So What? What’s New with the LinkedIn Algorithm

So What? Marketing Analytics and Insights Live

airs every Thursday at 1 pm EST.

You can watch on YouTube Live. Be sure to subscribe and follow so you never miss an episode!

In this episode, Christopher Penn and John Wall explain how the LinkedIn algorithm actually decides who sees your posts. You’ll learn why reach drops off, why you can’t game the system, and what to do instead. Updating your profile headline, leaving real comments on other people’s posts, and reposting what already worked will help keep the LinkedIn algorithm working for you.

Watch the video here:

Watch this video on YouTube

Can’t see anything? Watch it on YouTube here.

In this episode you’ll learn:

What’s new in the LinkedIn algorithm since June
Why you can’t game the LinkedIn algorithm
What the current best practices are for the LinkedIn algorithm

Transcript:

What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for listening to the episode.

Christopher Penn – 00:26

Folks, this is So What? The Marketing Analytics and Insights Live Show. I’m Chris, John is here. Katie is speaking at MIMA at an event today, so she is on the road being the star of the show this week. We are talking about what’s new with the LinkedIn algorithm. For those folks who don’t know, for the last—I forget how many years now, I think we’re on the fifth edition of the LinkedIn guide, which is the unofficial LinkedIn algorithm guide—we’ve been publishing this. It’s got at least four years now of what’s new in the system, how it works, and what things you should be doing.

This quarter’s version is totally different than previous versions because we used to publish a very dense technical paper. It was 102 pages long, filled with all sorts of information derived from LinkedIn’s actual engineering teams—the words literally out of the horse’s mouth. We got some feedback from folks saying, “It’s great that what you’re publishing is grounded in reality, but we’re not going to read 102 pages of technobabble.” So this quarter, instead, we’re creating a much simpler version, which is the LinkedIn Algorithm Guide Playbook, which you can get at TrustInsights.ai/linkedinguide. There’s also a deluxe version, which is $20 US. That one incorporates some skills for your AI tools, as well as one-pagers, supplementary materials, and a whole bunch of stuff in there.

John, to start off today’s show, based on your experiences, what have you seen working and not working with LinkedIn?

John Wall – 02:26

This is a great question for me. I haven’t been able to make anything work on LinkedIn, so this is perfect for me. It’s funny because when we were doing this, I immediately jumped in. I was like, “All right, what am I even doing on LinkedIn? How is everything going?” As I dug down, I see that I have over 5,000 followers, so I have a decent follower count, and I have over 3,000 connections. In Analytics for Marketers, Katie had actually thrown out—she was like, “Hey, how many people do you know?” I take that back, I have 4,000 followers—4,099. But she just said, “Hey, how many people do you know?”

As I was thinking about it, I probably know 15% or so of my connections, like I’ve done business directly with them or know them. It’s weird for us because I will accept pretty much any followers from Marketing Over Coffee because I feel like I know those people to some degree or, more appropriately, they do know me. They know what I am working on, where I’m going, and a bunch of things about me, so I’m pretty open with that. I know other people are much more militant. But I’ve just been following the standard advice of trying to post daily, and I end up maybe posting three times a week. A lot of times I will share content from Trust Insights, that’s number one on the list.

Whenever I just find something interesting from somebody that I really feel is worth sharing, I’ll share that kind of stuff. So that’s what I’m doing. I’m constantly confused by my alerts that I get from LinkedIn. Every week I don’t understand how I can go down 30% every week in engagement, yet somehow that is what’s happening with my account, and I’m still getting likes and shares. All that to say, I’m the perfect student for this. I’m excited to see what you’ve got in this round, and it’s been interesting. I have at least followed all the previous papers. There have been so many times that it’s changed. We’ll see people saying, “Oh, this is how you do it,” and I’m like, “No, Chris’s paper last month said you have six months of info and that’s totally invalid, so get with the program.”

