top of page
Search

Fixing Broken Marketing Attribution with AI | Sayf Sharif, Three Bears Data | The Brainiac Blueprint

  • Acy Rodriguez
  • Feb 8
  • 42 min read

In this episode of The Brainiac Blueprint Podcast by Left Brain AI, we sit down with Sayf Sharif, President & COO of Three Bears Data, to untangle one of marketing’s biggest headaches: attribution in a messy, AI-powered world.

Sayf breaks down how Three Bears Data and their product Optimeasure use Bayesian regression, machine learning, and real data science to bring “enterprise-level” multi-touch attribution to mid-market companies - helping teams uncover 10–15%+ lift in their existing ad spend without hiring an expensive data science team.


We dive into why last-click and linear attribution are fundamentally misleading, why paid channels like Google and Meta still provide the richest “gold standard” data, and how most companies’ biggest barrier to AI isn’t the technology - it’s messy data, inconsistent taxonomies, and a lack of data culture.


Beyond analytics, Sayf zooms out to the bigger picture: AI as the next major “age of man,” the tension between a Star Trek-style abundance future and corporate feudalism, and why human-to-human interaction will always be the real premium experience. He also shares his journey from archaeology to analytics, his love of metacognition, and the wild dress-website insight that revealed how differently rural and urban audiences shop.


Full transcript below.


🎧 Watch or listen to The Brainiac Blueprint Podcast:

Apple Connect: https://bit.ly/3VMNtlH 


⏱ In this episode, we discuss: 

00:00 – Intro 

02:00 – From archaeology to analytics & understanding human behavior 

05:00 – What Three Bears Data & Optimeasure actually do 

08:00 – Why last-click & linear attribution are “fucking criminal” 

12:30 – Bayesian regression, AI & multi-touch attribution for the mid-market 

17:00 – Why paid media data is still the gold standard 

21:00 – Data readiness: taxonomies, silos & messy naming conventions 

26:00 – Building a real data culture 

31:00 – Using AI daily for content, analysis & faster workflows 

35:00 – AI as the biggest transformational age in human history 

40:00 – Corporate feudalism vs. Star Trek abundance 

45:00 – Why human-to-human service will always be “first class” 

49:00 – The dress-site story: designers vs. leopard print & rural insights 

54:00 – What’s next for Three Bears Data & Optimeasure 

58:00 – Metacognition, walking & becoming a better human


🔗 Sayf Sharif 


🔗 Three Bears Data 


🔗 Optimeasure


📲 Connect with Left Brain AI


📣 Subscribe & Share

 If this episode inspired you, taught you something new, or gave you a different lens on AI in Data & Analytics, share it, leave a comment, or tag us.


Let’s help more people stay brilliant.


Know someone who would make a great guest?


Episode Full Transcript:


Kyle: All right. Welcome back, everyone, to another episode of the Brainiac Blueprint, where we discuss the intersection of AI and how it impacts business and the world around us with our esteemed guests. I'm Kyle Lambert, founder of Left Brain AI and Action Hero Marketing.


In today's episode, we're going to be discussing digital analytics and reporting in the blurry world of attribution and AI. That being said, today's Brainiac is Sayf Sharif. Sayf, how are you?


Sayf: I'm good, Kyle. Good to see you.


Kyle: You like that intro?


Sayf: I think I'm always hyper. It makes me so critical about myself whenever I hear someone intro me. I don't know. It's just I get very - I don't know - very looking at myself for this kind of stuff. So it's weird. Yeah.


Kyle: Oh man. Here I am thinking you're just impressed by my news anchor kind of a voice.


Sayf: Oh, that too. Yeah. A hundred percent that. I don't want to take away. Everyone's a little, everyone's a little bit narcissistic. So my first thoughts are always going to be about like - but of course, of course.


Kyle: Well, Sayf, thank you for joining. You are founder and COO of Three Bears Data. So tell us a little bit about that. Tell us about you. What's your experience and insights like and everything?


Sayf: Yeah. So Three Bears Data, it's a new MarTech product startup that we just started under two months ago. And focusing on marketing measurement in particular - you know, if you will call attribution - and bringing it up to a level where we're actually using statistics and data science in order to do it correctly, right?


Moving away from - our product's called Optimeasure, Optimeasure.io. And it's - people call it different things, whether it's cookie-less, multi-touch attribution, cookie-less MTA, or RBA for regression-based attribution. Because we use Bayesian regression in it, which - it's the kind of thing where if you start - I just say magic sometimes, because for a lot of people, it's like, "Do you believe in math?" It's math, right? It's just math.


Kyle: But I can't wait for you to pull a rabbit out of a hat in the middle of this now.


Sayf: Right. I mean, I'll be on calls and it's like, "If you want to have a call about statistics, I'll get someone to talk to you. I don't - it works, you know, it's statistics."

But so we use AI, data science, complex math, statistics, Bayesian regression in order to be able to look at what people are doing, whether it's offline, online. Some people will call it light MMM, marketing mix, but real MMM is econometrics modeling, and it's a little bit different.


Just from experience, I'll speak from my perspective, not my partner's, but I've seen this. We've been working with this, doing this with enterprise clients. And we wanted to bring it down more into the mid-market. So a lot of times with bigger agencies who have the data scientists that can do this, they're turning away companies that have 15, 20 million in annual spend because it's not enough.

And we can do this with companies that a million annual spend and up, definitely half a million and up, depending on complexity, we can do it. So kind of bringing and bringing the price down for the market.


So bringing kind of good data science cookie-less, multi-touch attribution, which can get companies 10 to 15 percent opportunity lift from their current spend if they're not doing it and bringing that to the mid-market. So why should all the fancy enterprise companies keep all the fun stuff when we can share it with everybody?


Kyle: Share the wealth. Share the wealth. I love that. I mean, honestly, you took my next question right out of my mouth. I was going to ask what problem you're solving. And obviously, analytics is just important. That's a problem.


