Multi-Touch Attribution: What’s Working, What’s Dead, and How to Fix It

[00:00] Welcome back to another episode of the conversion tracking playbook. I’m your host Brad Reading. I don’t deserve to be on this podcast, but somebody has to click, record, and manage Riverside. I’m here with two special guests. Today is all about MTA, multi-touch attribution, and where we are today in that world. So, uh, I’m going to let Jacob and Fay,

[00:19] you guys introduce yourselves, and we’re going to jump right into what’s changed. The goal for today is to give you the listener, I don’t know, 10, 20, 30 actionable takeaways that can help you in your own marketing world and as the AI is taking over in 2025. So, Jacob, I’ll kick it over to you first. Do a

[00:36] quick intro. I’m Jacob Bro. Great to be here. Thanks, Brad. Uh, Fay, great to see you. Um, I run a marketing agency called Fluency Firm. We work with a bunch of brands in the DTOC space, cross fashion, beauty and skincare, CPG, home, and and other things. I spent a lot of time in-house at brands about 15 years

[00:54] before starting Fluency Firm and this has always been a hot a hot topic. So spending a lot of budget trying to get customers to come into the funnel get them to convert and so you know making the most use of your media dollars and how do you get them to convert and so this is topic that’s always been front

[01:08] of mind and things that we’ve used. So awesome. Thanks Fay. Awesome. Thank you Jacob and good to see you guys. Uh I’m the founder of Source Medium. We are a firstparty data aggregation and uni unification platform for first-party data for e-commerce brands across Shopify, Amazon, Tik Tok shop and chargeby. Uh I had my e-commerce experience uh from the

[01:37] mattress industry actually at a company called Resident Home which was acquired for a billion dollars by Ashley Furniture Q1 of last year. So if there’s any mattress related questions, I’d also be happy to help you answer. We got a full list of them. All right. So, so the way I’m the uh if anything talks about tracking, so actually collecting

[01:59] firstparty data, that’ll be me. So that’ll be about 2% of today’s episode. Uh the FA I was saying you’re the again extractor, the merging, keying, blending, all things first party data or other data sources. And Jacob, I think I described you are like, okay, how to turn this data or insights into dollars in my bank account or incremental

[02:20] dollars in my bank account. So that’s the way I’ve generally described. So Jacob, let’s start with you just as a level set. MTA really quick. What is it? Why do people use it? And then uh yeah, we’ll go from there. Yeah. Well, so MTA, multi-touch attribution. Yeah, I think the way the way that we’re using it and

[02:40] and the way that I think brands are are using it is looking at the entire customer journey, right? So, understanding that there’s multiple touch points for a customer to come in to your to your website to your to your brand and there’s the first time that they see it is probably not the first time that they’re making a purchase and

[02:56] then they’re coming back however many days, weeks, months later and making a purchase. And so the idea behind this multi-touch attribution is uh giving the appropriate credit to the the right sources that are driving that traffic or driving that customer to come back, right? And you do that for a lot of reasons. You know, one thing is

[03:14] understanding where the customer journey is, optimizing that spend across those different channels. So whether it’s, you know, paid social, organic social, organic search, Y, email, all of the all of the sources that people are driving in. Uh, but making sure you’re giving the appropriate credit. The the other few things are really being able to optimize your creative and messaging.

[03:33] So, being able to ascribe or, you know, give credit to a specific creative and then I think making decisions based off of like for your media spend is is a really big topic. So, how do you get the most out of it? We’re all trying to be more uh efficient with our media spend, especially now with, you know, margins

[03:49] are getting a lot tighter, especially this year. So, how do you get more efficient with your media dollars and make sure you’re putting them in the right places? And then I’d say the last thing is incrementality. So how are you going to get the most when you feel like you’re locked in in a really good place? How do you start to be able to scale

[04:02] that and and know that you’re putting your dollars in the right place? Yeah, obviously 2025 has been pretty wild with tariff changes and ping ponging there. Obviously AI, the raise of AI and LLM is taking over search and we’ve all seen the charts of how LM are a source. So potentially F, you might have some aggregate data to share in that, but

[04:22] maybe I’ll tee it up for whoever wants to jump in. what what have you heard just on the on the ground from your brands, your customers? What’s what are they asking? What are the questions that they’re asking today that they weren’t asking a year ago? Uh let’s start with you, Jacob. Yeah, I I mean a big thing

