E150: Making LinkedIn Ads work for you

Karin Kanamäe: you Joaquin Dominguez: welcome to another episode of B2B Marketing Futures. Today we are diving into how B2B teams are actually running paid media today, how they are balancing brand demand creation, demand capture, and how platforms like LinkedIn fit into that mix. We'll also explore the challenges around measurement, testing, and attribution. and how AI and automation are starting to reshape how campaigns are managed at scale.

But before we jump in, I'd love for our guests to introduce themselves and share a bit about their background and current focus. ⁓ you like to introduce yourself? Welcome. AJ Wilcox: Sure, yeah, I fell in love with LinkedIn ads back in, I think, 2011.

I was at a company where they became the biggest part of what we did, and I grew that to become LinkedIn's largest spending account. And the whole time I was doing that, I was looking around going, why is no one talking about this platform? It's amazing. And so just over 11 years ago, I jumped off and started to be too linked, and we're an ad agency that LinkedIn ads is all we do.

Joaquin Dominguez: Thank you so much, E.G. ⁓ James, welcome. James Dall: Hey, yeah.

yeah, I'm James. I work for TrustPilot, which is ⁓ the world's largest independent customer platform. I'm a global paid media manager for TrustPilot and a big, well, probably one of the biggest channels we run is LinkedIn ads. We from an two agencies actually, in the last ⁓ sort of months.

And yeah, we a significant focus as we have a big B2B focus at TrustPilot. Joaquin Dominguez: Awesome, thank you so much and Karen, welcome. Karin Kanamäe: Hi all, yeah, I'm Karin from Scoro. I work as a paid marketing specialist and I've had around seven years of experience in the paid marketing field across different platforms, including LinkedIn.

yeah, in Scoro I mainly focus on LinkedIn and Google and everything that comes with it. Joaquin Dominguez: That's great, thank you all for those great introductions Let's start by setting the context and understanding how each of you is approaching paid media today Are you investing more in brand, creating demand or capturing it? How does LinkedIn fit into your channel mix? This is an open question, so feel free to admit yourselves Karin Kanamäe: I'd say us in LinkedIn focus 90 % on demand gen, on thought leadership and ⁓ But LinkedIn is just one part of our whole paid marketing endeavors.

⁓ And is good for specific like outreach and warming up our prospects that we wish were to be our clients. Yeah, so we want to be on top of their vendor list when they especially come in market. So that's why we focus on LinkedIn. AJ Wilcox: say, you for our clients, they're all over the board.

We've got a bunch of clients and everyone's in a different stage of marketing. What I like to do, I like to start with more of a ⁓ demand kind of bent to all campaigns just to see will their audiences, especially even cold audiences, will they support more of a bottom of funnel kind of function? Cause if we can go directly for a demo, directly for a sale, and people will do it at a, okay price, then cool, our testing is over. But most of the time we're unhappy with those conversion rates.

We're unhappy with the quality there. And so we'll end up turning that into a two and then a three stage funnel where we kind of prime and educate and nurture those audiences until they're finally ready to take the action that we want them to take. James Dall: Yeah, I think that what the guys there resonates completely, to be honest with you. And I think that dilemma around what is capture, what is demand, you know, you could argue everything's everything.

LinkedIn plays a significant part in like what we do in that brand awareness piece. But as AJ says, you know, there is, there's opportunity there to drive leads, to drive demos. And it's about finding that sweet spot. Maybe you max it out for a little bit, you know, the whole like 5 % in market thing creates a problem too.

And then trying to balance that with your budget. and the other channels that you have to play, right? know, Google is the one that captures all the demos, gets a lot of the credit, but you've got to drive that demand and that brand from somewhere. So yeah, it's, it's a fine line to tread and it depends on lots and lots of factors, to be honest with you.

Joaquin Dominguez: And did you see LinkedIn playing a role in capturing demand as well? Or only creating that brand awareness that thought leadership that you were mentioning Karen before? Karin Kanamäe: Yeah, we do a bit of demand capture on LinkedIn as well, but our main focus is on Google in terms of demand capture. So yeah, I think LinkedIn is best for the mansion, at least for us.