You got to know where your dollars are going and what it's bringing back for you. But it sounds like you're trying to help the little guy out a little bit too, so that it's not just the Amazons getting all the data and all the money.


Sayf: Yeah. And it's I have plenty. It's all human to human, right? DTC, B2B, it's all human to human. And I love my friends at Google. I've worked closely with Google for years, so I don't want to malign anyone.


But I don't think it is going to be shocking to anyone to say that if you look at your reporting suite within, say, Google Ads, that Google is not going to be able to do that. Google is going to tell you what they want to tell you to get you to spend more money on Google.


And Meta is going to tell you what they want to tell you in order to - they're not going to lie to you, but they're going to tell you this stuff in a way that - so they're not - any platform that you're using, it's like - I've compared it to trusting the casino to be your accountant or something like that. It's one of these things where you shouldn't do it.


And then even for other stuff they claim, whether it's Google Analytics and attribution in there, that's never been accurate. I don't know why anyone has ever claimed it as accurate. It is a complete farce that you would say, "Oh yes, here is your attribution." Oh really? You have everything that happened?


Yeah. And because like attribution these days, you know, I'll be - that's the story I always tell to tell this is like, just think about how you interact with the internet right now and interact with our customer journey.


I'm - I did it before last. I have to - I always had dated. We're close to Thanksgiving now, but like, I'm searching for a new cutting board for the turkey. And then I go to a couple of sites and I don't really find anything. I want to kind of like - but whatever, I go away.


And then later my partner gets a message saying, "Oh, you know," they see a sponsored ad on Instagram or something saying like, "Oh, turkey cutting boards." And then they come to me and they say, "Were you looking up cutting boards?" And I go, "Yeah, blah, blah, blah."


Next thing you know, we buy a new cutting board. We've gone through four or five different devices across two different people. There's been impressions that we haven't clicked on. The last thing we did was a paid search campaign, a branded search click at the top of the page. We went right to the place. So Google is going to say that gets all the attribution because it's the last click. That makes no sense.


Kyle: Yeah.


Sayf: So I think like we're - and then also like a lot of agencies will - I mean, no offense to agencies. I've worked for agencies my whole life, but agencies - I've seen them. I've literally seen agencies that I've worked at basically turn clients away from spending money on attribution to save them money and take that money and put it into spend because they get paid based on percent of spend.

So why would you trust anyone who gets paid the more you spend or get paid just by spending? And so for us, for me, I wanted to create sort of like this fiduciarily responsible third party where we're not a platform. We're not doing your ad spend. We are here to just do measurement and tell you what you pay us flat.


We don't make a cut of whatever you do. If you want to save money, save it. If you want to take it and reinvest it, amazing. We can tell you the best places to reinvest it, help you with different audiences and whatnot.


But there's too much kind of these - you know, there's too many factors where trust becomes an issue. And I wanted people to say, "Hey, you can trust us because we don't make - you pay us to tell you the truth, not to make money off you."


Kyle: That's great. I mean, attribution is kind of everybody's biggest pain. So it's nice to have someone that's just coming in and be like, "Hey, I'm here to figure it out for you." And that's it.


So I want to dig in a little bit more into that. But before we do, I want to get your response to the prompt. So I think AI is - what is AI to you, Sayf?


Sayf: So this might take our entire conversation off on a different - 


Kyle: I'm excited. Let's go. 


Sayf: So I think AI - and I was trying to think of the best way to put this - but I think AI is probably the biggest transformational change in an age of man that has existed.


If we're going back to Stone Age, Bronze Age, Iron Age, Industrial Age, Information Age, there's people that are saying this is the fourth Information Age, the third being computers, the second one being in the late 19th century, the first going back to 1760 in England. I think this is bigger than all of them. This is bigger than - and it's bigger in what could be an extremely good or an extremely bad way, depending on how things go, because it's so big. And because humans are using it.


And humans are using it. And so, you know, I ran some calculations. I wanted to think about, let's say, like a blanket. You have a blanket, a standard blanket. Blanket for your bed? Maybe not your bed. You're huge. So like, I don't know what kind of California King you have, but like, a standard blanket - how long would that have taken to make in the Stone Age, like 20,000 years ago?


And the answer is hundreds of hours. And that's why people then, they wore animal skins and fur and things like that because that didn't take as much time. And when they had cloth, it was smaller pieces of woven material because it just took so much time to gather and prepare and make the string.


And then the early industrial age, 50 times faster, and then so on and so forth. And if you were to actually have one of these dark factories right now with robots growing the textiles, harvesting the textiles, making the blankets in a dark textile plant with no humans there doing it, it would be approximately three minutes of human oversight per blanket.


So we've gone from something that - this is someone's life making this, multiple people for weeks, hundreds of hours, 300 to 500 hours of manpower to make a blanket. And we're going down to minutes of someone's time just overseeing it.

And so then we get into this aspect of like, "Oh, okay, skyrocketing." There would be no scarcity. Like, well, there's no scarcity for food right now, but there's plenty of hungry people. And so my concern is the transition to this. It's what direction do we go here?


Do we go in this Star Trek - everyone has a replicator of sorts. We're all fed and we all have some sort of, not even universal basic income, but universal high income and blah, blah, blah. And it's actually wonderful. And we can do make art and music and, and explore the universe, whatever. Hey, that sounds good. I'd like that. That's my vote.


But also, or is it going to be one where it's corporate feudalism where, you know, you all work for Weyland-Yutani, and uh, uh, uh, you know, that's, we're, we're in an alien earth scenario kind of thing, even if, or Elysium. So I, I, I worry that we're going to head towards more of like this sort of feudalistic future of people who control technology, but I'm hoping for the, for the nicer version.


Kyle: Yeah, I mean, I understand your point. And I think, you know, if you think about any tool, people who have access to that tool get a level up, right? You know, and not even just specifically AI, like, you know, if you have a calculator, you're going to be able to get done math quicker, right?