[04:37] is is the media mix. So, am I putting my dollars in the right place? Are they actually getting me what we’re saying that they’re getting? The platforms. So, when I say platform, right, the platforms like Meta and Google and Tik Tok, they’re all saying NCTV, they’re all saying that, hey, we’re driving this much. Last click is obviously saying

[04:54] something totally different. We know that the truth sits somewhere in the middle. And so a lot of people are asking, do I really need to spend this money here? So in instead of spending a million dollars a month to drive that, you know, that $5 million or $6 million in revenue, should I be spending less and and will I be more efficient? And

[05:09] and I think really trying to understand that, you know, the relationship between those things. Uh the AI and search thing is is really interesting. Um we just talked to Perplexity and and Chachi BT. um or heard some you know some things around what AI is doing with search and all that and it’s really going to change the game and so it’ll be really

[05:29] interesting to see how that changes this sort of multi-touch um yeah are you getting more a lot of questions about are brands coming to you like hey what do I need to do with LLMs or how do I get in chat GBT or how does that impact paid paid yeah a lot of questions on the search side uh we’re helping a

[05:46] lot of brands out on the organic side just you can upload your product catalog now there’s some things that you can do in chatbt to upload build your product catalog. Same with complexity. So you know that organically if people are searching your organic catalog is coming up which a lot of brands if you haven’t done it yet there’s some really easy

[06:01] things to do. You could search on Google but how to catalog my you know my products for chat GBT. You could do it today. It’s you should do it as soon as possible. So we’re telling all of our brands to do that. And Fay, what about you? uh we have a direct integration elevar and source medium. We have a an

[06:16] integration behind the scenes that it’s basically the the raw event click stream from elevar gets piped into source medium and then f you guys can either visualize that on its own to compare to G4 or other attribution data sources. So same question what what are you hearing since you are in the weeds on a lot of again the reporting the analysis with

[06:36] customers? Yeah, I think there’s a few major themes. One is additional use cases beyond marketing attribution. So things like CRO, right? Because you know once you have the sort of high quality on sample data set from Elevar, you can actually do very sophisticated page value assignment exercises understanding the best entry level landing page assist right etc. And also this becomes uh a

[07:07] lot more this gives advertisers a lot more confidence when they’re doing things like landing page AB tests especially when they want to know for example well which one is best at acquisition orders that converts into a subscription right not just kind of like any order and that also goes beyond to email and SMS right because all of that

[07:27] metadata does get passed into things like UTM params so that’s one two is around trust in auditability. I think there has been sort of more concerns around u you know blackbox solutions that uh you know are changing their algorithm without notifying brands and people all of a sudden seeing their numbers go different. So what people have been wanting now is more around the

[07:55] ability to audit the numbers but also having direct access because you know one of the big things with AI is working with data in a data warehouse environment is a lot easier now. So people are a lot more capable in terms of doing their own statisticalbased MTA modeling, but what they really need is a clean, reliable, consistent data source, right? So that’s

[08:21] a really really big role that we’ve been able to help with our customers is to give them that uh you know because it’s first-party funnel data that they have already been generating. So then they can sort of implement their own business nuances and invest into more data science sort of resources internally to take that last mile and use AI to really

[08:43] kind of like customize their own modeling to what they need to do. I I have a quick story of a a brand Steo uh they’ve been customer for a long time. Pre-elevar I was at an agency and actually when they launched as a brand I think they’re going on 11 years right now. helped them design a site and

[09:01] launch a site. And they were really one of the first brands that I worked with and I’ve been ecom for a long time that uh Danny uh who you just own their their brain so their own attribution. So they would ingest data from Google Analytics for a long time and then pull in different data sources and basically get

[09:19] to that point of okay if we have a 100 orders out of those 100 orders how many can we attribute back to marketing touch points whether it was last touch, first touch, anything in between. And it was so interesting like that’s how that was one of his metrics and how they would they would analyze their their performance of their data quality

[09:39] collection is how how close and they get can they get to that 100%. So if they know they’re spending they would they’re expecting 80% of orders to come through whether meta Google whatever it might be email SMS are they actually able to attribute those and pull those back to campaigns etc. So I think what they were doing again going back five, seven,

[09:58] eight, nine years, I would think they’re probably they were very unique and I would say ahead of most brands, which is obviously they’re amazing big brand now. They have retail stores I think in six or seven different retail stores. But Fay, what I’m hearing from you is that’s becoming the norm now where brands just want to be able to validate and do that

[10:17] gut check of okay, it how accurate, how what percentage of of those orders can we attribute back to marketing dollars is that or marketing campaigns specifically in landing pages, which I do have a question on landing pages I want to come back to, but is that is that basically what I’m hearing from you is is brands want more of that today and