James Dall: I think it depends, doesn't it? It depends what you're trying to give away. I think if you're trying to drive a demo, that's a pretty hard sell these days. Yeah.

People's time is precious and ⁓ battered over the head with a B2B demo is pretty nauseating. So I think that it depends what you're offering. If you're trying to gate a pretty flimsy white paper, then good luck with that. And then if you're passing that to your sales team, even better luck with that, say, to be honest with you, ⁓ but is value, right?

And if you catch people at the right time, you never know. So ⁓ we a mixture and Also depends on the region, brand awareness, but we've seen success with LinkedIn with driving those bottom of the funnel demos. But again, I think it's really nuanced. You have to keep testing and working out what works and what doesn't.

Joaquin Dominguez: James, you mentioned managing multiple accounts globally, so how does scale influence your strategy and priorities? James Dall: ⁓ it's tricky. I think, as I mentioned, obviously AJ runs a successful LinkedIn ads agency. We, we in-house from one agency and then a second one when I was brought into Trustpilot, we, I mentioned, we run nine accounts.

So nine accounts globally, nine different countries. And when we in-house, we restructured everything ⁓ we wanted to create a structure that worked. Not like a template. I don't think that ever works, but you need a sense of logic so that if you add to the team, then they can pick it up and understand what they're going to do.

And then each region, right. Has different budgets, different total addressable markets. So has to be nuanced to a degree, but having that sort of functioning structure where if you went to Spain or to the U S then you could kind of pick up and understand where things are running in what formats that was particularly helpful. But it is a challenge.

And I think manpower, you know, the more eyes, the better. And I'm sure AJ experiences similar problems, like when you have multiple clients, right? They have different, different challenges, different ways of working, different structures. But yeah, that's the approach that we took when we in-house from the, from the agency.

AJ Wilcox: Well, what I love about what you've done is you've broken up your accounts by geography. And the reason why I love this is if you have two separate accounts who happen to be targeting the same individual, then you actually start bidding against yourself. You're now paying more just because you're, have things in two different accounts, but when you have them broken up by geography, especially if your geography is set to permanent only, not recent or permanent, ⁓ not bidding against yourself.

James Dall: Yes. AJ Wilcox: And now you have a really clean cut way of saying, ⁓ this is just performance for Spain and we're going to give Spain exactly what it wants. James Dall: 100%. Yeah, I think that that was a very clear decision, especially around that permanent location, which is a classic, one of the classics of LinkedIn.

We love to give you a default setting that isn't actually going to help you. That's probably one of the many ones that the location targeting, but yeah, it gave us that sort of clear dividing line, as you say, AJ, because otherwise we get bleed around spend. And, ⁓ you have regional budgets ⁓ or different who want to spend separately, that becomes a problem. So yeah, absolutely.

Joaquin Dominguez: yeah that's a really helpful overview of how you're thinking about strategy let's build on that and talk about where things get difficult in practice you anticipated James that Google gets a lot of the credit, right? and what are the hardest things measure today? where does attribution feel unclear or incomplete, particularly beyond direct conversions? Karin Kanamäe: of the hardest thing to measure is such a stupid answer, but add fatigue, because you can't just look at ad frequency and make conclusions based on that.

You have to look at CTR if it has dropped over time and no nice ⁓ graphic, like graph view on LinkedIn where can that. So that's like super annoying to me. And I like I have to implement some external tools to make it easier for me. But terms of ⁓ attribution, ⁓ we a tool called Fibler, which is really nice.

We can see the influence pipeline there per campaign everything. But ⁓ I still thinking that whether the default setting that we use on Fibler, which is the pipeline will be influenced or is considered influenced after seven paper impressions and three organic impressions, whether that's actually doesn't make sense ⁓ or it be still a coincidence that ⁓ someone that company has seen our ⁓ ads then they are also in our CRM. yeah, that always ⁓ a thought the back of my mind. whether those numbers are real, should make decisions based on those.

But at least it's ⁓ better seeing ⁓ like conversion per month for my demand gen campaigns and the campaign ⁓ platform. yeah. ⁓ Joaquin Dominguez: by organic impressions you mean be anything could be an email that someone opened or you're just talking about organic on a channel like LinkedIn for example Karin Kanamäe: I'm talking organic LinkedIn, Like, you've seen our company posts on LinkedIn. Joaquin Dominguez: I'm gonna link then, okay.

