Saturday Night Data Party: Slack Community Analysis

Welcome to Saturday Night Data Party. Tonight we are talking about Slack. So Slack is one of the most popular private social networks available. It is right up there with Discord. For those who are not familiar, Slack is a chat room essentially.

If you look at it, if you look at Slack, it Slack is actually for those of us who are really old and have lots of gray hair. So if you remember internet relay chat from the 90s, this is it. So here's the analytics for marketer Slack. If you'd like to join it, um I'll put the link in the chat. Uh it is trustinsights.ai/slash analytics for marketers.

And it really is an IRC system, right? 1200 members in here. Now, here's the question. How many people took the advice that we had? And B, how would you know how popular Slack is?

Right. Well, when as you saw in the channel there, when somebody wants to join a Slack, they will get a link to share that will get them to a Slack instance. So you can see there's the URL. But it all of them are uh something.slack.com, which means uh we can figure out how many of these things are there are by using SEO tools. SEO tools, uh this here we're using RFs.

Tell us how many inbound links to a domain there are. So in this case, we have Slack, right? Slack.com. Let's look at the backlinks. Cool, let's get that downloading.

Um we see, you know, Feedly has its Slack community. Uh WordPress has its Slack community. You can see WordPress.com. You can see where we're going with this, right? By using an SEO tool, we can start to extract the data and get a sense of just how big is this thing.

Keep on going here. Join uh so there's a join one here. Apps, uh let's see, keep on going. Nomad. Cool.

So let's take this data and get it into a format we can analyze. So let's bring in our Slack data here. And let's call this. All right. Did we put the file in the correct folder?

We did not. Alright. Let's go. Up one level here. There we go.

Let's save this. Alright. Oh, you are not going to play nicely, are you? Oh, the the file is at the wrong root level. There we go.

Quit and try that again. Okay, let's open up our Slack folder here. That's better. So this is 86 megabytes of good old raw text. What do we have?

We have total backlinks, domain rating, referring page URL, referring page title. I see a lot of duplicates already, so that's worth paying attention to. There's the link URL. That's the stuff we're looking for. There is the text uh of the link, the link anchor, the text post, the type of link, the date first seen, the date last checked, traffic to that link.

Alright, so what we're gonna do, let's first get a sense of what don't we need here. Hi Ashley, good to see you here. We don't need the number, don't care about that. Uh let's do I'd like total backlinks, I think that's useful. Uh we don't need the domain rating uh because it's all going to Slack.com.

Um we should get the we don't need the referring domains either. Uh let's see, we want the referring page URL. What the referring page title. We want the link URL. The link anchor text.

We want let's take a quick look here at the descriptions because I want to see how we describe these things. Text pre. Oh, okay. So this is essentially the text that is just before and just after the actual link. That's kind of handy.

I like that. I don't want it in that order though, but it is kind of handy. So let's do text prext. And first scene traffic. So those fields I think are a good selection.

And we want to weed this down to just uh one unique Slack instance. So let's do a distinct on link URL keep all that. And how many slacks do we end up with? 79,381 distinct ones. Okay, now there's still a lot of noise in here, still a lot of junk.

I think what we want to do is extract out just the domain from the URL and get rid of all of the other stuff. So we're gonna need domain equals URL domain link URL. Let's see what this looks like. Oh, helps to run the function. Okay, we should have 10 variables.

Let's see if we got domains. And yes, good. That looks great. Now let's do a distinct on that. There.

So if we rerun that now we're down to 19,000 of these things. And that looks really good there. Jam stack, stuff like that. So now what if we wanted to see the most popular slacks? Arrange in descending order by traffic.

Let's see what we got here. We got secure set cybersecurity academy, national data service, software engineering, get a GitHub project, baking bad. So that's awful. Oh dear. So those are the page titles themselves.

So the GitHub idea portal, the text post. Okay. Planet Football. Envoy proxy HubSpot Slack, of course. Makes a lot of sense.

Traffic does not look right. That does not look very well. That oh, you know what it is? That is not numeric. Now I'll try.

Still does not look correct. Yes, it type traffic is typing as a character. Alright. We're going to force it to behave. Much better.

Okay. So feedly, most popular Slack, besides Slack itself, of course. Soylent, that's Slack.com. Uh some API dev stuff, the BabelJS. So this is now we've now got a good sense of you know just how popular some of these Slack instances are.

Gosh, look at how look at the dates there. Alright, let's do let's turn that fr date field to something a little bit more useful. Uh mutate. First scene goals. Uh first scene is coming in as a character, so we want this to be as date.

YMD H I Switch there. HMS D. There we go. Okay, so our our dates now are a little easier to see. Very few of the top slacks are more than a year old.

Look at that. There's I yeah, there's there's a couple there, like twenty sixteen, but a lot of them are twenty twenty, twenty nineteen. So you have that were first the links were first seen to them. Okay, next. Um things that might be interesting to know would be uh what is what are the the main slacks about?