[10:35] DIY it. I think there’s a large distribution between the people that are doing it uh and the people that are not. You know, I think there’s the gap between more sophisticated brand and less sophisticated brand is probably increasing faster than ever right now, right? So, you know, a lot of brands continue to sort of like not adopt sort

[10:59] of like first party data ownership and rely on existing tooling and then they’ll hit some kind of a growth ceiling, right? And that’s that becomes like okay then we have to do something and I think the brands that are winning is uh you know they really understand the importance of that early on and really a lot of the things related to

[11:21] attribution health is just UTM params and destination URL auditing right and of course a lot of these vendors don’t want to tell you that because they want to tell you there’s AI in the MTA right but ultimately that’s kind of what we help our customers do like the data and the identity resolution is pretty decent, but really it’s just about the

[11:43] human part now. Like make sure you’re using consistent naming conventions and UTM and naming convention is 50% of data and like you can’t sell that for $100,000. Uh but that’s just what it is. You know, that’s a F, that’s a really interesting point about like that gap and I think I saw that when I was in house and you

[12:02] know, you’re at a bigger brand and and Brad, you obviously have some experience, but like I think I think you’re right when you’re at a brand that’s let’s just say it’s doing less than $50 million in revenue. I’m just throwing out a number, but like you you don’t have a 200 person team, right? You you don’t have a 100 person team. You’re

[12:16] you’re probably pretty lean, especially now. And so I think to have the team that can support this is becomes, you know, becomes tough. And so you’re right, there is a big gap. Um, and I think I don’t know if we’re going to talk about tech stack, but I think there’s a pretty easy tech stack that’s not that’s pretty affordable and more

[12:33] affordable than what the incumbents want you to believe, right? I think I won’t I won’t name them, but there’s so there’s, you know, a handful of players out there that have been around doing MTA for a long time, but it’s really expensive. Uh, you know, if you’re a brand to to want to pay 10 to $15,000 a month for an

[12:48] MTA solution, like you might Yeah, you might get the incremental, you know, worth out of it. you probably will, but still like that’s a tough that’s a tough pill to swallow. But I think there’s a really, you know, I think having a good partner and, you know, a couple good partners in a pretty lean tech stack, you can do some amazing

[13:06] work. That’s what we’ve been doing with with Source Medium. We use Source Medium to help out on some of our data aggregation and almost all of our brands we push to use Elevar. And like that stack right there is is all you need. that GA and you know and and Elevar is like a pretty a pretty inexpensive stack

[13:22] to be able to get up and running and do some really deep analysis. Yeah, I appreciate the shout outs. We uh I didn’t pay you for that, but yeah, it was just an easy it was an easy layup. It’s I’ll send you champagne later. Let’s go landing pages. Some cookies. Cookies, please. If if I’m let’s say I’m still a heavy GA

[13:45] user, G4 user, you you’d mentioned landing page analysis and just getting deep into okay, focusing on where where CRO efforts would go. What if a brand just says, well, can I just go to J4 and sort by landing pages by session and do filtering by conversion rate, event value, etc.? What’s what is different? Why why would they why would they not

[14:03] use G4? Not that there’s anything wrong with G4, but what’s the difference? Yeah, there’s probably two layers. I would break this down to one is on the user experience layer. You know, people just don’t know how to use G4. Yeah. Period. That’s fair. No if ends or buts about it. Uh so that’s that’s the first problem. But the other problem is, you

[14:25] know, most of the G4 installations are client side, right? So all of the cookie level uh because it’s all about how many touch points can you link to a purchase, right? The more touch points you can link to a purchase, the better the MTA output, the more trustworthy the data and that’s where you have to do server

[14:44] side identity resolution and uh you know you have to have a pretty sophisticated vendor to do this. I won’t name any names, right? But we have we have some good ones. So you know from there you can do the analysis but of course like okay maybe it’s easy to say you know it’s defaulted to last touch right so

[15:07] okay well you know my cart page is my number one performing landing page right like that’s not right what’s your what’s your top performing first touch landing page right because then what we but across different channels right what’s your top performing on meta versus app versus YouTube versus Tik Tok right that matters What about when it comes to

[15:27] actually for acquisition orders versus repeat subscription versus non-subscription? So this is where you then need to join that data to Shopify to get the right net revenue figures, your P&L figures. And then there’s the the secondary problem of like what about your assisting pages, right? So at uh resident with mattresses, you know, we had this like 10 reasons why page that