Interesting. James Dall: I think this is a, sorry, I was going to say, just think that obviously attribution is like the best and worst topic for marketing, right? Because we're just clawing away at like, well, we did a thing. give us credit for it.

When we know that there is significant value in what we're doing, we're serving impressions and there is engagement. But I think, you say, Karen, ⁓ is the threshold of that? What is actual influence? ⁓ seven impressions on a Small business with 10 employees is a little bit different to seven impressions on a 10,000 employee enterprise business, right?

And finding that sweet spot is incredibly hard, I think, and is a challenge. And even impressions or number of engagements, these are great. These are signals and can paint a picture. But as we know, that journey is getting so, so much more complicated, especially with AI search and things like that and what happens offline.

I think, yeah, we're in a sort of a weird period where something like Fibla or Dream Data is wonderful and it helps paint that picture. But we're really, I feel like we're really only seeing a small part of it and making the right decisions. say Karen is very difficult. Like, you know, do you go solely based on what this data is telling you or are you going down an avenue that actually the wrong avenue?

So making those decisions is very hard and ⁓ don't have an answer for that. I think it's about ⁓ data and as much data as you can and then making calls off the back of that. But yeah, this is, think as you go back to your question, attribution is, is always going to be a challenge, I think always. AJ Wilcox: Well, I think to echo what you've both shared, attribution for LinkedIn is just inherently really hard, but there's a big blessing with it as well.

You know, the reasons why it's hard, number one is, you know, it's a more top of funnel platform. We're able to target people by who they are, but not necessarily by what they're interested in right now. So Google captures them so well, but LinkedIn's where they're going to hear about us first because that's the only reliable platform we can make sure that our exact ICP is seeing it every time they come. But then that's the other weakness is people don't tend to spend a lot of time on LinkedIn like they do on other channels.

So we don't get very many shots on goal with them. you know, Fiddler, Dream Data, Factors.ai, they're all really good at what they do. for a very first step into this, LinkedIn has a free report called the revenue attribution report.

⁓ you know, it requires you have to have one of, of two, maybe three CRMs. I hope it gets a lot more flexible in the future, but what it's doing is it's saying every time a company makes it to a sales qualified lead or a closed deal, we're going to send that company name back to LinkedIn. Fibler ⁓ a paid tool that kind of does the same thing. And, going to look at that and say, Ooh, this company, this is how many.

organic impressions, how many organic engagements we've generated in the last 90 days or 180 days. Here's how many paid clicks. Here's how many paid impressions. And so it's going to look like LinkedIn's trying to take so much credit for what it's doing.

But, you know, don't don't listen to that. Think of like, OK, this is proof that my my LinkedIn is is having a positive influence, a touch point. on these audiences who've gone on to do something. And that speaks louder than just, okay, how much should I pay for a demo on this channel?

Joaquin Dominguez: And what you do with ingesting back the data from your CRM to LinkedIn is teaching the algorithm to find more people similar to the interests and that people who close deals with you. AJ Wilcox: I think eventually that may be the case for right now, the revenue attribution report and the company's reporting, it's a signal that hopefully LinkedIn's learning from, but the one signal that they really do learn from is from their conversions API. ⁓ And that's if you set a conversions API conversion to say every time someone becomes an MQL or an SQL, send that over to LinkedIn.

And then they're actively, there are two different campaign types where you can use that conversion event to try to optimize to find more people like that. So of course this is early days. I'm sure it's just going to get better, but that's the direction LinkedIn's heading for sure. James Dall: Yeah, think that, I think that's a fantastic point.

And I think, you say, AJ, they're not like ⁓ right? This, this, these information, and if anything, it might overindex. We're at a point where some pay channels don't get in, like don't get enough credit, but then ⁓ almost likes to potentially overindex the importance of the ads that we've been running. ⁓ it's finding that sort of educated guess in the middle.

