he’s at that thought, Peter, then you have the middle group we call network hubs, or, you know, mayors, if you will, the people who know everybody, and, again, those are folks that don’t have to be loud, but they know everybody, you know, and you say, hey, I need to get a meeting with the VP of Marketing at Dell.
And oh, yeah, Bob, and I went to college together or, you know, Bob’s my cousin’s friends, whatever.
And those people that particularly for b2b, you know, that’s a, those are the influencers that will land multi million dollar deals.
And then the third category is super, like you said, the Kardashians, so allowed people and stuff.
One of the challenges, though, that you have with a lot of influencer marketing software out there is that it’s only really good at that latter category.
Katie Robbert 4:26
Yeah, it’s because I think that there’s still not a lack of understanding of influencers, but on behalf of companies sort of a lack of, you know, long term planning of how influencers could be used to boost their brand.
And so a lot of times, it’s a well, we have this event going on, who can we get who can reach the most people to promote our stuff, which is a fine and decent plan, but there’s no real long term thinking in that you know, is that person going to stay with your brand long term To really, you know, help be an ambassador for it, because that’s really what you’re after is a brand ambassador, someone who’s going to constantly be known and associated with your brand.
And you know, so Chris, to your point, you know, someone like Seth Godin, like if he’s not on social media, but people are referencing him.
If he starts saying like, Oh, you know what, I exclusively work with Trust Insights, will everyone’s going to move to start paying attention to what the heck it is that we’re doing.
And so that’s the kind of longer term planning that, you know, we would want to be thinking about, especially in the b2b space, in the consumer space.
I do think it’s a little bit different, especially if you’re just looking for that like, celebrity thumbs up of endorsement, because you can kind of like switch out influencers, depending on who’s hot that minute.
Which, again, is it’s short term thinking, unless you’re going to be building up that influencer along with your brand to keep them long term.
Christopher Penn 6:00
Yeah, that really is the catch.
And so one of the challenges I think that marketers face in the influencer marketing space is the fact that most of the software really is not tuned to identify anything other than a lot of people, like a lot of people are easy.
You can, and many pieces have reduced, like this person has, you know, 142 million followers, you know, Taylor Swift has the follower base of a large, you know, nation like the only difference between Taylor Swift and a nation is that most nations have navies, and Taylor Swift does not yet we’ll see
Katie Robbert 6:32
is that is that on our list of things to do was to build a Navy
Christopher Penn 6:38
writing, writing albums during the pandemic, and stuff like that, who knows it’s possible.
But the other categories of influences are harder to see.
Because for two reasons, one, not all the data is available.
Like certainly, you can see some what happens on Twitter, which is probably the most used and possibly the most overused network for influencer identification, you can see some of the data on Instagram, you really can’t see much on Facebook, and you can see zero on LinkedIn.
And so it becomes very difficult to identify those other classes of influence.
But the other challenge that people run into, is they don’t have the technology background to digest that data and turn into something useful.
Katie Robbert 7:25
I’m assuming that there are I’ve never personally had to look for this kind of software, but are there off the shelf pieces of software that will identify influencers for you? And to your point, Chris, do they just look for like, you know, largest number of followers?
Christopher Penn 7:41
There are tons tons of influencer identification, you know, companies out there.
And yeah, a large number of them either look at one of two things, who’s got the loudest mouth or the biggest audience? Right.
And and on the b2c side, at least, you know, who gets, you know, decent amounts of engagement on a per piece basis.
That’s about it.
There isn’t much else out there.
On the b2b side.
There are a couple of additional pieces of software out there, like tracker Analytica, a few others that do have slightly more sophistication.
We actually were at a marketing conference, we had a chat with the CTO of Analytica for a while and we’re discussing the different algorithms they use and stuff and have a philosophical difference between them, but at least that they’re using more advanced.
Katie Robbert 8:33
So it sounds like there’s a couple of limitations based on what’s available right off the shelf.