Christopher Penn – 04:33

That’s a good segue into a high-level overview of what the system even is, because one of the things that people get the most wrong about LinkedIn is actually the phrasing itself: “the LinkedIn algorithm.” There isn’t one—there are two major subsystems. Actually, that’s not true: there are three major subsystems, two of which are large language models. This comes from the engineering posts and from some papers that the engineering team has published for conferences like NeurIPS, ICML, ICLR, etc. The first model is built on Meta’s Llama 3, the 3-billion-parameter version. This is called the Causal Recommender Engine. What this does is it looks at words.

It looks at the words that you use in your posts, in your profile, and in your comments, and it creates what are called embeddings—essentially topic maps of who you are and what you talk about. It does that for everybody. When the LinkedIn system is assembling your feed, it’s looking across all these members, all these different words, and these topics, and saying, “Which 500 or 600 pieces of content would best align from a word perspective, from a semantic topic perspective, for John to see in his feed?” That’s step one. Step two is what’s called the Generative Recommender model. This is a much larger model. This is the ranking model that says, “Okay, based on who John has engaged with in the past, who John likes, who comments on John’s stuff, who shares, new connections—especially given all that data out of the 500 candidates that the causal recommender created—which ones and in which order should I show them to John?” So it’s two different models, and there’s a third model, the trust and safety model, which says, “Is this spam? Is this garbage?” and things like that.

As practitioners, we have three different levers: two gates, two models, and three levers. The three levers we can control: what’s in our profile—headline, description, job tasks, and things like that. All those words are part of that first model to help the model understand who we are and what we’re about. The second is the stuff that we post and how we engage with it. The third is how we engage with other people, especially people within our network. For the purposes of engagement, it’s likes, comments, and shares, but comments tend to be the most useful.

When it comes to using LinkedIn, the bare basics are you’ve got to have words that are related to the people that you care about. If your headline is “I help people enlighten their marketing” and you actually just do social media, those words don’t have as strong a semantic or topical relevance to, say, CMOs. You’re like, “Hey, I used to be the VP of social media at this company.” Second, you should be publishing content as often as you can. Creating useful content that gets engagement comes with a big asterisk. The third is you should be going out, interacting, and engaging with other people, and engaging with people who engage with you on your stuff. Those are the three levers that people can control.

John Wall – 07:58

There’s a lot of heavy lifting in some of those phrases.

Christopher Penn – 08:03

And then the third is you should be going out, interacting, and engaging with other people, and engaging with people who engage with you on your stuff. So those are the three levers that people can control.

John Wall – 08:16

That sounds good, and thankfully none of those are a surprise as far as where I’m at. But the profile one is interesting because I am in the same boat. I see all these statements, people saying crunchy, new-age weirdness like, “I build connections across and find synergies,” and you’re like, “Buddy, you work in HR, what are you talking about?” Will you basically just dive into the three here so we can see what’s going on on these three fronts?

Christopher Penn – 08:49

Fundamentally, this is the system. Where this gets challenging for a lot of folks is these are language models under the hood, so they’re probabilistic, which means that you will get inconsistent results out of LinkedIn in the same way that if you copy and paste the same prompt into ChatGPT three different times, you’ll get three different answers. They’ll be directionally similar—if you ask, “What’s a good coffee shop near me?”, you’re going to get directionally similar answers, but you’re not going to get the same answer every time. The same is true for these two systems.

You can publish the same post once a day for five days in a row, and you’re going to get different results because you have two different language models under the hood that are taking in constantly changing data. That data is part of the prompts that these tools are fed. They actually are essentially generative AI models that are being prompted by LinkedIn’s own system, which contains stuff like your post, your engagement data, and your connections. Underlying that is the LinkedIn Knowledge Graph, which is this ridiculously huge graph of everybody on LinkedIn. I think they said it was 16 petabytes across six different data centers planet-wide with 10-millisecond latency. It’s insane.