I think that point about data going back and forward is obviously really important. think LinkedIn probably is lagging a bit behind with things like Meta or whatnot. But also what I like to do is then understand as a base point, if you're going to do that, do it that way, then let's look at all the closed deals in the last 12 months and what are those companies and what are those people look like on the platform? Cause that actually might look a lot different to what you think it might look like or what you think it looks like in the CRM.

Cause obviously with LinkedIn's targeting, it can be a little bit off compared to what you might buckets a certain industry, for example. So that's a, that's a quite a nice way to reverse engineer success and then try and expand upon that depending on. how big the audience is basically. Joaquin Dominguez: Yeah, I think all this conversation is leading to something that is maybe that comes before attribution, which is targeting because in my opinion, in B2B everything starts there.

If your targeting is incorrect, whatever you do after will be biased and with a lot of probably So How do you approach targeting today? Are you doing demographic targeting or are you uploading lists of people that you want to target or you have a blended approach? Yeah, I would love to hear your thoughts on ⁓ on LinkedIn especially. Karin Kanamäe: I use both in Scoro, but I've seen that our ABM list that we directly, which is a dynamic ⁓ we sync it directly from Housework.

We don't do the manual upload. This much better pipeline ROI, at least according to our attribution settings. It has around three times higher pipeline ⁓ ROI our demographic ⁓ and job title and the job function based targeting. Maybe this just tells me that we could still do a lot more testing and drill down on our targeting, but still the ABM list is my like gold mine at the moment.

But I keep them separate. I don't blend them in one ad group. ⁓ I keep them separate. AJ Wilcox: I think that's really important to make sure that the source of your data is something high quality, because I see this all the time.

A client will give us a, an ABM list ⁓ it perform worse than the, ⁓ the ⁓ I'll it the native targeting on LinkedIn. And I'll scratch my head and go why. And then we'll have another where the ABM list is much higher quality and we see better interactions. And I think it just comes down to like, where did that list come from?

Joaquin Dominguez: Interesting. AJ Wilcox: company lists, ABM lists are a lot easier for LinkedIn to match at a really high rate than contact lists. And then depending on how you enrich contact lists, how you upload them, LinkedIn could have a 40 % match rate on the low end up to a 98 % match rate, just totally depending on how you've handled that list and how you've uploaded it. So list quality is huge too.

James Dall: Yeah, I think that's it. That comes down to that, right? If you're going to use a list and it's good quality, then I'm a huge advocate of that because what you give LinkedIn, it will work with demographic data or the native targeting. You say AJ is extremely powerful and there may be companies that didn't make your initial list that ⁓ just weren't in the criteria.

know LinkedIn has access to ⁓ that are constantly emerging, but you're relying on obviously the quality of how they've been classified in the platform, which, ⁓ know, is the revenue correct? Were they bucketed in the right industry, et cetera, which I think I mentioned earlier. So. Yeah, think starting with a really good core list of actual accounts that would work with your business, which goes back to my point about working backwards around what you've actually closed before will be the best place to start.

If you're just starting out with LinkedIn ads, or if you're looking to choose between the native targeting and the lists, I think. Joaquin Dominguez: think James, you mentioned earlier that the challenges of ⁓ for example, a large account, right? Where maybe not the same targeting seven people in a smaller account and seven people in a very large account. However, if ⁓ you and you have very good contact data, targeting a large account, if you know specifically that those are the right decision makers that you should be talking with.

Then maybe you solve that problem. James Dall: Potentially. think at the start I mentioned how we have this always on mid-market activity and then separately a little bit like Karin, we run sort of ABM enterprise activity. I you could do your approach where maybe you really know the people you want to speak to, but do you though, because ⁓ buying are ridiculously large.

I strongly doubt that the head of procurement is already in your CRM. So you probably need to think about, especially with an enterprise level, the absurd amount of decision makers that are involved in a new piece of software, et cetera. So, ⁓ mean, I'm still an advocate for job titles because they're not really job titles, they're super titles and they actually combine numerous when you pick a marketing manager. ⁓ But I think start with an account list and then work with the job titles that you think are going to be influencers.