[15:50] was really effective as a sort of a midfunnel landing page. And the only way you can really give that landing page credit right is more of a uh linearbased attribution model or you know some kind of a customized version of that. So if you know G4 just you have you can obviously link the data to BigQuery and

[16:11] do that yourself. We’re actually going to try to open source some of that uh SQL queries on the G4 BigQuery side but it’s just not really easy to do right. Um, so there’s a lot of different aspects. You could in theory just look in G4, filter out any unassigned or not set sources for landing pages and and if you

[16:33] have 50% of your landing pages are attributed to a marketing channel and you look at okay, your first party clickstream data and source medium or or another tool and say actually we have 70% of landing pages are attributed to a marketing channel. That is that basically the the back of the napkin. And how can I just validate data quality

[16:52] in 30 seconds or less? That’s if you assume it captures the maximum amount of actual touch point that happened, right? So let’s say given a purchase that’s worth a hundred bucks, the user actually had, you know, 10 touch points. It’s only cap G4 for whatever reason only captures three, right? Then you have to spread that 100 bucks across

[17:15] three landing pages. But there’s like additional sort of nuances, right? You don’t want to give credit to checkout pages. You may or may not give credit to uh cart pages, right? You only really want to be giving credit to your marketing landing pages because that’s ultimately going to be connected to your spend. Like that’s the sort of like

[17:34] sensible defaults that people need that G4 is not going to offer. But also a lot of brands uh that we discovered this with many of our MTA uh sorry u all of our customers. they have thirdparty landing pages that don’t even have G4 or Elevar implemented, right? And they’re they’re sending 30% of their traffic paid to this third-party landing page

[17:59] builder that’s essentially capturing none of the touch points. So then all of a sudden the data is just not usable at all, right? It’s actually just wrong. So the quality piece is very foundational because without that you can’t really trust the insights and then on top of not knowing how to use the user experience layer then that’s why people

[18:21] are like complaining about G4 you know left and right right Jacob I’m going to kick this back over to you looking at our our list of questions here when’s when is the right time to use an MTA solution versus okay I’m I’m going to look in combination of meta, Google dashboard, my Shopify analytics. So, when’s the right time and

[18:43] and maybe there isn’t, but what are your thoughts there? Yeah, I think it goes back to some of the things we were talking about is is the team uh and your stack a little bit and and all that and I think talking about it’s it’s a little bit easier now because there’s some other tools and you could be smarter

[19:00] about it. Um I think you could do a lot without without MTA and doing your own sort of analysis. I think you know a lot of brands are not doing it like Fay was saying like they’re just not looking at it. You could do a lot right and I think we do we’ve been doing a lot of sort of

[19:13] correlation analysis where you’re looking at last click you look at Shopify data so your blended sort of total data let’s say you look at last click from from GA and G4 and then you look at the platforms and then you can do some correlation analysis and try to look at things also G4 has a couple different models right data driven

[19:29] linear and last touch and so you can use G4’s you know uh data models and then having clean UTMs right you could do a lot as a small brand or a brand that doesn’t have a tech team or a data analyst, you could do a lot. I think coming into when you start spending, you know, more more dollars and, you know,

[19:48] to put a number on it, but I think if you’re spending anywhere between, let’s just say 200k and and and plus, I I think you could start thinking like, all right, how efficient is my $200,000 that I’m spending. And I think as you get across a couple different social platforms um and and search platforms and you’re on multiple channels, that’s

[20:05] when you start to say, “All right, I need to know, am I actually incrementally are these dollars I’m spending on each platform incremental?” Right? So, if you’re just on Meta and Google and you’re spending less than $200,000, let’s say a month, you’re probably okay to do your own analysis. I think as you start adding on Tik Tok and

[20:20] Pinterest and you’re doing CTV, you start to see all these things and you’re wondering, am I actually incremental on all these other platforms that I’m spending? Uh that’s when I think you should start thinking about the partner. You should start making sure that you have the right data points, you’re it’s clean data. That’s probably what I would do. Yeah,

[20:38] in theory you could just look at the new customer report in Shopify and if you added Pinterest one month and then you cut it the next month and see how many new customers did I get from Pinterest. That was like a very like what was what was the delta? What was the change? Yeah, we look at new uh Oh yeah, we look

[20:55] at new visitors. So total sessions and new visitors and then new customers. Yeah, but it’s that same idea which is like is it actually incrementally driving more traffic that’s new? Uh that’s a very easy way and we have a lot of brands that are small that are not doing that and it’s pretty easy to implement. Yeah, sorry Fay.