John Wall – 10:15

This has opened my mind a little bit because I hadn’t thought about it, but because of that, you are pulling the slot machine lever every time. It’s not deterministic at all—it’s the opposite of deterministic. You’re going to do this, and who really knows what the heck is going to happen today?

Christopher Penn – 10:30

That’s what makes it so difficult. One of the things that is in the paper—and in the premium version—there are analytics skills that you can install into your Claude or your ChatGPT or whatever, where you take your data out of LinkedIn, export it from your profile, and start doing analysis to figure out what is actually working. As an example, I took my LinkedIn data. It’s funny because the way we have our social media sharing system—we use a company called Agorapulse—our account manager, Kelsey, schedules posts and I schedule posts. For example, I schedule a shared post about the podcast, etc., and then I will also just get on LinkedIn, show up, and rant in posts.

It’s interesting when you put your data from LinkedIn into an AI tool to say, “Analyze this for me,” and you give it some requirements. It’s interesting to see what works. For the posts that I publish (this is a year’s worth of data on my posts), when I just show up and rant on LinkedIn, it outperforms everything else by 4x—easily 4x, almost 5x. Those other things that I’m doing, like scheduling the short videos for the Trust Insights podcast that Kelsey is scheduling on my account, do not generate near that level of engagement.

John Wall – 12:17

That’s interesting. Seeing that, my first thought would be, “Okay, well, everything scheduled by the team is like, ‘Here’s our podcast, here’s a white paper,’ just kind of the standard stuff. Whereas the native post is something that makes you angry enough to go to the keyboard.” Is that the truth?

Christopher Penn – 12:37

Right. For the share of impressions: native posts account for 90% of my impressions, 94% share of comments, and typical impressions count is 1,136. For posts scheduled by me—like the Trust Insights podcast and stuff—3% of impressions, 3% of comments, and 273 impressions is the normal count. And by the social team: 113 posts, 8% of impressions, 3% of comments. So if I wanted to maximize my LinkedIn performance metrics, I would stop scheduling things.

John Wall – 13:16

I thought of that. Do you think there’s an opportunity to make the scheduled stuff sound more like you off the cuff, or do you think it’s different content entirely?

Christopher Penn – 13:25

It’s different content.

John Wall – 13:27

Okay, so it’s a content issue.

Christopher Penn – 13:29

Yeah. Interestingly, when I look over the overall history for my account, you can see that since March, there have just been fewer impressions on a per-month basis; it’s been suppressed. I don’t know if this is seasonal or if there’s something else going on here, but remember, this is my account—I would encourage people to make sure that you do this on your own account. This point is interesting: the typical post with a link gets 38% fewer impressions than a typical post without a link. Even though there’s nothing in any of the engineering content that says to do or don’t add links to your posts, and no analysis that shows one is better than the other, for my particular data, posts without a link do better by a substantial margin.

John Wall – 14:22

38% better.

John Wall – 14:24

Jesper threw a question in here, too. He’s saying, “Your average impressions are tiny given that you are a LinkedIn Top Voice.” I can brag about that here for you—you have over 45,000 followers, but the average number of impressions is not a big number by any amount. What do you think about that?

Christopher Penn – 14:49

That does not surprise me. Even with one of the things that’s true now of almost all social networks is that the account-level stuff doesn’t matter as much as individual pieces of content. The whole idea of today’s algorithms is that any piece of content that resonates well will take off, regardless of whether your account has one follower, 1,000 followers, or 1,000,000 followers. It’s not uncommon, particularly if you look around at some of the big accounts out there: they’ll post something, and they’ll get about the same percentage of visibility for posts that don’t land. So it can be challenging to try to figure out what actually does work, and it is on a post-level basis, not an account-level basis.

John Wall – 15:41

That’s an interesting point. So then it is a matter of anybody being able to break through. If you’ve got good content, you’re going to appeal to the model with that one individual post, and it will run. So that’s actually quite a statement for you to be able to consistently stay that far ahead of the pack, because, again, that goes back to what you just said: how frequently can you publish decent content or engaging content? It’s not a matter of if you crank the knob up to 20x of what you’re currently cranking out of spam—your odds are probably going to get worse, not better, and you’re going to end up worse off.