In terms of the activity you run, that's a separate question. Do you then go crazy and break it down by the persona or do you run verticalized content? There's obviously lots of different ways you can splice it, but with the sort of Catch 22 with LinkedIn is that every time you have an idea of a new campaign, that's a new audience and that's a new format. And then you start running away with yourself and you've got a crazy structure.

So you kind of have to find again, that sweet spot based on resource and budget and really thinking about again, like who you actually want to speak to. Joaquin Dominguez: And how do you approach testing if you want to test if this audience is interacting with your audience, specifically when you have limited budget, right? How can you ensure that you get that statistical significance? Is it something hard to achieve?

AJ Wilcox: statistical significance can definitely be hard to achieve. It's kind of the metric that I can't ever tell from my gut, like have we hit it or have we not? You've got to bring a tool and you've got to calculate it. But what I try to do is ⁓ my testing, I will have multiple stages where I can find statistical significance earlier.

So for instance, If I just look at the ad level and say this ad versus this ad, I've probably got to spend, you know, $500 or more on each variant before they become statistically significant to the click through rate. And then if I, what I really care about is some sort of a conversion or, you know, something higher than a click, uh, I've got to spend even more. I've got to spend more like three or $4,000 per ad variation. But if I've run that those same two ads across multiple audiences in the account.

or there's something else in common between them, then I can combine the spend across the whole account, start comparing likes, and then I may end up being able to find statistical significance a lot earlier. So, ⁓ one of ways I approach it, but I'm really curious to hear how you guys do too. Karin Kanamäe: That's a really cool approach. I'm going to have to write it down combine.

But ⁓ first question with combining ad variants across different audiences is that maybe one ad works ⁓ good with one audience and it doesn't work with the same and it can skew things around. But yeah, I'm have to try it because I struggle with like a low or limited budget for testing. But my approach has been to just ⁓ slow it down a bit. don't test like audience's objectives in that format all simultaneously.

I just pick one and I ⁓ one thing at a time. And I think it was in your course, AJ, that where you spoke about testing that I use personally, but correct me if I'm wrong, that I let each ad variant to get at least 3000 impressions and then I can make some kind of an initial understanding or decision based on that because if my objective is website visits with 3000 impressions, I maybe only have up to 10 website visits and doesn't retell me anything if I have eight or 10, but the CTRs, for example, might still speak a bit louder at this point or the engagement rates.

yeah, I usually base my testing based on this. James Dall: Do you guys, when you run those tests, do you tick the option to run the ads like rotate the ads evenly, ⁓ do you just let the algorithm go nuts? Because I tend to see you shaking your heads, ⁓ reassuring. yeah, curious about that because obviously it's an option, right?

And usually, I mentioned, the default options tend to be not ideal, but this one feels like actually the right one. AJ Wilcox: Yeah, this is the one situation where I say LinkedIn got the right default. it used to be, well, here's the reason why the rotate evenly is problematic is because it sounds great to us as performance marketers, like rotate my ads evenly. Yeah, sounds great.

But that's not what they actually do in practice. What they do in practice is they enter each variation into the auction evenly, but they've still given the lower performing of the ads a poor relevancy score. So when you enter something with a worse relevancy score into the auction more often, you're going to lose more auctions and the ones that you do win, you're going to win at a ⁓ cost. So I call that the charge me more and show me less button.

sounds great in theory, in practice, it raises costs and then still see a huge disparity between like add A and add B, add A is still gonna have, you know, 70, 80 % of impressions, just like if you let LinkedIn optimize. a few years back, LinkedIn made the right decision, which was in sponsored messaging campaigns, that ⁓ evenly as the default. And that makes perfect sense because when paying percent of a sponsored message, they really can make sure it's like, okay, send one of A, one of B, one of A, one of B, and they'll split it.

So we that. I used to have to change that manually, but for anything in the newsfeed, I think they think they're too smart for their own good and they overthink it. James Dall: Nice. Love that.

Joaquin Dominguez: Yeah. And AJ, before you mentioned this idea of testing multiple variants of an ad across the same audience, is that something, in my understanding, you do manually, right? Or is it something that you could do with LinkedIn the platform? AJ Wilcox: Well, we're a little bit nuts.

