The Best-of-Breed vs. Consolidation Debate Is Wrong | Vinicius Rodrigues

00:00:01 — 00:02:40

Welcome to the making sense of MarTech, where the rabbit hole goes deeper than the headlines. This is a show that pressure tests ideas, not just platforms. My guest today built a career arguing that platform consolidation isn't always the answer. He also got 100 K in martech costs through consolidation, and earlier this year inherited a stack that became dramatically more complex.

So which is it? Principled fragmentation or pragmatic consolidation with better branding? A little bit about my guess first. Vinicius Rodriguez is head of marketing, technology and operations at Mind Body, now part of playlist, the $7.5 billion parent company of Mind Body, obviously, as well as Booker ClassPass in Egypt.

He owns the global martech stack across North America, EMEA and AIPAC. Three regions where almost nothing about how customers respond to marketing automation is the same. He knows that firsthand. At a previous agency role in Brazil, he ran a WhatsApp commerce pilot, a channel. Most US martech leaders have never touched.

That generated 30 x ROI and identified 20 million in revenue potential. Now he's applying that same regional instinct at my buddy, where he built the company's first marketing AI agent from scratch, and cut six figures in marketing spend through that consolidation. He didn't theorize about localization.

He ran the experiment in his native Brazil and then took the lesson global. Welcome. Hey, Jacqueline. And, uh, yeah, really excited to be here. So, as you said before, my body, I spend a couple of years in one of the largest fashion retailers in Latin America, uh, leading CRM lifecycle marketing, martech strategy.

And, you know, I had the privilege to work on big initiatives across personalization, CDP, AI. So, honestly, I think what really motivates me is the intersection between technology and marketing strategy. Like, how can we actually use technology, the right tools data to drive measurable business outcomes.

So excited that you're here. So let's dive in. And first up we've got some rapid fire questions. What was your first martech tool. It was oracle responses you know in marketing and automation platform. And of course not considering sales CRM and ad platforms because they had some experience with those, you know, prior to that.

But yeah, I think that was the first one. It's a common starter esp for folks. Yeah. Okay. Name one channel that overperformed in Brazil and underperforms in the US or vice versa. Definitely WhatsApp in Brazil in email in the US.

00:02:42 — 00:15:38

That tracks for sure. Yeah. What is the tech stack you're working with today? Okay, honestly it's quite a big and I'll say kind of complex martech stack for a business of our side. But we use Marketo as their marketing automation platform. Salesforce is our sales CRM, so Marketo is kind of very interconnected to that.

We use Bres as our B2C kind of marketing hub or you know, we use it basically to send our lifecycle comes to our consumer database. We also use psyllium as our CDP and our tag management system. And we have some, like I would say, adjacent tools such as chili pepper we use for meeting booking espera, which is really cool, vendor we use for an AI web chatbot.

We use optimizing for a B testing. So yeah, we do have the central ones like Marketo and some adjacent ones for, you know, other specific use cases. It's always a wrangling of of tools when you have both the B2B and B2C side of the business. Yeah, that's the challenge. Yes. So right now I think there's so many buzzwords that are very overhyped.

So I'm curious, what is the term, in your opinion, that every vendor is pitching right now? That's just shouldn't be the term. You probably agree, I guess, but it's a generic or even AI because I think, oh yeah, 3 to 4 years ago, if you were browsing for a new martech platform, you know, you would just go to the website and find the best CRM, the best, uh, marketing automation platform, or the best CDP.

Two years ago, they all became the AI, CRM, the AI market automation platform, the AI CDP. And then now it's the authentic CRM, the authentic CDP, the authentic, you know, market automation platform, which is for me is exciting, but I still think it's a little overrated because it's not fully authentic, right?

At least not to the business expectations yet. Yes. Well, in my opinion, agent tech is just automation with maybe an extra bell or whistle. Yeah, an easier automation, right? Where you can just tell AI what you want and then we'll do it for you. But in theory, it's still automation. It's just trying to make it easier.