And it the limitation might be what it is that you want to do with an influencer.
And so if you’re looking for someone to, you know, grow along with your brand, you might not be able to find that person through these, you know, pieces of software because of their lack of following for now.
You know, you could probably too soon to say I want someone who’s in this space with this many followers, but you really need to do the work to get on those platforms.
Before I this is my opinion, I feel like you need like, if you want an Instagram influencer, you need to get on Instagram, and start looking and searching and doing that manual.
Like, what is this person look like? How do they engage? You can do that with some of the software but nothing replaces really just getting into that platform itself and seeing who’s there.
Christopher Penn 9:31
And one of the things I think is philosophically different that in the way we approach it than than others do is in what constitutes influence, right.
So again, most platforms most software most people think okay, who is talking the most, right who’s who’s the conversation starter.
And, at least in you know, in the graphing software we use, we use what’s called network, graphing saw We’re to look at the interactions between different people.
I like to look at who is most talked about, because again, Seth Godin, not on social media, but yet Seth Godin, everybody talks about him.
When we look at Twitter mentions, or Instagram mentions, or any of the mentions that are available on these different data feeds, it’s, it’s more important, I think, to see who’s talking about you, rather than who you’re talking about, because you can wax poetic all day long.
And you know, post 100,000 pieces of content on LinkedIn, but if nobody ever talks about you, then you don’t have presence of mind, your brand is not top of mind.
Whereas if people are talking about you constantly, you know, you gotten into people’s heads.
And so from a computational perspective of influence, I think there’s a lot more value in knowing who is most talked about.
And then, you know, obviously, that whoever those people are, if they start referencing you, like you were saying, if you have someone who’s talked about as the revered expert, and they mentioned, you know, TrustInsights.ai, like, wow, people will pay attention to that, you know, there’s there’s folks in every industry who just, they are the folks that everybody looks to and follows to see what they recommend.
And when they recommend something, you know, the most classical example in the B2C was when Oprah mentions a product, that company had better be ready to sell out immediately.
Katie Robbert 11:32
So, you know, we looked at some of this data on last week’s live stream, we looked at the hashtag marketing, Twitter.
So we’ve talked a little bit about how there was this movement at the end of last year, to really build up this sub community within Twitter, for the marketing.
And so you can see the largest bubble there is the handle that Christina Gee, that’s Christina Garnet.
And she started, I wouldn’t, I don’t know if she like originally founded the idea of the marketing Twitter hashtag.
But what she did was she tweeted out this, you know, post talking about if you want to make more connections, if you want to get more followers, then respond to this thread.
And it really sort of took off in what you could consider a viral way.
And so now, you know, three and a half months later, it’s something that people are still talking about on Twitter, which is a big deal, especially in a social media space for people to have that long of a memory on a social media platform, because things happen instantaneously.
And then you forget.
And I constantly see two things happening.
One is, Christina is constantly boosting other people.
She is retweeting people’s posts, constantly, all the time.
She herself is tweeting, conversation starters.
And then when people are posting, they are referencing her without her saying, hey, talk about me.
And so that’s why her bubble in this network graph is so big, because she’s doing a lot of talking but not talking at she’s actually having those meaningful conversations and conversation starter, she is engaging with people and people are talking about her an awful lot because of what she did for that community.
And she’s still working to build it.
Christopher Penn 13:18
And for those who are listening to this episode, and can’t see what’s on screen, if you go to Trust insights.ai slash YouTube, you can see the video portion of this, we’ll also put a an image of this in the blog post over at Trust insights.ai slash ti podcast, you can see what’s going on what we’re referencing, but it basically looks like a big spider web with lots of little and big circles in it.
So part of the reason why I think this is so important is that it network graphs and graphing databases, these are not new, these have been around for 50 ish, some odd years.
And they’re pretty straightforward to understand conceptually, when you have, say, two nodes, you and me, for example, and we communicate to each other in any fashion, you create, these were called edges.