So you're right. So that gap could could increase. I'm trying to stay positive, trying to say, you know, I like the thought of, you know, the people who get good at AI are going to be able to, you know, not get replaced and all that kind of stuff. But again, people don't have that luxury to learn a new tool or have access to it and some of that.


So we'll see what happens. It could be a little bit, it is a little scary to think about for sure. 


Sayf: It's going to be, whatever happens, it's going to be a tough transition.


Kyle: Yeah.


Sayf: And I - 


Kyle: It's going to be disruptive. 


Sayf: Yeah. And I'm hesitant to say everyone needs to learn how to use AI. I do think that, right? But what does that even mean? And how are people even capable of following what that means? I'm not sure.


And things are going to change even with AI so rapidly that honestly, if we get to an AGI, ASI kind of situation in five years and you've spent the whole five years trying to learn how to use AI, well, too bad - ASI is here and now you don't even have to do that anymore.


I think that what I'm telling people is we can't really see what's going on the other side. It's going to be transformative in a way that we cannot predict. So it's kind of like going to blacksmiths in in 1750 and saying 200 years from now, you're going to have to be able to build a car and be able to work on a car.


And that seems absurd because it's 200 years in the future, but that 200 years is so - we're in this exponential growth pattern. So that's actually five, right? Or one.

And so I don't know what it's going to be, but what I do know is that or at least what I truly believe is that the human interaction, human to human interaction is something that here and there will be, you know, having a reservation, right? Having my AI deal with the American Airlines AI to work out our flight issues. Right.


It's like the stuff that no one wants to do. That's just following logical patterns. Yes. But having your child speak to their therapist, right? Yeah. Having a, having a, um, having someone who consoles you about your death of your mother, like the person who is playing the guitar at the, at the beer garden, you're at the person who is giving you a selection of the flights of beer and saying, "Here's our current six things that we made," you know, and even to a degree that you had, you know, you know, who made the beer, how they made the beer.


Then people are okay with a level of automation, but I think keeping a mind on the human side and not letting that go. I think that if you're looking at something where you don't have to do any interaction with another human being, you don't have to understand the psychology of another human being, I think that job is going to be automated away.


So if you're not really good at interacting with people, it could be a problem.


Kyle: That's an interesting point. Thinking about the psychology of things, you know, like, you know, like you said, like interacting human to human for a customer service standpoint. You know, I don't want to say it's easy, but it's straightforward.


You know what I'm saying? But when you get into the complex nature of human psychology, you know, who knows? I mean, again, we can't necessarily predict if AI is going to be able to do that, but you know.


Sayf: I mean like one thing, someone, I was talking about this and, or I think I had something on LinkedIn and, and one of my former employees posted on about how they had just gone to the, they just did a flight.


They had gone to the airport and they had used the they hadn't talked to anybody they they walked in they used the the interface to you know get their ticket they put that on themselves they put it on the belt carried it away yeah and they never talked to anyone and I said and talking about automation and ai and I said the interesting thing is that if you were to fly say emirates first class you wouldn't have a touchscreen.


They might come pick up your luggage at your - some of the first class things, they'll pick the luggage up at your house. You don't even carry it to the airport, right? They have a human come to your house and get your - 


Kyle: Wait, that's real right now? That's a thing right now? 


Sayf: Yeah, yeah, yeah.


Kyle: Oh, I had no idea. Wow. 


Sayf: For like real first class service, you know, when you're $50,000 for whatever thing on certain airlines, they'll come pick up your luggage at your house. 


Kyle: Wow.


Sayf: Show up at the airport, you get off at the curb and you walk in and there's people there. Right? And they know your name. Maybe they use facial recognition scanners, but they also know here's who the first class passenger are. You have a nice little thing. They welcome you. A human welcomes you.

They have to do the perfunctory things. The kids can go ahead and like someone will get them a juice and we're going to do this. And it's very quick and easy. And of course, you're not going to take care of your bags. We're going to take care of your bags, but it's personal and it's human and it's interactive.


And so I think even today, if you think about high quality service, it is not touchscreens. High-quality service is not going to Taco Bell and hitting a touchscreen. High-quality service is sitting down at a Mexican cantina and having Luna come over and take your order.


That's what you want from service, is human service. You don't want the touchscreen.


Kyle: You're making me feel like a peasant right now. I didn't even know that these options were out there. Yeah.


Sayf: The thing is, you don't want to experience them because you start experiencing that stuff and then everything else is pain. 


Kyle: Yeah, yeah. I was like, I got to do my - I'm going to get drop 20 grand and get one of those Neo guys to do my laundry. 


Sayf: Yeah, yeah, yeah.


Kyle: So, all right. So, Sayf, I know that you are just - you know, we used to work together. I know you're a smart guy. You get a lot of stuff. Obviously you're in tune with these kinds of things.


So I'm curious just from your day to day, whether it be the business side or the personal side or just internal processes, you know, I could tell from your camera, you've got a little AI connection and everything going on. So how are you using it today? Um, you know, what's - where are you seeing benefits right now?


Obviously, ChatGPT is big. I'm sure you use that or an alternative. But I'm curious where you're using it.


Sayf: Yeah. And depending on my mood, I'll use ChatGPT. I'll use Claude. They're all different for different purposes. Sure. And they kind of have different approaches that I'm still kind of trying to wrap my head around whether i'd like the chat gpt go out and do it the cloud kind of more cyclical go out into the wild and do things and then come back and agentic and and that I would say without getting into I mean I use it on a daily basis yeah and an example of I can crank out so much content that the problem is I can put out too much content.


Whereas back in the day, it would be more of a struggle. And what I can do now is I have an AI trained on my writing samples. I can put together my thoughts relatively quickly using ai and say give me a bulleted list on this and this and bring this over here what is you know the idea and I have it kind of package up into a structure that I want and then i'll just talk through it i'll just talk about that and and then have take the transcript of what I talk feed that back into ai have that then kind of clean up my spoken word into being written word using my writing sources as that.