Exactly. The US versus Brazil. What is one word for the biggest cultural gap in each market and how they respond to automation? I mean, it has to be one word. I would say agility slash velocity. Ooh. Okay. All right. WhatsApp as a CRM channel, is that inevitable in North America within the next five years, or is it permanently a Latam and APAC thing?

I hope it's inevitable because it's a really relevant channel, but to be honest, I'm not very optimistic about it because it really hasn't taken off yet in the US, even though, yeah, it's one of the major channels in mostly all the other countries around the world of the world. So and it's been like that for a year.

So not very optimistic, but I would hope that it does. Yeah, it's kind of mind blowing. How long? If you have any sort of international friends or family. If you're using it, you've been using it. It's just if you're based in America and don't speak outwardly, you don't use it. It's very strange. Yeah. And it's a really good channel.

I mean, I get impressed by. I think you guys, like, use SMS a lot, you know, in different ways. But for us, use, like, WhatsApp is so different. It's it's like much better in some, some way. So yeah, it's gonna be interesting with RCS and how that re maps everything. And if it does even in other countries. Yeah that's true.

And it's funny because it's owned by meta. Right. So I wonder what strategy they have for the US since it's been like like that for so many years. Okay. Buy or build for AI tools. Which do you trust more? Can I say both? Fair buy when you can view and build where you shouldn't buy? I think that's my philosophy.

For example, our AI chatbot is a good example. We use the vendor called Spera because we just couldn't build something like it with our current capabilities, right? But on the other hand, we've built custom AI workflows where we saw it was probably the best option because we would be able to move faster, and it makes made sense to that point instead of like buying something for that specific use case.

That makes sense. Okay, last but not least, what is your hottest take? I would say it shouldn't be consolidate or I mean, the question shouldn't be consolidate or best of breed. I think it could be consolidate and best of breed. So kind of explaining a little bit, I think we should try to consolidate the foundation.

But best of breed, where there's clear incremental value and differentiation. I think that's probably my my take. I wholeheartedly agree, if only everyone thought that way. All right, let's get into it. So underneath all of our conversation is a real tension in your own career. You've built a career arguing the best of breed.

Contextual fragmentation is not a failure state. It's sometimes the answer, but you cut hundreds of thousands of dollars through consolidation, and that's no longer a fragmentation win. That's a consolidation. One, to your exact hot take point is context dependent. Best of breed or a real framework or more sophisticated way of saying you couldn't get the budget to consolidate, so you made it work.

I do believe the context of pain and best of Breed is a real framework, and I think it should be because, to be honest, consolidating saved us money. And I think as martech leaders, we should always aim for that, you know, cost savings and also, you know, management effectiveness for sure. But I still believe that fragmentation when it makes sense is good because different tools do different things.

Well. And, you know, let's be honest, there's no tool in the market that can consolidate everything that the marketing art needs to do, right? So even if you consolidate a few pieces, you would still be using separate tools for separate use cases, because there's no tool can account for all the channels or tactics that your marketing team has.

And again, just to be a little more practical, I think going back to the AI chatbot example with before signing with Spera, we actually evaluated 13 different chatbot vendors to make sure we were making the right decision. And one might ask, okay, well, but you use Marketo as your marketing and automation platform.

They do have a web chat bot or well, use Chili Piper for, you know, meeting booking. And they also launched a AI chatbot. So why didn't you go, you know, with them? I think honestly, even though like those chatbots are good for our use case, it didn't work well. So we wanted something that would enable what we ideal scenario.

Right. And that's why we thought, okay, well only Sarah could do that and we signed with them. So that's you know why I think that when there is incremental value and it makes sense, you know, Best of Breed is a good path because they outperforming. That's a specific use case that we had in mind. Agreed. But also 13 that is a big eval in RFP.

How long did it take your team to do that? I think they're kind of two categories of martech tools right. Like one is, let's say, the big martech tools that are kind of like core and foundational maps CDPs. I would probably not evaluate 13 for those. And they also take more time because there are a lot they have a lot more requirements.

I see the web chatbot as a JSON one, and that's why I think the evaluation is we could be we could move faster with it and be a little less tricky. Yeah, it was kind of quick to rule some of those out. So we started with 13. But after the first round, like we could rule let's say six out because they just didn't have what some of the basic functionality we were looking at.