Christopher Penn – 16:19

What’s really interesting when you take a step back and look at the big picture about why social networks behave like this is that it overcomes the cold start problem. So a new account can land pretty well, but it dramatically undercuts influencers and personalities because all these big networks are like, “No, we want the ad dollars coming to us. We don’t want brands paying you, the person—we want the money.” So we’re going to diminish the effect of the account and focus on the content. Obviously, if people enjoy the content, they stick around, and they don’t get that person-level boost. Which is, again, no surprise: big tech wants all the money.

John Wall – 17:04

Still, with all of these networks, there’s usually a point where you can crack through and pull more than the average amount of pull to make it decent. But again, social networks ruin everything—that’s just a given.

Christopher Penn – 17:18

Exactly. The other thing is you can and should use your AI tool of choice to do some topical analysis of your content to see what stuff sticks around, how your content maps to your audience, and what the different data means. Then, compared to our guide, you might have things that you need to go test. I have a lot of things that are marked as “not enough evidence.” The guide says don’t spend much time or effort on time of day. That’s generally true because it’s not taken into account. Mine says there might be something there, but there’s not enough data.

So I would have to build an actual testing plan to start testing at different times of the day in a more structured way so that I could prove or disprove whether this statement from the guide is actually true.

John Wall – 18:16

Unfortunately, again talking about tech ruining things, LinkedIn is large enough that we’ve seen plenty of cases where the guides put out there give all this generic advice, and when you do your testing like this, you find that it’s not actually true. A bunch of these things work, but PR doesn’t want that message to get out there, so they say it’s not done. But again, only trust data.

Christopher Penn – 18:45

Yeah, exactly. The big thing for me was I did not give the AI tool enough data. What I think would have been helpful—and I will do for the next iteration for myself—is also taking stuff like Google Analytics data to see for posts where I do provide a link, does anyone click on it? Because if I post a post with a link and it gets less reach, but I actually get people to leave LinkedIn and go someplace I care about, that’s a win. If I post a post with a link and nobody clicks on it, then I’m sacrificing visibility for no benefit.

John Wall – 19:27

That’s the one grain of truth in all this: you can always count on the clicks that you’ve got. That’s proof that you can put everything against.

Christopher Penn – 19:38

Exactly. So let me take a look at what else is in terms of the guide. The versions of the guides that we have, like I said, we have the two versions: we have the one-pagers and we have the full guide itself. The big thing from a language perspective is focus on the language of the people that you want to be in front of. If you want to be in front of CMOs, like in my case, I should probably not be as absurdly nerdy and technical, because your average CMO does not care about what level quantization in a model you’re using. That is not their concern. When you look through the best practices from the guide based on what LinkedIn engineering says, these are all things that hopefully should be fairly obvious.

Also, one thing that the engineering papers don’t say, but we kind of inferred, is if you’ve got a post that did well, repost it. When you repost your own stuff, it just shows up in people’s feeds again as though it was almost new. It just says, “John Wall reposted this,” and it’s John Wall’s post. If you’ve got stuff that resonates well, rather than making a net-new post, bring it back. The reason for that is the original post has that network graph data of who commented on it, who liked it, and who engaged with it. As new people engage with that original post that you’re reposting, it influences the graph of the people who have already engaged with it. So it’s your greatest hits album.

John Wall – 21:22

This idea that it has momentum—once it has momentum, it can keep rolling. Which is crazy though, because that’s as clean an example of duplicate content as you can get, but it makes sense. It’s still driving more people to click and look, and it’s keeping the traffic there, so they’re all on board with that.