Then I take that and then I take another pass at that because it's never, it's never perfect. And so on and so forth and take a pass at that. And, um, sometimes rewriting all of it, but honestly having 2000 words written out and then looking at each paragraph and rewriting each paragraph is still, I think easier for me than, uh, uh maybe just writing 2 000 words on something from scratch it's easier for me to talk about it and so I mean I can do it but just it's the speed of it so like in the span of a few hours I can crank out a 20 minute video uh you know long form content piece based on it you know all these use ai to like take the video and like create all these like short video hooks edit out all the you know have ai edit out all the ums and ahs and you know the you knows and all that. And it's just so - 


Kyle: I'm leaving that "you know" in, we're not editing that out.


Sayf: Yeah, yeah, yeah. So it's tough. And then I think for our product, obviously, we're looking at how do we use AI in the product to, again, save time or do things that otherwise you just wouldn't have the time to do.


Really, I mean, what our stuff does is to find that signal in the noise that a human just doesn't have time to do the math. You're talking about a calculator. It's just having that many more calculators and being able to do that much more math.

Whereas a human will never necessarily... You know rationally and you can see it. Sometimes if you do turn on, turn off tests and just do something or geo test, well, we turned off all our ads in Chicago over the weekend and our conversions in Chicago tanked by this percent.


So we can assume that this accounts for that.

Um, and of course you've lost all that revenue from Chicago that week in order to do it. And you got to convince someone like we might come on, cost us like $50,000. Uh, you know, we can do that with math to a certain degree or change the mentality from feeling like we can prove it to, we can, we can probably predict it.


But I think the AI bringing that into there, looking for other patterns, machine, you know, it's really more machine learning. I mean, it depends on how you want to define AI, but looking for different patterns, bringing in seasonality, looking at data, trying to understand is that looking for problems in the data, data readiness is a huge problem.


Like I would say, arguably, I mean, what's the bigger problem with people adopting AI? Is their data really ready? Do they have good clean data? Or do they have a system? Do they have a system in place, a process in place in their org to actually adopt AI?


AI processes are two big problems whether they can actually orchestrate it or whether they have the data capable of it and having ai help with both those really but particularly with looking at data for for issues um and and is something that we're we're trying to leverage uh just because it is one of those problems where when we talk to people they'll frequently say things like I mean i've had people tell me it sounds too good to be true but the problems tend to be we're not ready.


Our data is not ready. And so if we can use AI to kind of get in there and help do some of that, then great.


Kyle: So let's unpack that a little more because I think using AI is, you know, it's obviously super important, but to your point, you put trash in, you're going to get trashed out.


So, you know, what is a - what's some like common mistakes that you're seeing some, some common things, or even just a best practice that anybody can kind of implement right now? You know, even if they're not ready to use AI, maybe it's just like, "Hey, put this in place. So a year down the road, you are good to go."

You know, I'm a marketer. I, you know, to your point earlier, I certainly understand how messy the consumer journey and the customer journey is nowadays. And like you said, different devices, different people, like all that kind of stuff.


So, you know, I'm curious if whether it's marketing data or just website data or whatever it is, what can someone do to just prepare and be ready to go and make sure they have good data to put into their AI?


Sayf: So, like, I do have a - I'm not going to look it up. I'm going to see what I can remember from my head. But I did put together a post on data readiness recently. And what kind of things, what are kind of some major areas that are problems? And what are some questions that you can ask about our data readiness? And what are some things you can do about it?


And there's a lot there. I would say if you are really the first part of it is having clarity on what are you doing with your data. And I think that a lot of companies don't even have at the basic level understanding of what they want data for.

And they have this mindset of like, "Let's collect data and figure it out later," as opposed to really having a good data strategy and saying, "What are the business questions that we're trying to answer? How are we going to measure that?" And there's so many problems, but a lot of it is just how people think about their data.


And before you even get into the technology side, before you even get into the collection or the engineering, and I think people, you get into these conversations in orgs and they're all about pipelines and ETLs and APIs and they don't have a single defined taxonomy or schema for what their data should be named.

Right. So many people don't think about that. They, they, they don't. And the - I want to use a different example, but we used to call it from a previous client at a previous agency, the candy gum problem, because they had a gum and sometimes it got labeled in the dimension as a candy. And sometimes it got labeled as a gum.


And just by doing that, it screwed everything up. I'm sure. Yeah. And, and so, how did that happen? Somebody, they didn't follow it. They didn't follow a taxonomy and they put up something on a site or they put something into a tag manager and here it is, or you name a campaign something. "Oh, who really cares if it's an underscore or a dash? I like dashes better." Well, great. Yeah.


You didn't follow the taxonomy and now that is not getting mapped and it's automated and no one looks at it and no one's aware. So really just having a culture about your data and knowing this is important and we use this to create value.


This is not about vanity metrics. This is not about dashboards and reports. It's understanding how do we use our data to make our business more successful? And we are going to do that in these ways. These are the questions we're going to ask.


And then we are only going to collect data about those things. We're going to have this all documented. So everyone in our company knows what this is. You get it in your onboarding training. You get trained on what your data is in your onboarding.


You have a data council that meets at least quarterly. You have people whose jobs - they are, you know, RACI, R-A-C-I. They are responsible and they are accountable for the quality of your company's data. You have different people on that data council responsible for your product data or your marketing data, your financial data.


You encourage people to talk about these things. You take this seriously and say, "You're in charge. You have this council. We have this documentation. We train our team in it. And we reward people that are answering the questions that we're asking. What is driving leads or whatever that is? Or how do we get to a better operational efficiency on our ad spend?"


whatever these are that you're trying to do, focusing on that and using it as opposed to just these siloed, don't want to share, don't want to integrate, don't want to talk, don't want to... Here's a screenshot of the dashboard because I don't want to give you anything you can interact with.


I mean, I've seen - I think it was at Sear where we had a client. I think it was. We had a client where they wanted to buy a second license for Google Analytics 360 for their company because the other part of their company that had it, they didn't really get along and they didn't want to share their data with that company.