So it eventually went down to let's say the main three. And then we decided to okay, maybe let's run a POC with the main one. And then it worked and we just move forward with them? Understood. Okay, well, we've been talking about consolidation and best of breed in terms of like reducing budget and making sure we're being smart about our finances.

Would your perspective change if you had an unlimited budget? I honestly, I think it would stick to the same philosophy. I would consolidate the foundation, especially by foundation I mean where data orchestration, governance and even analytics live. But I would intentionally spend on best of breed tools where they could create incremental value or deliver the specific use case that we have in mind.

Um, now, if you allow me, let's complicate a bit because budget is, I guess only. Yeah, only one part of the question is budget. But there's also like team resources and business prioritization which which is I think such as important. So for example, imagine that there's a vendor that claims it could consolidate three of your, you know, main tools so you could consolidate three into one.

I mean, it makes sense in theory, but especially if it's a big tool to make that consolidation transition would take a lot of effort in time, like at least a few months, you would probably have to dedicate a team only to that, or at least like two team members. That would have to be exclusively working on that.

And the reality is that the business doesn't want to stop doing other things for a technical work or for something that would just improve data governance. You know, they want to launch new tactics, new channels. So the question is, are we willing to stop or go slower with other growth related initiatives just to make this better data governance, you know, transition.

So yeah, so it's honestly like a business strategy and business prioritization decision. And sometimes it's good to kind of to just not have the ideal stack. As long as you were kind of delivering the results that businesses wanting because one of the martech goals is also, you know, make sure we enable the the business needs.

Right. 100%. Okay. So I want to dig into one of the channels we've kind of alluded to. So you ran a WhatsApp pilot in Brazil and you had a 30 x ROI. If you tried to port that exact playbook to the US market tomorrow, what would break first, the channel, the message or the customers expectations? The channel for sure, because it's basically WhatsApp is core to how people communicate here in Brazil.

So it's kind of people use it daily like many times a day. It's yes, the main channel we use for everything. That's why it's so relevant. Right? You have the rich is so big, right? It's just not the case for the US, right? So even if you use the same strategy, same message, you wouldn't work I guess, because people are not using it.

Right? And also the rich would be limited, right. Because less people using it, the rich would be limited in there. And so the impact would be. So yeah. Let's move from the philosophy to the actual plumbing, because you've actually built most of the things that people just talk about. And so you've built this AI agent yourself.

Rather than buying into Salesforce's Agent Force, Adobe's AI assistant, HubSpot sprees, you name it. Every platform vendor says agent AI, which we've already kind of talked about, but those all belong in that suite. So what do you think they get wrong? And at what point does a custom build become a maintenance liability versus a suite that could just be absorbed?

Yeah, that's an interesting topic because honestly, we built the agent ourselves because there wasn't an option within our stack at the moment. And we really needed that process enabled, like as soon as possible. And I think that connects to what I was saying before to allow fragmentation when you have, you know, a good business case for it or you have to move fast, right?

However, I think this is the interesting piece. You realize later that one of the platforms that the sales ops sales ops teams actually use would enable the same functionality, and they would also be able to maintain it for a similar cost. So we then decided to shift that flow to them, because then the mob team wouldn't have to spend time maintaining it, and the cost was basically the same.

So when we had the opportunity to kind of consolidate into an existing tool, we did it. So but when we couldn't have that option, we kind of just build it ourselves. And also, I think the landscape has evolved since then, and Marketo has introduced its own AI assistant, a AI agent features. But that was that only happened like two months ago.

If I'm not wrong and but so when we launched it, we you know those weren't available. Now they are. So if it was today, I think before trying to set up those workflows ourselves, I would probably try the features within our stack first. And I would only try to build something else if we kind of prove that it didn't work, or if we had, like a really good reason to.

Well, spoiler alert, I have yet to hear good feedback about it.