[21:11] No worries. Uh my apologies. Yeah. So I think to add on to what Jacob’s saying, the way that we advise our customers is really around the complexity score almost, right? Which is, you know, how many marketing channels do you have, how many uh sales channels do you have, right? Do you have additional sources of attribution data like uh you know how

[21:35] did you hear about a survey that you rely on or landing page right so for example one of our customers um you know elements they they rely on landing page URL tracking quite a bit for attribution and a lot of people don’t right discount codes so as these complexities add up that’s where you you’re going to need a

[21:57] more complex solution but you know if you’re just starting out you have one funnel that’s working you know across meta and Google and that’s your main thing is to optimize that right right or optimize creatives on one of those two channels then like Shopify analytics will probably be more than enough look and then look at platform uh and then

[22:18] the other thing is use case driven right so how complex how many flows and campaigns are you running on claio right and how many landing pages do you have so you start to have these combinatorial complex complexities between uh marketing channel times sales channel times landing pages times email SMS flows then you really need something a lot more robust to really understand

[22:41] what’s going on what’s what what’s the answer to view through like hey we do a lot of either connected TV or podcasts or we’re active on Tik Tok what’s the view through answer Jacob I’ll let you take that yeah uh so I think F you said it the complexity uh the customer journey what the product is uh we try to

[23:02] go pretty much no view through on most of our at least on the social platforms where we’re looking at only click clickbased attribution. We think it’s the most direct it’s the most uh sort of like linear. I think the complexity thing is is interesting to think about. So like what is the customer journey? How long is that

[23:18] velocity between from a first touch to a purchase? Right? On a on a $30 AOV, you know, lip gloss or skincare product that’s $30 or, you know, $20, that’s a much quicker life cycle. But we work with a furniture brand and we work with brands that have a, you know, we work with a very high-end jewelry brand. It’s

[23:35] a $5,000 AOV. Customers not coming in and purchasing on the first touch. And so I think understanding the user journey and the customer journey and and you know what what that looks like is is important. Yeah. It’s almost a psychology shift for I’m 40. How old am I? 43 42. I don’t know. You stop counting after four. Yeah. Yeah. Yeah.

[23:56] Yeah. Uh but the the 20 year old demographic, you know, we probably think like, well, but someone’s going to go direct to their browser and and it’s going to be the, you know, autocomplete or whatever it might be, but that that’s not their behavior. They’re not using Google. They’re is that is that do you see a big difference in demographics

[24:16] within your customers where a certain age group is going to behave and you’ll see that see that through trends in your your different reporting? Is that what’s Yeah. What’s your uh we def we definitely do I think email and SMS have actually been very telling by the way of like you know a younger you know let’s say Gen Z gen alpha brand email is not

[24:36] it used to be let’s just say 20% of a brand’s revenue that was like a rough number 15 to 20% was like pretty standard for email then SMS started coming taking some of that but if you added those together maybe it was 20% together now you’re seeing email is down 10 10 or less or 8% SMS is taking more

[24:52] but also those two own channels are not are not actually driving 20% of total revenue and it’s changing. Yeah, I think the psychology shift is is changing and and you know how you have to look at that and that’s again why it’s important to understand the touch points of where people are coming in and what first and

[25:08] last look like. Yeah. And to add on to that you know I think there’s just a lot more games now in DTOC than Shopify. So you know for example we have large Tik Tok shop brands like Neurogum and Guru Nanda right they generate a lot of revenue on Tik Tok shop and they are not caring about Clavio performance right they care

[25:33] about affiliate performance they care about offers they care about Tik Tok to Shopify crossover right and then for very large Amazon brands right you can actually get to kind of like a lot more granular data like skew level, conversion rate, and things of that nature. Uh, and there’s just a lot more different types of complexities that comes with the type of data that you

[25:58] have access to. And what’s important, right, reviews is very important on Amazon that could tank your entire brand if you’re not watching it. Um, and like it’s not as important on Shopify because you could kind of like audit, you know, you can edit your you can choose to show your reviews, right, in ways that you want to. Nobody trusts Nobody trusts

[26:17] reviews anymore. Wait, are they Amazon Are they Amazon reviews real anymore either? Yeah, a lot of them probably aren’t. Yeah, I got a guy I got a guy for that. Yeah. Yeah. All right, so let’s transition into common pitfalls and best practices. Who wants to take common pitfalls? So, let’s let’s assume it’s a 20 to50 million a year brand. What are some