And we actually had to go directly to Google to just basically be like, "Can we even do this? Like, how do we..." Normally you can't. So we had to figure it out because they literally didn't even want to share their GA360 account with another part of the company.


They didn't have to spend. They spent $150,000 to not share their data. 


Kyle: So that company has way more problems than data management at that point. 


Sayf: Anyway, so I think when we talk about this stuff, there's a ton there.

And it's something that I'm happy to help people with too. But there's so many things. But really, I think the core of it is having and having it be leadership that feels that way and values that and instills that and does it in a healthy way.


Kyle: It's funny. So you mentioned Sear. You're kind of giving me PTSD from the 2018 days of back when they were just starting to get into the data warehouse and Power BI and all that kind of stuff.


And all of us paid media guys had to go and adjust all of our campaign names for every single client and have like 13 consistent names in our naming convention and stuff like that. I was like, "Oh my God."


But we saw the impact from it at the end. Like when the data was coming in, it was clean. It was nice to see. So it works. It works. 


Sayf: And the more you do that and the longer you have it, you get year over year data. Compounds. It's just gold, right? So you do it right. And it's just so, it's so valuable.


Kyle: So I'm curious, let's say a client comes to you. They've done absolutely none of this. Things are super messy. You know, they have, let's say, five years of marketing data and stuff like that. But again, it's not organized.


You know, how do you use that or do you not use it at all? Do you say, "We have these assumptions and we're X percent confident" or something like that? Or, you know, what does that look like? I mean, I hate to say it's worthless, right? It has to be used for something. But I'm curious how you approach it.


Sayf: Yeah, I'll be honest, like at a certain point, the juice is not worth the squeeze if your data is that garbage. It depends on what's there. Again, like using tools, using AI, using even just a good find and replace, you can sometimes correct a lot of problems and your data that may have been...


Again, like if you had half your team using dashes and half using underscores, that's not a huge problem. And so for us, for Optimeasure, part of this is like, it's very human in the loop.


It's kind of like when you buy our product, you're not just buying the racing car, you're buying the F1 team driving the car and you're the owner, but you get humans that help you utilize this technology. And part of that is to understand and map that data because I don't if I've ever really seen anyone that has perfectly clean data.


It doesn't exist. It doesn't exist. If the instant you get it, it's going to start falling apart. So it's just a matter of time on that.


But yeah, I think a lot of the good thing for what we're doing, a lot of it is using primarily stuff from the ad platforms, data right from the ad platforms. So as long as people have been doing consistent taxonomy or if they've changed a few times, but we know how to map certain things, we can deal with it pretty effectively.


It's really only when it's super chaotic. And generally, once you get into the position that you're using two to three different ad platforms, you have two to three different channels, you have different things. Usually you have some kind of naming convention going on or multiple, but something that can be identified and cleaned up.


So, um, and I think for for everything else, it's just a matter of us working again. Cause like the human, no data is clean and we have to determine what data matters. Sometimes, um, you know, the, maybe people care about the mortgage rate.


People care about whether there was a natural disaster. People care about what the weather was like. Uh, I've had people, one wanted to know if we could pull on a daily basis, what the value of scrap metal was.


And I had to go find out, was there an API that exists somewhere that can pull me in daily? What the value of scrap metal is per pound. I, it's just like, "Okay, we can see," but it's, you have the human involved, you bring in what does it matter for people? And then we we make sure it's clean.


So you'd have to - it'd have to be pretty bad for - and I think those people are not going to be coming to us at this point. They might come to us to help us fix our data for the long run, which I'd be happy to do, you know, but, you know, I think there's, there's, we were talking about it earlier today internally about if someone wants to come buy the $200,000 Frank Lloyd Wright-style Airstream trailer, you don't turn them away because they don't own a truck.


You say... "Awesome. Let's get the paperwork ready on that $200,000 Airstream trailer. What truck do you have? You don't have one? Amazing. Let's walk over here and talk about trucks." I think if your data is not then let's talk about a truck and then we can move on to the trailer afterwards.


Kyle: Sure. Sure. So, you know, you talked about these different channels and, you know, you talked about ad platforms and some of that, you know, I'm a paid media guy. I'm a little bit biased, but, you know, I think that these ad platforms are so powerful because, you know, you know, this is how they make these channels make their money, right?


Google makes their money through ads. Meta makes their money through ads. So they empower us advertisers with more data, you know, so that we can make actionable insights and we're not just assuming and all that kind of stuff.

You know, you are an analytics guy. I'm curious if you kind of think about that the same way. Is there a specific channel that you're like, "This is the gold data, like this is our standard"?


Sayf: Paid.


Kyle: Paid? That's what I figured. Yeah. Is Google or Meta better than others? I mean, those are the big two, of course.


Sayf: I think it all depends on the audience, right? Who is your audience? Who are you targeting? What are you targeting them with? Some people, it's going to be Facebook. Some people, it's going to be Instagram. not Facebook.


I mean, it kind of goes back and forth, but some people it's going to be TikTok. Some people it's going to be YouTube, a lot of YouTube videos, like YouTube folks. Some people it's going to be TV, right? I mean, but there's, there's a lot out there, but I would say the the number one effective thing going forward has been in going forward is paid.


And, and, I say that in the sense that, so I think it's going to get more so too, just because I think that we continue to see this evolution of, um, again, it's, it's, it's this, it's Weyland-Yutani capitalism, right?


It's like, how can these companies make more money? They always need to make more money. And anything that they are giving away for free is that they're not making money. So, you know, they are Google organic, right? is with taking away deeper scrapes of like this the google searches with the ai promoted.


I mean is there organic on the front page of google searches I mean you got to be real obscure if that's going to happen you're going to get an ai response you're going to get paid responses uh and you know like you're going to go into chat gpt and but chat gpt or whatever else that are you're going to you're going to start asking questions of that where are they getting their information from it's a similar kind of thing um algorithmically i've done different tests on this but you know and i'm not you know just personal but I mean there's not anyone would say this but whether you're if you're creating content these days they want to pay people who are keeping people on platform.