00:15:40 — 00:22:44

Yeah. Although I haven't tried using it as much as I should, but yeah, haven't really seen game change or anything. Same boat. Okay. How are you measuring quality of leads? And how do you know that the model isn't just getting better at filtering filtering leads that are hard to convert rather than leads that are actually unqualified?

Because those are not the same thing as much as sales would like to pretend they are. Yeah, we have different layers of qualification, right? We use the kind of traditional sales qualification criteria like marketing qualified leads, sales accepted leads, sale qualified leads and etc. basically, what we wanted to have was a better marketing qualified lead because, um, I think one of the challenges in the martech landscape here is we are a B2B, to C company.

We sell the mind body platform to business customers, but we also have the Mind Body app for consumers that want to book classes. And although we have distinct websites for H, there's still a lot of consumers that go to our B2B website and end up filling out forms saying they want a demo, which they clearly don't write.

They just want to sell something related to booking, or actually just want to book a class and don't know how to do it. One of the challenges is how can we fit there all those consumer leads, at least to kind of route them to the appropriate place and not to the sales reps, right? So that's where the agent really helped us to kind of figure out, okay, well, these are likely consumers and these are likely businesses because.

So it is easy if your prospects actually all have business email domains, which is not always our case because we do serve very small businesses or solopreneurs that use Gmail addresses Hotmail addresses. So it's a little more challenging. So it's a perfect job for AI or otherwise, you would have to have a rap kind of, you know, looking at each specific record to figure that out.

So are you relying solely on the agenda AI for this filtering, or do you also have a human in the loop? Because to your point, if you have a boutique studio, they might actually want to talk to sales. Yeah. Good point. So before actually launching it we back tested it. So I think I kind of liked the idea. So basically we launched the model but we didn't take any actions in the beginning.

So we launched the model and then just let the model kind of run and categorize those leads for certain. But for a few months and only after we had enough data, we then back tested the data with the actual Rap disposition for those same leads. And then we were able to compare like, well, who AI is actually saying it is a low quality lead against what raps actually disposition as low quality leads.

Once we saw that those predictions were really accurate, only then we decided to take actions. So that kind of helped us to make sure they were being good predictions and were doing what they intended to do. I'm going to push you again. Do you re-order this every so often, or is it a continual conversation, or is it kind of a set and forget.

Yeah we do re audit it and we'll definitely keep doing it since it's recent. But also those low quality leads like we're not gonna just block them or filter them entirely. We kind of like we just work on this, but we have like a separate prioritization queue for them. So we're not going to remove them entirely from the flow.

For those that are in that separate pool, we can some of those might still be relevant leads and we don't want to miss them. As for the consumer side of things like if, well, what if they were consumers and if they have questions? And now we were not going to be answering because we're not going to be connecting them to sales.

We do have some processes in place to if we do believe that there were consumers in have like real questions, we'll route them to the customer support team and to the appropriate team. We just don't want to take the sales team bandwidth to handle things that are not within their scope for sure. I think it's one of the hardest parts of B2B, B2C or B2C to be, depending on where you started.

When I was at Grammarly in-house and building this out, we would have students who were actually the original audience of Grammarly. It was originally just built for students, and they would unintentionally or accidentally submit for contacting sales when really they just wanted to discount or they meant to sign up.

Yeah, and it was a daily, daily struggle of trying to make sure we didn't speak to them on the B2B side of the house and instead suppress them, and then made sure that they made themselves available to the B2C lifecycle team, uh, in nurturing them in their own journey. Exactly. Yeah, it makes sense. It's similar to what we're doing here.

And also, I mean, the kind of filtering consumer is just one of the pieces of the agent that we build. But we also have some sort of ICP qualification within the agent as well. That kind of help us to with prioritization, right. To prioritize the better leads. But we're not going to take any filtering actions for those for sure.

A couple of different layers that will help us both hopefully prioritize the better leads, but also be more efficient by not focusing on leads that are not true leads like your consumers, just with support questions for example. Exactly. I think the beauty of these, this use case with agent AI is it's everything we've been doing just with some elevated features.

It's lead scoring, lead grading, ICP fit. It's all of those things in one. And then the extra filtering of okay, lets air traffic control to the correct team as per their profile and request. So that's where I think the agent piece helps, but still under the umbrella of league qualification. Exactly. And to your point, earlier, it's kind of like the foundation was already set.