Christopher Penn – 21:41

Yep. So that’s sort of what at a high level is going on with these algorithms: it’s two language models, and you’re having to deal with them. We got another question here. Jesper says, “Normally, using a hashtag results in that word being included in the post URL. Is that not worth considering for SEO, or are posts not really indexed?” I’m not sure I understand “normally using the hashtag results in that word being included in the post URL, is that not worth considering?” When LinkedIn posts show up in search results, not all posts do, and I’ve not been able to figure out a rhyme or reason as to how a search engine like DuckDuckGo or Google surfaces what posts are and are not available.

Because I’ve run into all kinds of issues where I’ve not been able to even get data from one platform to the other. I use the hashtags in posts more as a discovery feature, because in both my own data and in LinkedIn’s technical papers, it’s not something that helps or hinders. Using a certain number of hashtags doesn’t seem to matter one way or the other.

John Wall – 23:01

Right.

Christopher Penn – 23:01

When I sat in on a LinkedIn Top Voices briefing earlier this year, they said three to five in a post is fine. It’s not going to hurt anything, but it also may not help anything. So I would say, from a URL perspective, I don’t know that from an SEO perspective. First of all, I wouldn’t necessarily want to be putting link juice for LinkedIn to take as opposed to my own stuff; I’d rather spend my SEO efforts on my owned properties.

John Wall – 23:44

Yeah, right. And then we are privy to no data on that. We have no idea how the heck that works as far as what gets in there or doesn’t.

Christopher Penn – 23:53

Yeah, I know Google and other companies have licensed LinkedIn stuff. Reddit has a licensing deal specifically with Google, so I’m not sure what the terms of that deal are and what data does and does not flow through. I do know that Google itself will restrict you pretty heavily if you start trying to scrape LinkedIn data through Google, just from firsthand experience.

John Wall – 24:24

Nothing that we would perjure ourselves with, but we’ve heard from many people that this is true.

Christopher Penn – 24:33

Exactly. If I had to boil down and say what two things should you do: number one is you should make sure your profile is up to date and contains language that you want to be found for. That can go to some degree even to your older experiences to find your profile, but definitely your headline and definitely your About section need to be updated with how you want audiences to perceive you and how you want machines to semantically perceive you.

And then the second thing is, if you have the time, I would spend 5 to 10 minutes a day intelligently and very humanly commenting on other people’s stuff. Do not use an AI comment bot. Do not do it. In July, LinkedIn debuted the “Seems Like AI Slop” button. They first debuted it for the feed, and it got some use, and then they debuted it for comments, and people started smashing the slop button left and right. It turns out I saw a post from LinkedIn’s press room saying posts that have been flagged as AI slop get 40% less visibility.

John Wall – 25:56

Yeah, no doubt that they would immediately lock you down because they know you’re trying to game things.

Christopher Penn – 26:01

So do not use AI comment bots. You’re not fooling anybody. If you go to somebody’s LinkedIn profile and you look at their three most recent comments, and they’re all structurally identical—”keen insight,” a one-sentence restatement of the post, and then the usual “Curious, what’s your take on this?”—it’s like, “Oh, it’s the exact same thing over.” No, you’re just getting slapped with the spam button.

John Wall – 26:26

Yeah, again, it leaks everywhere, and there’s nothing we can do about that. But try and be human—we can at least ask.

Christopher Penn – 26:34

Yeah, but again, with LinkedIn’s corporate press room saying stuff that gets flagged as AI slop gets 40% less visibility, that’s a pretty good bat to the knee.

John Wall – 26:47

Yeah, that’s definitely—if the ethical argument isn’t enough, that’s definitely encouragement.

Christopher Penn – 26:57

Exactly. Profile and comments—those are the two things that you should do. If you have more time or LinkedIn performance is important to you, I would spend some time having your AI tools do the analysis of your LinkedIn content: doing stuff like topic modeling, doing statistical tests to test hypotheses and saying, “Hey, I think five hashtags or more is the way to go.” Have the tool ingest your spreadsheet, do the analysis, and say for your account. Because one of the things that’s really important is that—and this kind of goes back to what Jesper was saying originally—my account is different than other accounts. Everyone’s account is different. But like the Top Voices accounts, even those get different buttons and features and stuff. There’s a whole set of backend systems that are not exposed to regular accounts, and so I can’t say what works for me is going to work for somebody else.