And it’s basically it’s a it’s a line connecting two dots.
And the more you and I communicate to each other, the bigger that connection should get growing.
Now if you add and say our partner, john, and that and you know, john only talks to me my.on the map would get bigger than then say yours, John’s would because we know there’s there’s more communication coming into into my node.
And what we want to do is take this data of who’s talking to or about whom, inside of Twitter and be able to or Instagram and map this out to understand Oh, this is this is who is most talked about, and that gives us the ability to say, Okay, I need to talk to Christina Garnett, or Michelle Garrett or any of these other folks and say like, yeah, let’s, let’s have a chat with them and see if there’s a way to organically answer sensibly work ourselves into, into their mind space, you know, the share of mind.
So that when somebody says, Hey, who do you know, that can help us with our analytics we come up to in their mind.
And then that recommendation spreads to their community.
The challenge with network graphing, well, there’s a bunch of challenges.
One, there’s not a lot of technology out there that will process the data itself.
Like the software, there’s there’s a ton of good software packages out there that do it, many which are open source, like Neo for j, GFI, AI graph, etc.
But they all presume that you’ve got the data pre processed and almost nothing doesn’t, we have to write our own software to do it because it wasn’t anything else you could find it.
And then you have to have some level of domain expertise in network graphing databases to know which algorithm to use.
This is going back to what we’re talking about earlier.
With Analytica, they use a harmonic centrality algorithm, we use an eigenvector centrality mechanism.
And they’re they’re two almost opposite ways of looking at communications, we look at who’s most talked about, and they look at who’s doing the most talking.
Katie Robbert 16:19
It’s interesting, because I think that this is one of those really misunderstood pieces of the process.
It’s it always goes back to what is the question you’re trying to answer? It’s so having that plan.
And, you know, part of that plan, as we’ve been talking about on the live, so what live stream every Thursday at 1pm.
Eastern is, you know, that exploratory data analysis process and really sort of coming up with the business requirements, the data requirements, and so not skipping over that piece of the process to say, Do I even have the data in order to, you know, answer this question? If the question is, you know, who’s the best influencer for my brand? Then you need to know where that data is coming from, you know, or do you want an Instagram influencer? Do you want a Twitter influencer? Do you want a Snapchat or clubhouse influencer? You know, you need to decide and then you need to figure out does that data even exist?
Christopher Penn 17:24
And again, this is your absolutely, this is so important.
Knowing your goals, determines which algorithm you’re going to use in a graphing database.
If you just want, you know, a human being that you can stick a credit card into an ads come out, that’s, that’s fine.
You know, that’s, that’s a kardash.
Katie Robbert 17:43
Christopher Penn 17:45
I don’t know the Kardashians seem to be doing reasonably well.
That’s one type of influence.
And in that case, you’d really do want the loudest most engaged people in the end for that application makes total sense.
The challenge is, those people tend to be very, very expensive.
If you want that person who can broker deals, you know, then analytics choice of like, a harmonic centrality mechanism, actually makes great sense, right? Because you want somebody who can say, Okay, yeah, I know this person, you know, and I can get you in the door with this person, and so on and so forth.
But if you want somebody who is that thought leader, who can boost your brand and reputation that you want, and the first album, the eigenvector centrality mechanism, because that person is the one everybody else looks up to as an actual authority.
And so if you’re not clear on your goals, you won’t pick the right algorithm, and then you’ll end up with the wrong influencers.
Katie Robbert 18:40
So let me ask you this question, Chris, because you’ve just described three algorithms.
And so let’s, let’s just go ahead and assume that the majority of marketers don’t necessarily know which algorithm that they would want, if they do their homework to say, these are my goals.
This is the data that I have this is the outcome that I’m looking for.
Do they need to understand how the algorithms work? Or can they run the data until they get the right kind of outcome.
Christopher Penn 19:10
So marketers without the technical background are a little bit stuck, because and this is not their fault.