Now it's the additional accouterments or the additional benefit. And then is your agent outside of your primary martech stack. And in terms of like your Salesforce instance, your Marketo instance, and then it filters back in or gets pushed back into Marketo or what is the mechanism? Yeah, it's pushed back.

So we started by pushing them back into Marketo. Now they're pushing them back in directly into Salesforce. So yeah it's basically it's a it's a process kind of like outside of the main stack. But yeah the data is kind of managed and pushed back to the to the main platforms. So yeah, kind of everything happens.

You can see everything there within the stack is just like, yeah, the AI process is happening around it. Also brought to you by our sponsors. If there's one thing that's followed me at every stop in my martech career, it's trying to get good data into the hands of marketers. That's why I'm so excited to tell you about our sponsor, High Touch, the leading composable CDP and AI decisioning platform companies like Domino's, chime, AirAsia, and PetSmart trust high touch to power their data.

And here's the kicker 90% of customers have a real use case live in production within their first week. That means you can implement a world class CDP in months rather than the usual years long headache. That's why top brands choose high touch to personalize every customer interaction at scale. See what high touch can do for you at high touch commerce.

00:22:45 — 00:45:46

And now back to the hot seat. All right. Switching gears slightly, the you reported additional conversions from your meta server side work. And the additional reported is doing a lot of work there. So. Net new conversions or conversions that were already happening but weren't being attributed. Those are completely different business implications.

And I want to know which you're claiming. This is related to setting up copy or server side conversion integrations with meta Not only like we've done that with many other channels as well, because when I joined, unfortunately we only had client side tracking, or at least for most of the events, and now we actually have probably more than 15 events being tracked, both client side and server side, and also offline events.

So this is related to us like working on the server side connection to meta and copy. And yeah, to your question kind of like a mix of both right. So two things. It's related to conversions that weren't being attributed and conversions that meta didn't have visibility that were happening because we not only are now sending conversions that are attributed to meta campaigns, we're actually seeing all conversions, all CRM conversions that happen, you know, in our business to meta.

So it can basically seize everything that's happening, like get all the conversions, then reads that not only the ones that are attributed to those specific campaigns. So it's not really incremental conversions that actually happen. It's I would say it's incremental track conversions. However, I will say that when Smarta tracks more conversions and can attribute more conversions, it does help to confirm more, right?

It kind of enhances the outgo in the feedback loop so it can improve and optimize, you know, the conversion tactics. So when you're deciding on a channel strategy for a new region, what are you looking to to drive the call. So as a data you already have local team input or purely trial and error? I would say it's a combination of all of them.

So usually starts with research. You know, make sure you understand what actually works for that region, which includes team input, you know, and feedback from locos. And from there I would say it's try and nurture. Right. Like you have to start with at least at research to narrow down to the specific things you want to test.

And then honestly, you're only going to know if it really works once you test it. And yeah, like a good example is for example, we know that G2 the software review like website is really strong in the US and across other regions as well, but primarily strong in the US for the software B2B landscape. But Trustpilot is stronger in Europe, especially UK and, you know, countries around the UK.

So this is a good example where by research and team local input, you kind of understand, okay, maybe we have to have not only like a software review tactic for G2 in the US, but, you know, a different one for Trustpilot in Europe. So we can make sure we have a, let's say, good software reviews in different regions.

I had no idea. That's a good one to know. Yeah. And that that came from sales reps feedback. So it's good. Okay. Yeah that's great. I mean I'm not a fan of G2 personally, only for the fact that it's very incentivized. But is Trustpilot incentivized in the same capacity or is it kind of set up differently? They work a little bit differently, especially in the way that they work on incentives.

So it's a different setup and different tactics. But I think the the strategy behind it is the same. Also, I will say though, that G2 reviews in Trustpilot reviews and basically any sort of software reviews are now influencing SEO and Geo a lot. So they're becoming increasingly important now in this age.