Right. Right. Because they want on-platform time. Right. And so that means 


Kyle: More ads showing up.


Sayf: More ads will show up. And so then they'll pay you if you make people stick around more and they don't care.


Followers, subscribers, this stuff is meaningless these days, almost next to meaning. I mean, you can show it as maybe for getting someone to pay you money to talk about something. I was going to say, you can sell it to a brand or something like that.


But the reality is that, you know, if you wanted to organically, you know, let's - what's the most generic side hustle business, like a Etsy t-shirt store. So like you set up an Etsy t-shirt store and you have, you know, whatever your Nathan Fillion fan club, whatever.


And you like make, you do paid ads to point to your store. What's going to, you know, you can do paid ads or you can just kind of post you organically on Instagram or TikTok? What is going to actually get you results? I think we know that's going to be paid.


Now you could hit on a specific video and you get 3.1 million views and oh my God, it's possible on any video, anything you upload, but the paid ads will get delivered. 


Kyle: That's how I always talk to people. I was like, "Well, do you want the free and slow way or the fast and paid way?"


That's pretty much all it is. 


Sayf: Yeah. And so I think it's one of the four core ways to do distribution these days. And I think it always will remain so as long as these companies make money off paid ads, then they're always going to push towards that. I can't imagine not doing paid.


Kyle: I agree. Well again I'm biased but yeah 


Sayf: Well um people don't do paid for - there's reasons why people might not do paid but it's it's generally not because of performance unless it's such a weird field that no one would be - 


Kyle: Right - just a super niche audience or something like that yeah.


So kind of two, two part question here. Have you throughout your career always been in like analytics and digital marketing? And B, do you have a, like a success story or a cool moment that like when you were in marketing and you were doing some analytics and stuff that you were like, "Oh, holy shit, this is super cool. I really want to like keep doing this," like kind of like a big aha moment or something like that?


Sayf: So yeah, first question is no. I start the - what I gloss, I don't always gloss over in my biography, but I actually started as an archaeologist. I went to grad school for archaeology and worked as an archaeologist before leaving academia for, um, initially nothing and then eventually uh uh where I am now but didn't didn't start with data it had a path through um development.


Kyle: And how is AI impacting archaeology these days?


Sayf: I mean I don't think I'm - I have I have not read enough on it to even give up my opinion but so I definitely, I mean, at this point, when I was in archaeology, I was the only person in my, in grad school, I was the only person in my class in grad school that had a laptop.


Oh, wow. And I had to plug it in during a class. It wouldn't last that long. And people looked at me as weird that I had a program to keep all my citations and stuff. And I was the odd one. But I guarantee you it ain't like that anymore.


Kyle: You're always on the forefront. Yeah. Always leading the charge. Yeah.


Sayf: And funnest. I mean, I've had so many different moments that just I loved and it usually was analysis. It usually was trying to dig into just playing with data, maybe not even doing what I would recommend as being valuable time. Right.

So like, you know, I think the thing for me was seeing interesting insights out of data that I don't even know whether you could use it for anything, but it just was interesting. Again, like that was that sort of archaeologist to me. I'm always like, what got me into it was I think hooking me on understanding human behavior.


And when you said, ask that question, the thing that jumped into my mind, we - I was doing work. This is years ago. We were doing work for it was like a - it was a dress website. Uh, I'm not going to name the names, but they, they, they, they did a lot of sort of like homecoming prom cocktail, high, that kind of level dress, uh, like event type dresses. Uh, and, and we did all the - I think I know who you're talking about maybe - um and uh we we collected all their data and we're just playing in a lot of different dimensions on products and product searches and things like that and I played with that data in so many ways I broke it up by metro areas and quartiles and rural and urban and and all this different stuff and was able to just, you know, as a human, dig through the data and find different things like the bigger cities, like the first quartile.


They would search differently than the fourth quartile. And so the first quartile, you know, like the New York City, Los Angeles, Chicago, big city, Metro areas. They liked designers. They searched for designers. They searched design names. They searched for, um, I don't remember the fullest, but designers was a big thing for them.


And then you went down into the fourth quartile, like the, the, the smallest 25% Metro areas, like, um, you know, places in Montana that you've never heard of before. Right. And they, they searched for like - and these were so these were like the smallest, most rural, rural places in the US. And they searched for like patterns.

They searched for animal prints, right? Like animal print patterns were huge compared to anywhere else.


They were looking at like discounts and bargains. So like they searched and used that brand completely differently depending on what the density of the population was in their metro area and so we could come back to that company we could say, "So here's the density here's you don't want to target by state you want to target by DMA based on density and here's and then you want to target them in different ways so for the big ones you want to target for designer brands for the small ones you want to target for animal prints" and like you know, that kind of stuff.


And I don't know. I just thought that was... And I think the second thing was the first time I saw how the weekly pattern for a different client, just like every week, it was sort of like Monday, Tuesday, Wednesday, Thursday, Friday, Saturday, Sunday, Monday. It was the same every week, every week.


And that was a long time ago, but that was one of the early things where I realized that sort of this aggregated data really can tell a story a lot of times better than like a cookie, "this is Joe and this is what Joe did." I don't care what Joe did. Give me a million Joes and then I can tell you like exactly what to do.


Kyle: That's so funny. I wasn't even involved in that, but you just got me so excited. I think that stuff is so interesting. Human behavior, like you talked about, psychology and everything.


It's so interesting to see how the same exact product gets looked for and purchased and the decision-making process is just different from just a location. You know what I'm saying? And I'm sure that it's culturally different and all that kind of stuff, but it's interesting to see, yeah, how people go from designer to, yeah, I want leopard print or something like that.


Sayf: And just like we, yeah, I mean, I can, a bunch of examples and every business, not every business, but different businesses, different things matter and different data that they don't even collect matters. Like I said, like, you know, not just seasonality, but other stuff. And it's just really interesting to see in aggregate. 


Kyle: It is interesting.


