Mastering Marketing Attribution: Uncovering the Secrets to Revenue Impact

Brandi Starr [00:00:34]:
Hello, hello, hello and welcome to another episode of Revenue Rehab. I am your host, Brandy Star and we have another amazing episode for you today. I am joined by Zeke Camusio. Zeke is a serial entrepreneur and the founder of Data Speaks, an AI powered analytics platform platform that helps companies identify what drives their sales and invest in the right marketing strategies. With a background in economics and data science, Zeke has spent the last 20 years developing AI machine learning and data analytics solutions, enabling hundreds of companies to make data driven decisions and accelerate growth. Zeke, welcome to Revenue Rehab. Your session begins now.

Zeke Camusio [00:01:24]:
Thank you for having me. I'm excited to be talking to you.

Brandi Starr [00:01:27]:
I am excited to talk to you as well. I love to talk about data and AI. So it is great to have you here. And before we jump into our topic, I like to break the ice with a little wooa moment that I call buzzword banishment. So tell me, what buzzword would you like to get rid of forever?

Zeke Camusio [00:01:50]:
I don't like when people say I'm gonna circle back with you because it's so, it's, it's not concrete enough. You know, I want to know when. I want to know. I actually like it when people say I don't know and I will get back to you, but I, it's over. Sometimes it's overused and not specific enough. So I just want to know what the next step will be.

Brandi Starr [00:02:16]:
Yes. So I'm with you. Let's not circle back. Let's get real specific on the action items and when we're going to follow up. So I can promise that we're not going to circle back in this conversation. So we can put that in the box, throw away the key. And now that we've gotten that off our chest, tell me what brings you to Revenue Rehab.

Zeke Camusio [00:02:38]:
Well, I love to share with you how your listeners can be using analytics, data and AI to move their companies forward.

Brandi Starr [00:02:48]:
Okay. And I believe in setting intention. It gives us focus, it gives us purpose, and most important, it gives our audience an understanding of what they should expect from our conversation today. So what's your best hope? What would you like people to take away from the discussion?

Zeke Camusio [00:03:03]:
Yeah, so I would love to talk about the, the one thing I have been specializing in for the last 20 years. Which is marketing attribution. And what is marketing? Attribution is really understanding out of all the different marketing campaigns, channels, activities, how much of your outcome, whether it's leads or revenue, how much comes from each of those activities? And that then allows you to focus on the right things.

Brandi Starr [00:03:32]:
Okay, well, that's a good place to start because I can say I've got a little bone to pick with attribution. And I think it's because so many organizations tap into attribution as a way of getting credit. Like, you know, it's like this, you know, the marketing team wants credit, the digital team wants credit. Like, everybody wants credit for the revenue. And I feel like attribution often creates friction. So I'd love to hear your perspective on how do you see attribution effectively being used in businesses?

Zeke Camusio [00:04:11]:
Yeah, of course. And you make an excellent point. You know, it's. This is not meant to be a conversation about, you know, you versus me. Right. Like we are, you know, if we're together, working together, we're a team and we're all pulling in the right direction. I think attribution has more to do with understanding how to optimally allocate your resources as a company. You know, if you have $100, what is the best way to allocate the $100? If you have 100 hours, how should we spend those hours? So I'm a big believer that what we decide to do is more important than how well we do it, because you could do something really well, and if it's the wrong thing, it's not going to lead to great results.

Zeke Camusio [00:04:58]:
So I think that being able to identify what is going to create the biggest impact for my business, you know, then allows you to focus. And that little focus and concentrating your resources is really what moves companies forward more than anything else. So that's a little bit of, on the, you know, the kind of the philosophy of attribution, I think it should be a way to help organizations make the best decisions possible and not, you know, try to take credit from you or vice versa.