So if we mutate a new field and call it uh full text equals paste uh let's do a paste zero on that one. We'll glue together the text pre link anchor text post. Ooh, you know what? I don't want it to be that I want to have spaces in between equals space. And now if I wanted to take a look at so we've got the full text now of what each slack is.

So it's the text just before the link, uh the text after the uh of the link itself and the text after the link. So you can see there's uh a decent number of uh there are a bunch of ones that just don't have anything, that's fine. Um the ones that do you know what let's clean out those NAs because I think text pre equals G sub NA for the space text pre. This will clean up the text to make it a little easier to read. Now let's do that for the text post as well.

Okay, now let's take a look at our frame. Yeah, it didn't work. Besides NA, which is gonna be the obvious one, is there a way we could get a sense of the general topics of all these different slacks that are out there? And the answer, of course, is yes, we're gonna need to pull just the text out. Slack names.

Start with our clean Slack here. And we're going to just pull out our full text. And we are going to need the quantity library to do text analysis. Okay. Next.

I think I want to turn that into a just a corpus. Let's take a look. Our Slack names. Nineteen thousand observations. Get rid of duplicates in the naming.

That took about a three thousand of them. And now we'll turn this into some basic text. Run that and then good so we've got a hundred thousand tokens let's do equals text stat frequency and this is going to be uh it's our tokens here from our tokens and equals five works at DFM, so oops, need to make it turn to matrix. All right. Yes, chip, you should learn to RS to follow this text.

Although to be fair, the syntax is not super ugly yet. Um this is going to be trying to DFM it from our tokens. So we should get a nice DFM out of that now. And okay, so there's our frequencies. Well, that's not helpful.

Alright. And we should be cle we should clean some stuff up. Let's do. Let's get punctuation out of there. Okay.

I have 500 observations. Better. There's still a ton of stop words in here. Um, but you're starting now to see uh some of the discussions uh some of the wording in here. Members, conversation.

What if we wanted to st if if there was a particular topic area that we wanted to look at? Um could we find that? The answer there, of course, is also yes. Let's take our clean slack, the one we were working on earlier, and topic annotated. And we'll take our clean slack and let's do let's do marketing.

Marketing's a good topic. We're going to do string count of full text, which is our field, and we're going to look for the word marketing or marketer. Let's see what we come up with. So topic announced. And filter.

Alright, so we have 24 Slacks now, and we have the traffic for those, which is pretty small. Um Bitcoin, DXW. I don't know what DXW, the DXW Slack is. Let's find out. No idea.

Where does that come from? That comes from Playbook, the DXW playbook. It's an independent agency. Private Slack channels. The DX double marketing check.

Okay. So it looks like they've got just a lot of stuff on their own Slack. So what we've come up with in this uh exploration of Slack information is a couple of things. One, uh lots of popularity of the tool itself, right? We have been able to identify those uh those Slack instances uh that are out there that exist, sort them by traffic, check out the dates of when they were first seen by this SEO tool, um, and then where those links are coming from.

Now what would you use this information for? To um I'm sure Ashley was would be very interested in in seeing because I know that uh her agency top rank does a lot of influencer marketing is this is a great way to find potentially influential communities by a specific topic. We just did marketing, for example. If we did, you know, node.js, if we did low code environments, if we did uh, you know, uh productivity, no matter what we did, we could tag we can go ahead in here and tag it and then look at the data frame that results and say, okay, these are the 10, 20, 30 Slack instances where this topic is could be being discussed based on the description of the tech of the Slack and the link text around it. So we could use it for uh influential community identification, and then that would be a place where you would go and participate, listen first, uh, and see if there are influencers, individual people within that community that would be worth having a conversation with.

Um, so that's the start of how you use this. Now, there are other communities out there. Uh Discord, we mentioned uh previously, is also uh just as big. And you could do exactly the same thing with Discord. So if you are B2C, particularly in gaming and entertainment, use this exact same setup to hunt down uh the communities where you might want to play a part, where you might want to have uh a role.

I would use this for influencer identification. I would absolutely use this for finding out where to be in these otherwise invisible communities, right? There is no master Slack directory, there is no Slack search function for other communities. The only way you know about these is if somebody links to their Slack invitation or their Slack join, you know, join us on our Slack community. Uh and in doing so allows you, allows us an SEO tool to find it.

But that's a good start, right? And then once you get into one community, you will undoubtedly see people discussing other ones. Uh, this happens all the time about on Discord. You know, I could go there's a server split and some admins got into a fight, and you know, and people ended separate their separate ways. Um but in doing so, um, you can then you know network around within the community, but you've got to have a starting point, and this is the way to get the starting point.

So um thanks for everyone who joined in. Love to see uh I see Seth there as well. How you doing? Um feel free to use this technique uh in your own work um and use the SEO tool of your choice. I'm sure uh all the other major ones uh do it just uh similarly similarly.

And uh good luck finding influential communities and influential people uh on Slack out there. Thanks for watching. I'll talk to you guys soon. Take care. Want help solving your company's data analytics and digital marketing problems?

Visit trustinsights.ai today and let us know how we can help you.