So I don't expect anyone to do what we do just because this is, you know, what I live and breathe and all that. But every single ad variation that we launch is totally separate. LinkedIn would look at it and see separate ad IDs. But for instance, if I'm targeting decision makers in IT, I'm probably going to have a separate campaign just for IT managers, another for IT directors, another for VPs.

and then another for CIO CTOs. But I'm not gonna tell the client we need to come up with new content for all of these. I'm gonna launch, let's say the same A-B test in all four of those campaigns. And what we learn over time is like, the ⁓ executive audience, they like this kind of talk track.

They like this kind of wording, ⁓ kinds of pain points, and maybe more of like ⁓ lower management. ⁓ They like something else. Also, because we have the same A-B test running in multiple campaigns, we also get that very quick roll-up where with a single pivot table, I can just say, hey, across the whole audiences, ad A versus ad B, which one's winning? Even though there's gonna be slight variations.

We know that LinkedIn's going to ⁓ make call about which ad in a campaign it likes. ⁓ And I'll say ad set, because I LinkedIn's trying to change what they're calling it. they're gonna pick their favorite ad in the ad set. if you're running that same A-B test across multiple campaigns, it might be that LinkedIn chooses the wrong winner in one, but in the other two, three, they pick the right one.

And so you end up getting a more even, a accurate, I would say, A-B test. James Dall: I was just going to add to that. I'm a huge advocate of always making, even if it's the same creative, same messaging, make like duplicate the ad. Like I'm just not a fan of like just bringing that ad over into another campaign.

I think it makes it completely messy. And then yeah, like we sound probably as crazy as UAJ and that every single ad has its own intricate naming structure. And that should be the same for the, campaign or your ad set, because the future you will thank you when you then have to find something or you ran a test or you wanted to check which creative did well in Q1 or yeah, those sort of naming things can be just a lifesaver and especially when it comes to reporting down the line in my opinion.

Joaquin Dominguez: And something everyone is talking about is the shift toward less manual campaign management and the use of AI to do things for you a smart way and reducing your time that spend managing manually things. So how do you see that evolving? Karin Kanamäe: Well, in my personal experience, the platform-side solutions have never proved successful. I'm happy if you have had better experience, but across Meta, Google, LinkedIn, they always try to push their automated systems where you should give all your control over your creative and ⁓ aspects to them.

I have given those solutions many chances over the years, but they have never proved to be effective because I find that they usually just spend more money and ⁓ back enough results as controlling your own budgets, targeting and creative assets. usually I give them the benefit of the doubt. I try them at least once. but still end up going back to manual campaign management, unfortunately.

James Dall: Yeah, I think I agree with Karen there. think that, I mean, look, spending lots of time doing intricate budget adjustments ⁓ whatnot is obviously not a great use of human time. And that sort of automation and that sort of AI optimization is happening should happen because then it frees up your headspace to make strategic calls and make actual ⁓ really decisions. But yeah, think LinkedIn has its own sort of accelerate, which AJ might want to speak to in a minute.

But I think the problem is that as we've talked about on this call, there are so many out of the box functions in LinkedIn's ad platform that actually work against you, that when it rolls out something that's going to then effectively automate the creation of your campaigns, you've almost got a bit of a trust issue there. So yeah, I think that's probably where the friction comes. LinkedIn probably has a long way to go to earn that trust before I hand over the keys to building me a campaign from scratch to execution.

Um, which is why I still, it's good to have, you know, people like those on this call to look at the campaigns and go, what the hell did you do here at LinkedIn? But, um, yeah, maybe AJ would speak to some of those automations that, um, they've rolled out. AJ Wilcox: just want to say amen to what you've both said. As long as a platform ⁓ upon telling that we should be doing something that is actually not in our best interest, it means every tool they roll out that relies on those ⁓ data ⁓ we can't trust.

So. If there is going to be something, and I'm sure there will, if there's an AI tool that's going to help us do things better, it's gotta be third party. It has to be from someone who understands how the platform actually works and what actually gets performance. Because for the reasons you mentioned, we can't trust it.

the accelerate feature. Sure, yeah, LinkedIn can look at your landing page and just try to decide who you should be targeting and what your ads should look like and what they should say. But it's gonna be... they're going to take away a lot of options along the way.