John Wall – 28:06

Right. And then you get the secret insider newsletter, too, that I’m not going to talk about.

Christopher Penn – 28:14

But for your account, you absolutely can do the analysis and say what’s working for you, and then use your AI tools to say, “Great, here’s where we are. Our goal is to get to whatever your number is—1,000 impressions a day or 2,000 impressions a day, or whatever downstream measure you care about.” Help me build a testing plan for when, what, or how to post different content formats to see what works better. For example, one of the things that it pulled out in my own plan is it said you really underutilize things like carousels. You should probably go in and test that. Okay, I have a testing plan to do that.

Jesper says, “Disturbing quick reversal links when they roll out. Seems like ASAP they explicitly—there’s a matter of taste—meaning they would not penalize both.”

John Wall – 29:16

That was assuaging the guilt for people who didn’t want to click.

Christopher Penn – 29:22

Yeah, the thing is, though, I think—and again, neither of us works at LinkedIn—I think LinkedIn saw pretty quickly how much people liked that feature, how much people used that feature, and said there’s a “there” there in terms of people being all-in on slapping the slop button as fast as they can.

There is so much absolute zero-effort garbage.

John Wall – 29:51

And it’s kind of rare that people get the ability to do the slap back. It’s usually downvotes and things like that end up getting abused and gotten rid of. So at least for a little while, you have a golden arrow where you can punch the gopher.

Christopher Penn – 30:09

Oh, God, that’s a throwback. So yeah, in terms of what you should be doing next, it’s those testing plans to say, “Okay, let me ask your favorite AI tool: Here’s my data. Do the analysis, pick the statistical tests that best fit this data, and come up with a testing plan for me.” And then you can start doing the testing plan, and even having the AI tools take your original ideas that are yours and just reformatting them. So I would not advocate that you have machines outright manufacture the content yourself, but if you’ve got a great LinkedIn post—particularly if you’ve done a LinkedIn post that’s done well in the past—hand that to a tool like Claude, ChatGPT, or Canvas and say, “Without changing any of the wording, take the highlights and turn this into a set of carousels for me so that we can see how it performs.” Because it’s still your words. Make sure the tool doesn’t manipulate words, but change formats pretty quickly. The big thing this week that a lot of people have been trying and we tried ourselves is you can give any generative AI tool that has coding skills the ability to generate motion graphics, like pretty nice motion graphics. So you can say, “Here’s this LinkedIn post. Help me generate a motion graphics reel.”

John Wall – 31:35

Yeah, right. Spin it up and have it go. That’s just another one of the classic… I’m just like, when am I going to take the time to figure out how to mess around with carousels? It’s one of those things that just never happens. So yeah, it’ll be very interesting to see if you optimize some stuff and throw it in there: does it get significantly better performance than just average posts?

Christopher Penn – 31:55

Even taking a look at the basics of just video and video with a voiceover, just if you were to do slides and things and do that. Tom says, “I wish the hate of AI transferred uniformly to dissatisfying ChatGPT-style video visuals.” Yeah, we all know them. And you know what causes that? When people prompt the AI tools, they do not give it any design language, so it always defaults to that exact same slightly weird, slightly off-putting font and that same general style. If you give it a specific design vocabulary, it does a great job with it, but nobody does because nobody knows anything about design.

John Wall – 32:41

Well, yeah, we’ve talked about this. That bottom level has been lifted up 30%, yet it all looks the same. I’ve never seen so many amazing bake sale and tag sale posters as I have in the past few years now. You’ll have somebody selling their pile of junk on the corner, but they’ll have Harry Potter flying over their house for their tag sale on Sunday. It’s just—yeah, I can understand how people get tired of that rather quickly. That’ll be one of those trends that future generations will look back on, laugh, and feel pity towards.