Cool. So we're getting to the end here. Before we jump into the rapid fire, I just wanted to go back to Three Bears. What's on the horizon? You know, you guys have your product. I know you just launched Q3, right? I'm making that up. 


Sayf: Yeah. Or no I guess yeah late q3 technically so we launched uh early september we're in early november right now uh so we're we're a little under two months in uh we actually have we are talking to a number of different uh we don't have any sales we're zero arr at present zero mr uh but that's to be expected one of my advisors uh they had their own a company that they sold off years ago for bank.


And it was a measurement company and took them 11 months to get to revenue so I as long as I do better than than them i'll be i'm happy but we have a number in the we have a number of leads actually that we're talking to that are interested that fit the fit the profile we want to work with we are going to uh hopefully you know I i think uh we'll we should start signing stuff relatively soon uh we are kind of giving good deals right now.


We might need to raise prices just because people seem to be reacting that we are almost too good to be true at the price that we're mentioning. So I think I'm like, does that mean we need to raise our prices? I haven't sold anything yet, but...


Kyle: Hey, you got to start somewhere and get a case study or two, and then you can start cranking it up.


Sayf: Yeah. So we have a few right now that we're talking to you about case studies and things like that. And I think then get those done. We have our team all lined up. And so getting those uh, getting those done, getting some case studies and then building off, building off.


That's the plan. And I, yeah, I mean, I, it's one of these things where I just, I, I just really believe in the products like to, uh, it's weird, you know, like the, the, I had a business before that I had my own agency for 10 years and it's a different world, but, and it's not that I didn't believe in it before, but it's just one of these things where I see the market and I see what the competition is and I see that there's legitimate companies out there that have 200 people at them selling a product that still does last click and linear attribution and they have that on their website. Can I swear?


Kyle: Absolutely.


Sayf: That's fucking criminal. It's fucking criminal. Yeah. Like how it just blows my mind that there, that I'm looking at competition that has 150 employees and they literally have last click and linear attribution is things that they do on their goddamn website.


And I'm not going to call them out, but like, I'm sorry. And other people who claim, "Oh, yeah, we do this, we do that." And there's so many people out there that are just confusing the space right now. And hucksters and people that are people that literally are because it's math and it can be hard to understand statistics, because someone is stupid and they can't understand statistics, they're not stupid.


This stuff is hard. It's hard. And sometimes you just have to trust that there's people that are smart that put time into that. And there's people that just go, "Oh, they're the ones who are lying to you. Come buy my course for me or whatever."


And yeah, so it's one of these things where I've seen it. We've done it. And we've done it in the past for other clients at former agencies. If you haven't done it at all, we're guaranteeing 5% because it's going to be 10% to 15%.


But we guarantee if you don't get 5%, we don't find 5% opportunity of your budget, then you don't pay anything. And we're just as likely to not do that as to get 20%. There's 30% waste. There's 30% or more waste in everybody's accounts.

And they're not doing this stuff because the agencies or whatever have been charging quarter million dollars to do it. And we're saying, "We'll do it for $5,000 a month for the 5% find that you've guaranteed you don't have to pay anything." And - and I think that sounds cheap to people.


And they say, "That's too good to be true that you could actually tell us that whole customer journey between multiple people and different things." It's like, "Yeah, you aggregate all that data together and you put it into AI machine learning. And guess what? You can. If you got enough of it, you can." And I don't know.


I just think there's a lot of value out there and it makes me excited to do it. So I think we get these case studies out and we keep on hammering and I just keep talking to people and just explaining how much I believe in it and then someday like we'll be - 


Kyle: Well I can feel your passion, Sayf. Let's let those con artists keep talking about linear uh and you guys can provide the real value that's right it's an opportunity there you go. Um all right cool five quick rapid fire questions for you you ready


Sayf: I don't know how quick I can be but go


Kyle: Yeah fair enough it's all good um All right. If you could snap your fingers right now and have one automation fully built out, what would it be?


Sayf: Wow. One automation fully built out. It would have to be something with distribution. I would probably say some kind of AI automation involving a warm outreach kind of thing where, you know, I don't know if I would utilize Apollo or or what, but some kind of some kind of AI automation that could both, uh, do warm outreach to people that I'm already connected with, but do it in a personalized way that doesn't sound like a copy paste or whatever else?


Yeah. Cause I need to do a bunch more sort of warm outreach. And so if I had that built out, that would be helpful.


Kyle: Never enough outreach. Yeah. I feel like I know what your answer is going to be based on our last conversation, but what is something in the world of analytics that you wish would go away forever?


Sayf: Yeah. Something in the world of analytics that I wish would go away forever. You can edit out my thinking. Pause. I don't know. I mean, I definitely think - I mean, definitely rules-based attribution, like weighted attribution, anything like last click, linear, time decay, position, any of that.


Because it - it's just, it's, it's lying to people in a way that it looks, it makes people think, "Oh, it's, that makes sense." But it's so just not accurate in any conceivable way that it bothers me that it's even available to people. It served its purpose years ago, but now there's, yeah, a lot more that we can be doing. So yeah. Yeah.


Kyle: uh all right cool so paint me a picture you have a messy data set um for a client and you're gonna dig into late at night you know if I was a fly on the wall what would I be seeing for you like in in like your your your room there what's what's that like


Sayf: uh well you get the reverse view and uh I think it would depend on the data set um I do like to, I do like to play a lot in whether it's even a Google sheets or Google Excel.


I, it's not necessarily the most advanced way to do it. I should do more where I'm throwing something into BigQuery or something like that. And, and I'm running SQL on it, which I could do, but there's, I don't know, there's something, it's like the digital equivalent of like a, of a, an open fire in the fireplace to me is to have data in a spreadsheet and be filtering on it and be pulling up some just you know pulling up some visualizations on it and trying to look at it in different ways um and so like if someone has a dirty data set - dirty data set sounds dirty and self - like um but if someone you know if something's it needs to get cleaned like just exploring a lot of times filtering stuff at the first level can, for me, get an idea of what we're looking at or just taking a sample of it.