Very, very true. Okay. You've alluded to it in different ways, but I'm curious. You launched an agent AI chatbot on WhatsApp before that was really even something available to the market in 2023. What all did that entail? How did you foresee the vision for this? Tell me everything. This was in my prior role.

We didn't really have a great business case to start testing it. We kind of started with stakeholders saying, well, there is this American startup we're partnering up, partnering up with because the company actually invested in them. And we believe that we should use what they're developing. But we didn't really have clear KPIs for Business Case.

And me as the martech leader there. They said okay, well, let's figure out the business case. You know, one of the main KPIs was kind of revenue. But anyways, I think the goal there was. Okay, well, we do send WhatsApp messages. It's a really relevant channel. And just, you know, as a baseline here, like at least on a fashion retail landscape in Brazil, the open rates got up to almost 80% click rate to about like 30%.

So yeah, it's a very engaging channel. But we thought, okay, what if we have an AI agent inside of the WhatsApp. So you can start with a campaign or you can just have it as a, let's say like more like a receptive channel as well. But what if we have AI there that can kind of chat with customers and the the overall goal is can we kind of have a personal stylist in there?

So like we could actually say, well, tell me about what are the type of, you know, your kind of what outfits or what lifestyle you like. And I'm happy to suggest looks for you. So you kind of started with that idea, uh, again, early days of ChatGPT and, you know, lambs being pushed to marketing tools. But it was really cool, actually.

We saw that customers really started to engage, and they started to say, well, I'm looking for let's say, um, you know, jeans or a black t shirt. And then the I would kind of figure out like, okay, well, this is your lifestyle and you're like a couple options that you might like, you know, would actually add the images, the links, like five different images and links of different, you know, combinations of clothing and add that to the website conversation.

So that was again, that was really cool because was in the early days, it was kind of like a genetic, even though that term didn't even exist, was basically just generative AI llms. And yeah, it was honestly like a fashion stylist recommending things. So it was not only answering questions through LM, it was more like recommending things by understanding what you want and also bringing the products from our e-commerce so that customer could actually see the image and then click and make the purchase.

So yeah, that was kind of like the the whole idea of that use case. On top of that, with open tax questions and answers. You were also able to gather a lot of behavioral and interest information from your customers that we didn't have a way to do otherwise. So, for example, if you started saying, well, I actually prefer a red t shirt or I'm looking for a red dress or things like that, that was like really relevant data for us to gather that we hardly could get by just your website browsing activity or your engagement with your CM campaigns.

So one of the goals was buy the things that the customers are saying. We will probably have a much better refined segmentation as well. That's awesome. And I think truly is one of the best use cases because it makes so much sense. Yeah. But then of course, there is that inflection point where, okay, now it actually needs to be hand off to a human.

That's definitely important, especially if those are like support related things that AI can solve. So we we had to think about that hand off layer as well. For sure, there's nothing more infuriating when you've got a complex situation and you're like, I need to talk to a human. Yeah, yeah. AI just says, oh, sorry, I canceled that.

You need to connect with the human, but then it doesn't connect you with the human, just as they need to do, and doesn't give you a number or a link to connect to them. So. And also the context doesn't travel. And it's just like, yeah. Oh, this is so frustrating. This shouldn't be so hard for any consumer, especially those of us who know the inner workings of how this backend works.

You have the data. So Marketo is for the B2B side of the business. Bres is for the consumer side of the business. You've got two different platforms, two likely different data models. As a result, two different sets of integrations all inside one company and by your own framework that is the vertical appropriate fragmentation.

But from a data standpoint, do you know what's working across that split? Are you running two visibility regimes and calling it a strategy? Or how are you managing this so that both sides of the business can see what's going on, or at least know that they're not interfering with the other? Yeah, that's a good push.

So we kept them separate on purpose, because we just saw that trying to merge everything into one system, like the customer and consumer database just risked a bigger data mess than running two clean, separate platforms. So that was the decision that we took, especially because the relationship between the different, like the data hierarchy for B2B is it's much different than the data hierarchy for B2C.