Christopher Penn – 33:19

Exactly. Here, this is a 12-second example of the motion graphics stuff that we were playing with earlier this week.

John Wall – 33:37

Yeah, I saw that go live on social. I was like, “That’s fantastic.” Just because you don’t want just the boring logo to slap up there for three seconds or whatever, but having it moving a little bit to catch eyes and catch attention—yeah, that’s bang on.

Christopher Penn – 33:53

Yeah. So if you’ve got an AI coding tool that can generate the vector graphics themselves, it can then assemble them into video. So even though a tool like Claude Code does not have image generation built in, it can do the coding necessary to create motion graphics. So that’s one of the things that, again, if you’ve got some ability to take existing great content that you’ve made, add it to different formats—do the short videos, do the vertical videos even. You could even use a tool like NotebookLM. If you’ve got a set of LinkedIn posts of yours that have done really well, you can turn them into 9×16 vertical video snippets and say, “Just summarize this post extractively,” which means using the language that you wrote in the original post as part of the video so that it’s still preserving your words. And then that can go into your LinkedIn testing plan to see if that kind of video resonates with people on LinkedIn.

John Wall – 35:01

Sounds good. That’s a good way to roll that.

Christopher Penn – 35:03

So to wrap up, we have two versions of the LinkedIn guide. There is one version that is totally free: that is the Trust Insights LinkedIn guide. This is the Playbook guide. It’s only 47 pages long, it’s easy to read, and there’s no technical stuff. And then there is the deluxe version, which is $20 US. That is the full 102-page paper, plus it incorporates specific guidance from the paper for four different roles: the founder, job seeker, marketer, and sales. It also contains four sets of skills for Claude Code, Claude Cowork, ChatGPT Work, and Anti Gravity for helping you think through changes you want to make to your profile, to your engagement strategy, to your content, and your analytics for LinkedIn. And then one-pagers again for your team for how do you do things like operate on LinkedIn, measure on LinkedIn. Because we got some feedback from the previous version of the guide that it was a little dense, so we put handles on it and made it a lot easier to digest. So the deluxe version is $20 US, the free version is free. Both are available now, and we’ll be pestering people much more about them next week.

John Wall – 36:24

That sounds good. Yeah, that’ll be going out on all channels, so watch your inbox.

Christopher Penn – 36:28

Exactly. Any final thoughts about LinkedIn, John? Anything you’re going to go try?

John Wall – 36:33

Well, yeah. I mean, obviously I’m going to check out the paper because I can get the $20 one for free, so I’m totally… And then I will dig in. But overall, the big message, though, I am at least happy that this idea of creating the best content you can and putting it out there—that’s the most important part to get started with. I mean, optimizing is great, but just keeping up with the normal flow of getting good content out there will keep you on track. But yeah, the big takeaways for me: carousels—I do need to play with that and experiment with that—and then the idea of just resharing stuff that has worked well in the past. I have a ton of stuff over the past five years—stuff that I could just pull out of mothballs and reshare it again to see if that runs. So those are two that I can put to work right away.

Christopher Penn – 37:20

Exactly. All right, folks, that’s going to do it for this week’s show. Next week, I believe we’re actually all hanging out, and the week after that we’ll be in Cleveland for the Marketing AI Conference. I don’t believe there will be a live stream that week, or if there is, it’ll be impromptu and probably from the lobby of the Hilton. But that’s going to do it for this week’s live stream. Thanks for tuning in, folks, and we’ll see you on the next one. Thanks for watching today. Be sure to subscribe to our show wherever you’re watching it. For more resources and to learn more, check out the Trust Insights podcast at TrustInsights.ai/tipodcast and our weekly email newsletter at TrustInsights.ai. Got questions about what you saw in today’s episode? Join our free Analytics for Marketers Slack group at TrustInsights.ai/analyticsformarketers. See you next time.

Need help with your marketing AI and analytics?

Leave a Reply Cancel reply

Pin It on Pinterest