A lot of the influencer marketing companies will not disclose what algorithm they use.
They say we have our proprietary in house bla bla, bla, bla, bla, you know, the usual sales garbage.
And they won’t tell you what’s in the box.
And so you’ll just get, you know, lists of influences back and you don’t know how they were chosen.
And chances are the person you’re working with, or the company also has no idea because they aren’t the ones who engineered the software.
The only way that you can be absolutely positively sure how it was selected, is if the company tells you which, you know, again, props to analytics for being willing to admit like, yes, this is exactly the algorithm we use.
And now I know when they have the use case for what they’re describing works and when it doesn’t, or you kind of build it yourself.
Because otherwise, you I don’t trust anybody who says this is a proprietary algorithm, we won’t tell you anything about it.
Like, you can tell somebody, the core algorithm underneath the hood, without giving away the secret sauce of how you process the data and how you know what cut offs you would make for saying, Yeah, this is the selection crowd.
You don’t have to give away the secret sauce.
It’s to say like, yeah, this is what is it’s like going to a restaurant saying, you know, here’s our mystery dish, Well, no.
Is it a sandwich? Is it the sushi, like, you want to tell me the exact ingredients, but just telling you what is I’m going to eat? The same is true of this stuff is like saying, just tell me what’s generally you’re doing under the hood.
And most of these companies won’t.
Katie Robbert 20:41
So ideally, if I’m following correctly, ideally, if I did my homework, I put my plan together.
And I said, I want someone who is a conversation starter.
I want someone who has, you know, a big network doesn’t have to be like millions of people, but it has to be the right kind of network.
If I go to one of these companies, they should be able to say, Yes, my eigenvector centrality software does that like so you don’t have to understand what I I get vector centrality is, but they should be able to say our algorithm looks at total number of followers, our algorithm looks at total engagement, total reach, likes, retweets, whatever it is, and combines that for one general influencer score, like they should be able to break it down for you.
In those terms, they don’t have to give you the exact like formulas of it’s like x y plus B plus r q squared equals, you know, 95.
But to your point, Chris, it sounds like they should be able to say, this is what we consider in our algorithm that gives you the output in plain English.
Right ideal, not
Christopher Penn 21:55
very common, it’s not very common, it’s not very common, you know, it is one of the reasons why we’ve had to build a lot of this stuff ourselves, because we could not get that information out of somebody else.
And I’m not saying you have to go and build your own software, although certainly it does help.
If you influencer marketing is going to be, you know, part of your core strategy, you might want to invest the time and the resources to build some of the software yourself or work with an agency that does it for you.
But if it’s not, then, you know, just, I guess, pick vendors and hope it works out.
But I’ve I’ve not found a lot of companies that are willing to even be the slightest bit forthcoming.
Because I think in some ways, they feel like that it’s a commodity.
And if they give away any part of the recipe, somebody else could steal it.
And I think there’s some truth to that.
But at the same time, it’s, you know, there’s so much that goes into the construction of software like this, that even small changes along the way, can make a big difference in the in the end outcome, it’s, you know, it’s just like cooking, you just put in a pinch too much salt at one point in the process.
And then you’re basically eating a salt, like at the end, as opposed to the dish you’re looking for something we’re just saying about I want who is most talked about, but I also want a certain following size.
Now you’re talking when you think about the data, okay, I have a cut off point and send some data will not get included, you have you have to know, if you’re building this stuff, where in the process, do you make that cut? naively, you might think I’ll cut it off at the beginning.
But then you won’t get all the people, the mentions from lesser folks, in terms of audience size, who talking about that other person.
So you have to know, in the process where to make that cut.
So there’s there’s a lot of nuance, even to a very simple set of requirements, like you just gave.
And so it’s really important that those requirements be very explicitly declared.