And also, I inherited that setup from before I joined. And and it's working well. So we just decided that well, trying to make a big data consolidation change was not within the top priorities for the past cycles. That kind of relates to what I was saying before that. Like our martech strategy needs to be connected to the business strategy, right?

So for us, it wouldn't be worth it to spend such effort to consolidate that if it's working. And we had like other better priorities. Agreed. I think a lot of people are just looking at the next shiny thing like, oh, they'll do it better. Like, if it's not broke, don't fix it. Yeah. Unless there's like a really huge cost saving, which most of the times it's not the case.

I mean, there's usually like a good cost saving opportunity in consolidating, but sometimes it's lower than the cost of having to work through all that while going lower and stopping other initiatives. Exactly. People don't factor in always there. Sometimes they forget how much of a cost, both resource wise but budgetary wise, migrating and or implementing costs on top of day to day.

Not to mention net new initiatives like you can. For the most part, keep the business running as usual, but that doesn't exactly help the bottom line. Typically, the agent you've set up runs on OpenAI through Naden and Marketo, and it's built and tuned in English language environment. Does that same set up perform as well when your qualifying leads in Portuguese or a different language, or is there a real quality gap in LLM performance outside of English that most US based martech leaders have literally never thought about?

To be honest, I haven't really seen a real quality gap there. I think now the LMS handle other languages pretty well at this point, and the output still comes back in English for us. So honestly, it's been working well for us as is. I'll say, though, that you do need to create a more complex LLM prompt to account for specific region scenarios.

You know, for example, Arabic letter form fills for eastern, Middle Eastern. It should be fine. But like if you start to get Arabic form fills for, let's say the UK, it might not be as a good quality lead that's coming there just because, you know, our sales team in the UK won't be as good as the Middle Eastern team to support those type of leads.

So you do have to take like to kind of give the prompt some additional context, local context to, to be able to, to run properly. But it's not really like a language issue. I think it's more like some context that if the Elm just read like the plain language, it would probably not get. And that's a nuance that you also have to learn as your team is growing or changes and people move to different roles.

It's a tricky one that I don't know if any lead routing part of the business can actually sort out correctly without some human intervention in those corner cases, maybe at some point in the future. Now I want to take a step back. We've been talking through some incredible strategies and implementations you've conducted, and because our industry data is telling us something different than the story, typically than we're living with, and platform vendors keep selling consolidation.

And they're almost always us built us tested. And when you're running a global platform optimized for American channel norms, but it's deployed in places like Brazil or Southeast Asia or the Middle East. Who is being left behind, forgotten, and what are you actually losing? I would say that now the major platforms already have the capabilities that are needed across different regions.

So for example, WhatsApp connection, I think most of the major platforms already have that integration. That's sad. I still believe that the platform's roadmap is built for its biggest market, and anything outside of that norm becomes a local teams problem to solve. Right. So I mean, going back to the Nan agent, it's something that we built because the main platform that we had didn't have a feature that would enable that.

And yeah, and I think that's the case for specific business needs as well. I mean, the big platforms will hardly have all the features that local teams need because, you know, each business is so different. There's different business sizes, you know, different company sizes, etc. and I think they'll always have something that the local team will have to kind of figure out.

And that's where I think ingenuity comes in, which is probably, for me, one of the best skills that martech leaders in marketing operators should now have. Right. Instead of just sticking to that same processing platform, they have to kind of be able to figure out how to do things that that current platform currently did not support because it's not part of their main market.

Right? So agreed. And now I want to double click on that. How do you filter out candidates that will be on your team that you're hiring for, to ensure that they have that level of not just ingenuity, creativity and problem solving, but also part of a growth mindset and not fixed to what the platforms dictate.

I think it also depends on the role, right? I like to kind of have different roles where one is more general and can take off general marketing ops projects. And of course, we might need one that's more specific. For example, the marketing automation manager, like it needs to know that marketing automation platform very well.

That's sad. I think even for that, let's say marketing and automation manager, I kind of always like to look at do you have a broader marketing knowledge? Do you know other tools besides Marketo, HubSpot, Salesforce? Are you willing to learn, you know, like CDPs. Are you willing to learn how ads channels work?