[26:42] common pitfalls that that you guys see? I Fay, I think you’re probably best suited to talk about sort of the data structure and that’s where I see the most the biggest mistakes. Yeah. Yeah. I think um any level of brand up to a billion dollars, I’m not kidding, is not having consistent tracking parameters, right? I always say, you know, UTMs will

[27:04] survive way past humanity. No matter how many versions of iOS we have, UTM will always survive, right? So, just making sure they tried sliding in the the Facebook click ID and G-Click ID. They just slid into a recent announcement a day or two ago that they’re going to start cutting those from all browsing. Big shocker since they’re doing a

[27:24] private browsing a year ago. Sorry, I didn’t mean to cut Didn’t mean to cut you off, but I’m just waiting for Apple to say UTMs. UTMs, I think they will not, right? Because it’s it’s it’s at a point where it’s so universal even though the U stands for urgent. Um, you know, so I think just making sure

[27:41] there’s consistent naming convention, you’re making sure that your influencers are tagging their links consistently, your affiliates are tagging their links consistently, right? Making sure that you’re tracking on the website is properly tracked. If you’re sending paid traffic, you better have your LR pixel properly implemented. Better have your G4 as a fallback, right? making sure you don’t have session breakage and you you

[28:06] know you can’t manage what you’re not monitoring right so like that will probably be the the main thing that I’ll always say to every single brand because pretty much every single brand can improve on this front and then from there they can use much simpler infrastructure right that that’s focused on good data quality but actually you would get higher signal quality and

[28:30] higher match rate across the board I I got to reply for the biggest mistake uh is yeah do it the floor is yours. pretty simple is a brand that’s only looking at last click I think is a mistake and and we have a lot of clients that come to us and they’re like we’re only looking at last click and that’s

[28:45] how we gauge the effectiveness of our marketing and that’s I think when we hear that and see that uh we’re like this is going to be a tough because we know that the model all the things that we’re talking about we know it’s changed and again these are you don’t have to have a $20,000 a month tech stack to to

[29:03] not look at last click only but that that I I think is the biggest thing uh probably stifling a brand’s growth because how do you answer the question why is 30% of my revenue from branded Google branded search declining but like why yeah why is paid social you know a 0.9 or a 1x return I shouldn’t spend any

[29:27] money there YouTube is perfect example for this you know like YouTube is the second largest search engine in the world and I haven’t seen seen many brands that genuinely have cracked uh understanding YouTube performance, right? And it’s it’s chronically under reportported as opposed to some channels that are chronically over reported. So really having uh you know and then that’s one

[29:52] thing and then the other thing is understanding the role of a deterministic model versus a probabilistic model, right? And having a deterministic model uh next to a probabilistic model to make sure that you know there’s that fallback and there’s that check and balance is super important because you know if you’re only trusting a probabilistic model algorithmic model uh and when that model

[30:17] changes and maybe that ch is not conformant with the way that your brand works and you continue to optimize towards that right that could be very very dangerous. Yeah. Um, let’s let’s wrap up with some AI, future AI. We for you listening, I don’t know how deep you are into the world of AI outside of just either despising it or all in and using

[30:41] it, you know, in all aspects of your personal professional life, but we we pretty much already have the ability to have our own digital assistants. They’re clones of ourselves. Chat GBT launched chat GBT record last week which can just basically sit in on meetings, listen, transcribe. So it’s like the note the note killer, notetaker, category killer in my my opinion. Uh I would imagine

[31:05] most have already most that are well ahead of the curve are starting to add them into meetings to listen and be kind of like your executive assistant that sits in on meetings. Uh Fay, we were talking a little bit before we recorded of if people are doing if we are already seeing AI disrupt that and just our

[31:23] general day-to-day and managing our business, we are not that far away from it doing the same thing when it comes to analytics and helping answer all the questions that we have never had enough human time and power and capacity to extract insights and have the mental awareness to keep every meeting and every dashboard you’ve looked at over

[31:46] the last 10 years, five years, whatever, whatever it might be to comprehend and stitch those dots or stitch whatever stitch something together. So, I I I’m excited about that future over the next year because Fay, I think we’re pretty close to that, right? Where you don’t you don’t need a dashboard. You can be you can text some you can text your

[32:05] analytics agent, whatever it might be. Yeah. And I’m going to come back to the first party data and that’s why, you know, Eloar exists. So, yeah. what what does the next six to 12 months look like in that world? Yeah, I could maybe speak to it from both the brand’s perspective and the consumer’s perspective. You know, from brand’s perspective, it’s