Katie Robbert 23:43
Well, and I think that, you know, your point is well stated of, you know, the way that I would program an algorithm versus the way that you would program the other, we could take the exact same algorithm, but because you and I are different people, it’s going to come out differently, because we have our own set of ideas, even if we start with the exact same set of requirements, the exact same algorithms, the exact same data, I think that’s something that, you know, these companies that are trying to blackbox it are forgetting is that, you know, their team, their, the composition of their team is unique to them.
That’s what’s proprietary, then, you know, so if you if you can, you know, reverse engineer what the algorithm is, your outcome is still going to be different from theirs because you are not them.
And yeah, I know, we always kind of come back to like, at the end of the day, humans are the ones who programmed these algorithms.
But that’s the bottom line.
Is that you the human are the start and the end of the process.
Christopher Penn 24:44
It’s like the shows on YouTube is that there’s a hilarious series of like, you know, celebrity versus a you know, an expert chef, they both try to make the same dish and obviously the celebrity has no cook experience is completely hosed is it same tools, same ingredients, same process, but the human body gigabit but the master chef makes this amazing edition this lovely capelle flaming garbage.
Like, it’s, it’s very clear that that expertise does matter at something.
And so I guess to summarize when it comes to this to influencers and network graphing and all this technology, you have to know what’s in the box.
If you don’t know what’s in the box, even conceptually, you don’t know that the results you’ve been given are good or bad.
You know, again, going back to a food analogy, if you don’t know how addition is made, and you say you have an allergy, you could be in a lot of trouble.
You know, so those requirements upfront really matter when it comes to influencer marketing, the process matters.
And the the people, not only you know who you’re trying to manage for influences, but who’s building the stuff for you, or who’s built who the agencies you’re working with.
You got to know the people too.
It’s it’s almost like, you know, people process and platform actually matter together more than they do individually.
Katie Robbert 26:02
that’s a that’s a weird concept, Chris.
So if you can bring up that graphic again, one more time as we close out.
So you know, as we’ve been talking about the different algorithms, so if you were to post this on a social media platform and say we did a network graph of the influencers, the first question that most people should be asking is, what is this mean? And we can tell them, well, we use Gaffey.
We pulled Twitter data.
And the question that we wanted to answer was, who is being the most talked about? If you are looking for influencer marketing software to tell you who are the influencers? And they can’t answer those basic questions for you, then you should probably, you know, run away screaming, or, you know, gently walk away either way, you need to know how it works.
And so we are telling you straight up, how we do it, we use Gaffey.
We pull the data straight from the networks themselves.
And oftentimes we are looking for Who is the most talked about and who is making those connections.
Christopher Penn 27:04
Yep, as you’d like to see more of this, go tune into our live stream show.
So what we’re going to do in a couple weeks do a walk through from beginning to end of the process so you can actually see a good chunk of the process yourself.
So that’s Thursdays at 1pm.
You can find that over at TrustInsights.ai dot AI slash YouTube for our YouTube channel when MAE places that area airs.
If you’ve got questions about what we’ve talked about, in today’s episode, head on over to Trust insights.ai slash analytics for markers, our free slack group who are 1500 1600 people now chatting, all things analytics, and a bunch of other fun stuff too.
We just added like 12 new channels on a whole bunch of work life balance stuff.
And again, wherever you’re watching, listening, if you want to head over to Trust insights.ai slash ti podcast for other ways to consume content, which is whatever is most convenient for you.
We’ve got all the options there.
Thanks for watching and listening and we’ll talk to you soon.
Take care want help solving your company’s data analytics and digital marketing problems.
Visit Trust insights.ai today and let us know how we can help you
Need help with your marketing data and analytics?
You might also enjoy:
Get unique data, analysis, and perspectives on analytics, insights, machine learning, marketing, and AI in the weekly Trust Insights newsletter, Data in the Headlights. Subscribe now for free; new issues every Wednesday!
Want to learn more about data, analytics, and insights? Subscribe to In-Ear Insights, the Trust Insights podcast, with new 10-minute or less episodes every week.