They're going to automatically turn on the audience network. They're automatically going to be bidding by maximum delivery, which is the most expensive way to pay, you 95 % of the time. And, you know, I don't know what it looks like on LinkedIn's end, but from my end as an agency, if those were the options I ticked, I wouldn't have that client next month. They would have pulled their budget and said, sorry, LinkedIn's not for us.

So I don't know why LinkedIn insists that they should roll those out as standard features and test it on new customers, because they're going to all swear LinkedIn off and they're going to walk away and turn the account. So anyway, that's my little rant. Joaquin Dominguez: That's a very interesting perspective. let's ⁓ you go ⁓ the manual way doing things, right?

You tag every single ad with an ID, same thing with the audiences. Of course, there's a lot of data, right? It's really hard to manage something like that. How is AI impacting in that data analysis or managing the...

Karin Kanamäe: mic drop Joaquin Dominguez: The insights that you can get from all that data, I don't know, it would be great to hear your thoughts on that as well. Karin Kanamäe: Well, I personally haven't used AI for data management or like data. What's that? Oh my God.

Data analysis. But I think a tool called Samy.chames as mentioned before, could be something useful again, a third party tool like AJ mentioned that you don't have to. and import anything anywhere.

I think that's the most annoying part about data with an AI. So third-party like Xamarin could really make it easier. But ⁓ in terms creating management, they use Claude for taglines and copy creation. And they just launched Claude design last Friday where you can prompt your design workflow and even ⁓ like manual tweaks because for me at least I haven't wanted to use like nano banana etc really because you can't really manually tweak the design afterwards and I this ⁓ option least my brain thinks I do ⁓ so yeah happy to test that out AJ Wilcox: I found that AI is not very good at interpreting data, but what has been really helpful to me for is building the workflow of how to transform, how to calculate the data, and then let a program, a Python script essentially do the calculation, do the reporting for us.

A reporting platform that would have taken me months to Claude helped me build it in like two days, and now it's... what we send to 80 % of our clients every Monday for a weekly report. I think the next step is going to be how do we ⁓ have actually do some sort of an analysis based on what it finds? I don't have a whole lot of trust.

I think that still has to be ⁓ a human experienced mind now for interpreting, but ⁓ I know is just going to get better and it's probably going to a lot more fun. James Dall: Yeah, I think that, I mean, what I would love is if I'm in the LinkedIn platform, you know, and I'm in a particular region, if I could ask it a question, I would absolutely love that. Like call me the top five best performing ads from the last month for this particular job title. And it just ran the query and that would be lovely.

Whereas you say AJ and Karen, got to really, you have to work quite hard for that to pull it out, to then put it in, to then do the sense check. So I think if that was in the platform, that'd be a glorious. Yeah. A glorious moment.

Um, but yeah, and then I think I've, uh, Carrie mentioned Sammy. don't work for Sammy, but shout out Sammy, but I think that, yeah, it's using a tool like that is extremely helpful. I think it, um, it basically frees up lot of that manual work, frees up a lot of head space, to be honest with you. You effectively can build workflows that run checks all the time.

And actually going back to something that Carrie mentioned right at the beginning of the call around things like dwell time or engagement rates. Like we set up like benchmarks, right? What are acceptable benchmarks for us? And if this ad hits below that particular benchmark in the last seven days, we want to be told about that.

And that's where we can use a tool like Sammy to be told about it. So are just one use case. We've got like over 50 workflows built. It's nuts.

⁓ But just again, like trying to remove that manual labor that ⁓ takes a lot of like ⁓ human and then freeing you up to then think about bigger picture stuff, I think is important. Joaquin Dominguez: Thank you so much. think this was a great conversation, really practical and grounded in how things are actually working. What stood out to me is how much complexity still exists across measurement, testing, the platform mechanics, even as automation and AI increases, we still rely on human intervention and judgment.

At the same time, it's clear that the fundamentals, understanding your audience, aligning the channels, getting the creative right, and are even ⁓ important. Thank you so much, James, Karin, AJ, for your experience with us today. And thank you, listening. And thank you for joining for another episode of B2B Marketing Futures.

We'll see you in the next one. Thank you.