[32:22] more important than ever to just have your first party data aggregated, unified, metric definition, your own business level nuance together, right? Because the data analyst agent is coming, right? And we are launching something around this later this year. But you know it’s traditionally our biggest obstacle to selling source medium is like okay I get that you guys

[32:49] have the most robust data but like I don’t want to do SQL the dashboard is too complicated I need something easier right thumb dumb down and with AI then it it could be as dumbed down or as complex as anybody needs to be. So having all the data together and having the data good is kind of the only

[33:09] requirement and right you’re going to have different BI vendors that can sit on top of your data and then it’s just about evaluation scores and accuracy rates right and things of that nature when it comes to evaluating the consumption layer of your good data right provided that it’s good from a consumer perspective you know Microsoft just launched uh NL NL web protocol

[33:32] which is a MC MCP protocol that Shopify joined for every single Shopify store. Now, there’s going to be a MCP endpoint slask, right? And what shop, this is already live on every single store. So, you can basically ask it for product level information, availability, and pricing. And also, you can add to cart and purchase. Right? So, from a consumer

[33:56] level perspective, I already say just give me one recommendation for this uh, you know, fan that I’m looking for, right? And so it remembers my user level preference and shopping preference and everything. I don’t have to go and do a bunch of search queries to refine down to what I need. And then from there, you know, the your AI agent, whether that’s

[34:16] Claude or Trabt or Perplexity or whoever will directly query slash ask, right? Because right in today’s world without the MCP endpoint, it’s querying only about 20 to 30% of the time your your actual first party website. So if you go and ask a question like what’s the best hydration product rapid hydration product right it only query like two out

[34:38] of the five options that it gave directly the rest of it going to affiliates and Amazon right so like brands need control over this and and and the touch points now is no longer page A B C D E FG the touch point now is query one two three four five six and the data that’s been returned and then

[34:58] you can think about the optimization that happens can become Hey, if it’s a um agent that comes in with this level of user preference prompt, right? It’s a you know a father with a newborn whatever you can even have AI specific offers, right? So from an MTA perspective, right, you will have like this new level of MTA that’s about

[35:21] multi-querying or multi-prompting even across multiple sessions uh journey, right? So then the question becomes what are the different you know where am I getting the most amount of prompts from across perplexity chat GPT claude etc and what are the queries that are coming in and what are the user preferences that are being exposed and you can even now have query level or

[35:45] channel level conversion rate from there right so in my opinion I think chat GPT ads is a matter of time if they want to continue to offer the free version perplexity is for sure going to have that Gemini is for sure going to have that. So then there will be kind of like AI agent level rorowass right. Uh so you

[36:06] know I think the concept of landing pages and websites and things like that will still be here for the foreseeable future but the the truth is it’s still going to be multi-touch but customers may only be giving you like that one touch and the rest of the touches are essentially done by their their representative. Right. I had uh Sam from

[36:27] Blink SEO on a couple weeks ago and he he’s all all in the world of SEO LLM optimization and he was he talked specifically and that was one of his action items for listeners was you you need to focus on your product descriptions and just your metadata around products. Don’t have the generic out of the box description. Get deep in

[36:47] the use cases. uh I think we were talking about landscaping or I I forget exactly what the use case but just your categorization your collection your subolctions or subcategories is make them more about use cases because that’s likely what you’re going to see people asking they’re not going to be asking like best rake for you know whatever uh

[37:06] it’s I need I need a rake that you know won’t rust because of linear salt water or yada yada yada whatever it might be yeah but just one thing though right if I’m saying best rake and I’m just saying that to my agent, right? Chad GBT has memory. So, it’s going to know I asked about oyster. It’s going to know I asked

[37:25] about, hey, best neighborhood to buy a house in Charleston, right? So, these are this just wasn’t really possible before, right? So, it’s when that query comes into your MCP endpoint, it’s probably still going to have all of that, right? Even though the original query is now just like best rake, you know? Yeah. And that that’s where I’ll

[37:48] let Jacob you can have the last word here but just going back to your first party data collection even for years I would say to brands that weren’t advertising on Google or Meta or whatever it might be I would say still just set it up so with the day you decide that you want to advertise you’re not starting with a cold pixel or cold

[38:03] audience. I think that same analogy can apply for first-party data. Whether you’re doing whether Elvar is doing it for you or someone else or you’re diying it, but if you have your just constant stream that is your source of truth data, then you can plug and play different tools or reporting or attribution or MM in and out, but you