Depending on how many rows it is, I don't want to... But if you take a good sample of rows and then start taking a look at what's there, you can start seeing patterns and understanding.


And then from taking that, I can automate it a bit more and have a bit more to do actual cleaning. But, you know, the reality is, is that I'll be honest, I don't do that as much these days because, you know, if you want someone to play with data, if it's not an AI, it's someone much younger than me, that's going to do it a lot better. But when I do it for fun on my own, I'll admit, but yeah.


Kyle: Fair enough. Fair enough. Which of the three bears are you?


Sayf: So this has been, this has been a discussion. I think I am technically a bear. I mean, in a way I'm bear number three, I'm either bear number one or I'm bear number three, but I'm not bear number two.


Okay. And then we've done different things by to label the bears differently where it's like, well, you know, black bear, brown bear, polar bear, or, you know, that kind of thing. And, and, and to be clear, we have, we, you know, they're, they're, they're, there were actually four founders.


And so the fourth took the nom de plume Goldilocks.


Kyle: I was going to say, yeah.


Sayf: But yeah, I mean, we don't discern our numbers. 


Kyle: Okay. I feel like there's going to be like an internal battle when this gets released now. 


Sayf: I'm the bear that likes his porridge hot. 


Kyle: There you go.


All right. And my last one is, you know, I, I've been, uh, admiring your, uh, your room here. I don't know if it's your office or your basement or what, what, what is your favorite like figurine or, or, uh,


Sayf: Oh, well, everybody does. Yeah. You know, my favorite, my, my favorite, um, because I have more like if I turn this it's like more over there and yeah over there oh wow like I got a lot well I pulled everything out a while back during covid because I I started doing that it's it's just grown but it was sort of a reverse Marie Kondo where it's like I have these things they bring joy to my life why do I have them in a tub in the attic let me put them somewhere I can see them so I would say favorites uh one of them would be I keep over here um this is an original Cylon uh from the original 1978 I think it was TV series and he's got his gun and that's had that again I've had that for a very long time so i'd say that's one of my favorites but yeah I got it I got I got a bunch I have a lot of hulk I was gonna say yeah yeah I i i've become a marvel guy uh actually since covid um funny enough I got into marvel I love the mcu now nice very cool old old hulk there but yeah I don't know I have uh I also have some other star wars stuff I have my original 1980 Empire Strikes Back with asteroid field damage TIE fighter that I found in the attic. That's incredible. Just a lot of stuff. It just makes me happy.


Kyle: It made me happy seeing it. So I get it. Yeah. Well, awesome. Sayf, we're at the open forum section. Is there anything that you're passionate about that maybe we didn't discuss that you want to just share with everybody real quick?


Sayf: Gosh, passionate about. It's, it's, I would say, I think, I think one of the things that I'm, I've been passionate about is to, I mean, I don't know how far, I don't know how far field we go, but I have 


Kyle: No wrong answer.


Sayf: You know, it's, I would say making, making oneself a better person, but not doing it in some cheesy, let me buy a self-help book way, but actually, like, how do you change thought processes as a human being and doing, I've been doing a lot of, a lot of studying and a lot of work and a lot of meditation the last, particularly the last year, year and a half, two years. Uh, and, it's interesting, right?

The more you kind of, it's kind of like metacognition, I think they call it. Just thinking about how you think is something I do like every day now. Partly because I also have a year and a half old Bernie Doodle puppy who still has so much energy. She jumps up this high. Oh, yeah. She's insane.


I love her dearly. She's the first dog I've ever had that is like my dog. Okay. But I have to walk her a lot. So for an hour and a half a day over two walks. I have a lot of time to think and just been doing a lot of that metacognition and how do I change my thought processes and stuff? And it's just been, it's a very interesting kind of topic that I've been working on.


Kyle: That's very cool. I think there's a lot of people don't think about that kind of stuff. People don't realize how much the subconscious drives us. You just act or react without thinking about it. That's cool.


Sayf: I'm a big proponent of that. My first thing these days, I'll have an emotional reaction. or i'll have a immediate thought reaction so something happens and I have oh this is what I think about it this is what I feel about it and I almost immediately go why why did I feel that way yeah and think about how I came to that feeling and my you know that that and it's just it can be you know it's not it's just interesting it's this and how many times i'm and i'm gotten better at just being like oh it's this Right. But I, you know, at a certain point you understand your traumas enough that you pretty quickly are like, oh yeah, that's that.


Kyle: Hey, I mean, you know, just like with data, you know, you can't fix things if you don't identify it. Right. You got to know what the issue is. So yeah, that's a good point.


Sayf: I think it's just a, it's a micro expression of me just trying to understand humanity. My whole life has been about trying to understand humanity. humanity whether it's data analytics or archaeology or when I was like got in trouble I think I got a detention one time for asking how people like took a shit in a castle and I really meant it I'm just like where did they go and like I got in trouble for it but yeah I always want to just understand how how people live and how people do things including myself so yeah very cool very cool


Kyle: Well, Sayf, I appreciate you taking some time out of your day to share your insights and talk with me about a bunch of different stuff. They can find you on LinkedIn. They can check out 3Bears at 3BearsData.com. Anything else you want to plug or kind of get out there?


Sayf: Or just go right to our product page to Optimeasure.io, O-P-T-I measure dot IO. It also can get linked from three bears data.com, but that's the actual product is Optimeasure. Um, but yeah, that's, that's about it. Awesome. Well, thanks for having me.


Kyle: Yeah. As always, it's, it's, it's great to see you. I appreciate you joining the, the brainiac blueprint. Um, if you don't mind, look at the camera and say, "stay brilliant, brainiacs."


Sayf: Stay brilliant, brainiacs.


Kyle: Cheers. Thanks.

 
 
 

Comments

Alright Brainiacs! Let’s Build Your AI-Powered Business

Ready to transform your business with AI? Let’s talk.
Contact us today to schedule a consultation!

bottom of page