[38:24] still have your constant consistent data stream that if you once you start quering, like you said, the memory from a consumer’s perspective, it’s you’re we’re going to have the same memory if we are a marketing analyst and the way that I’m asking questions the way that I’m analyzing my Facebook performance or campaign performance or creative performance, it’s going to use that

[38:42] memory. So to make sure that you have that same consistent data stream, so your answers also are going to be consistent coming back. So Jacob, you got the last word here. What are your uh what are your thoughts? Yeah. Well, for uh every time I talk to Fay, by the way, I I’m more educated about what AI can do

[38:59] and and what it, you know, what it’s doing right now. So it’s it’s great and I feel behind always but uh I think I’m probably somewhere probably a little bit more advanced than most people but FA is next level and um so it’s great. I take my my sort of theory and and take on AI is probably a little bit simpler and not

[39:17] as technical but there’s already so much that people are doing with AI. I think that you said it, Brad, the the marketing sort of co-pilot, sidekick, you know, the things that AI can help you with as a marketing manager and just a marketer. It’s it’s endless. And that’s just chatbt. I mean, we’re not talking about doing all this super

[39:34] technical stuff. We’re talking about having it just be, you know, helping with prompts and doing things like that. It’s like, if you’re not doing that every day just to help you get better and learn and be smarter, you’re probably a little bit behind. But just get on there and start querying and asking questions. The next thing is the

[39:47] analysis piece. Like Fay said, if you have clean data, but it’s you don’t have to upload something some big, you know, data warehouse and database. You can upload an Excel file and say as long as you have clean columns and good headers, ask questions and have it help you do an analysis quicker than let’s say you sitting there doing it doing it out. You

[40:06] have to know some of the prompts, but it’s not that hard. Um, the next thing we talked about was AI search. And I think just understanding how that’s going to work and how consumers are going to search and use it. I think that’ll be something that we’re going to see change within the year. I mean that like you said, are there going to be

[40:20] paid ads, what will that look like? But also just organic, we talked about it earlier for any brand that hasn’t uh gone to Chatbt or Perplexity and and uh basically uploaded their product catalog, do it today. It’s it’s how brands are searching. They’re pulling that metadata. It’s really easy. But my last two things are creative. A ton of

[40:38] stuff you can do with creative for paid media there. I’m not talking about creating the p the the creative but getting ideas even just how what other things should I be doing here or what what are brands doing from a creative perspective that are working prompting the the GPTs to do that. Um we use motion uh for for all of our reporting

[40:55] and the creative side they’ve actually uh over you know started rolling out some pretty cool AI stuff that analyzes video where it’s giving some ideas of like how you can make the video better. It’s not perfect but it’s a great jumping off point. We have creative strategist and our creative strategist will look at some of the things and be

[41:10] like, “Oh, that’s interesting.” And they might use some of it. Um, but it’s pretty cool. Adobe has amazing AI features already built into the product. So, being able to remove things, so our designers are just working faster now. Um, we’re not replacing people. They’re just able to, you know, turn out a couple more pieces of creative in a

[41:26] faster time period, which is just awesome. Um, and then last thing, Shopify. I don’t know if anyone’s been using Shopify Magic or any like the Shopify AI stuff. I mean, you could basically build a a site using AI not having to do much code and that’s pretty amazing. So, I think there’s a lot of stuff to come, but those are my sort of

[41:43] points of of AI and what I’m looking at. Yeah. Awesome. Well, this is uh been an awesome episode for me. Got to learn a lot and hopefully you did too. So, Jacob Fay, where can uh folks reach you, follow you, etc. Jacob, I’ll take it. Uh, LinkedIn, jacobfluency.com if you want to email, but LinkedIn, I’m I’m there and active and and posting

[42:06] things and and having a good time. So, uh, yeah. All right, FA, I know you got a couple. You got MTA.source medium. What else? What else you got coming out? Yeah. So, we uh, so this is the first place I’m announcing it is we are launching our all MTA solution that works right on top of the L of our data

[42:24] set on G4 data set. So, very excited about that. Um, so we will have uh mta.sourcemedium.com uh launching some by the time this podcast drops. Uh, but you can find me on LinkedIn as well. Uh, I’ve been working for LinkedIn for the last year or so. Uh, and then you can also find me uh by email fasfrank ei.com.

[42:49] Awesome. And if you are an all of our customer listening, I we’ll be doing a separate webinar that’ll go through the visual and more of the nuts and bolts of the reporting with uh source medium soon. So, keep an eye out for invites for that. All right, guys. Thanks for uh thanks for joining.