Brandi Starr [00:05:36]:
Yeah, and I definitely agreed. That is where I see attribution being a huge benefit to organizations, because there's a lot of marketing activity that's happening. And, you know, way back when, it was a lot of throw it against the wall and see what sticks. And now we've got technology and data that allows us to actually see what is influencing our revenue so that we can do more of whatever's working and stop doing whatever is not. And I know just from some of our back and forth before this interview that you've got perspective and insight on how marketers can really understand what drives their conversions. And so, you know, we talk a lot about the funnel and what's important and how people move through the funnel and the importance of optimizing all of those different conversion points. So I'd love to hear more of your insight and experience around that as well.

Zeke Camusio [00:06:40]:
Yeah, yeah, absolutely. So I really like what you said about throwing stuff at the wall and seeing what sticks. You know, the. The challenge we have is that we often throw a lot of things at the wall and we're not able to see exactly what sticks. You know, so what I mean by that is that let's say that I'm running campaigns on Google, some campaigns on Facebook, some campaigns on LinkedIn. You know, I have email marketing, I have referrals, I have affiliates, I have influencers. And then at the end of the month, I have, let's say, a hundred leads or a hundred dollars in revenue. So we're throwing a lot of things at the wall, and we're not able to identify how much of those hundred dollars are leads, you know, was contributed to by each individual channel or campaign.

Zeke Camusio [00:07:32]:
So that's the most common challenge marketers have. And the reason for that is that for the longest time, we've been relying on these pixels, and we've been hoping that we could just put a pixel on our website and that's going to tell us what's going on. Well, there are a lot of issues with that. So there's, you know, ad blockers, browser privacy settings that don't allow us to track about 40% of all conversions. So we're not able to identify the source of a lot of those sessions. Right. There's also the fact that people use different devices. So maybe you find me on Facebook, sign up for my emails, then open my emails on a desktop.

Zeke Camusio [00:08:18]:
You know, I don't know it's you anymore. So the, this idea that we can follow individual customers across the web, across every channel and device, it's just not possible anymore. So what happens is we're relying on the attribution data that we get from the individual advertising platforms. The problem with that is that let's say that somebody sees your ad on Instagram, then clicks on your Google Ad, then opens an email. You can have three different platforms claiming credit for the same sale. So it's never going to add up because you got one sale, not three. Right. So the problem is when we look at the amount of revenue that say, for example, Google Ads is claiming, and we say, you know, I invested $1,000, I got $10,000 back.

Zeke Camusio [00:09:13]:
What a great return investment, I'm going to invest more. So that's, that's the issue, right? We, as marketers, we don't have any reliable data in terms of figuring out is Google a good investment or not. And if you think about it, this will be unacceptable in any other area of life. You know, if, let's say you have 10 different stock in your investment portfolio, and then your portfolio manager calls you at the end of the year, tells you that the value in your portfolio increased by 10%, but it's refusing to give you the performance of each of the 10 stock in the portfolio. Well, maybe one of them increased 20%, the other one decreased 10%. You want to know what to keep and what to sell. So it's a big challenge even for very, very experienced data scientists. It's really difficult to model that kind of stuff.

Zeke Camusio [00:10:14]:
And just to give you a few examples of why it's so difficult when, let's say you run ads to 100 people today, you're not going to see 100% of the benefit today. You're going to see some of it today, some next week, some next month, and so on. And that's different for all different channels and campaigns. You also have the, you know, the diminishing returns. So the return on investment you get when you invest $1,000 is not going to be the same as when you invest $10,000. So as you grow different campaigns, you're going to see different returns on investment. So a really good attribution model needs to take into consideration all these different behaviors, what's happening in your company, what's happening in the world, what's happening in all your marketing channels, and be able to come up to the, the closest explanation for how you got to the outcome, whether it's leads or revenue that you got.

Brandi Starr [00:11:17]:
Yeah, and I think you hit on a number of good points there. I really like the analogy of the stock portfolio because you're right in that, you know, if that happened, you'd be looking for, you know, another portfolio manager. Whereas, you know, from a marketing and attribution perspective, it just, you know, it kind of gets written off as it is what it is. And so while I do agree, the question that I have is what do we do about it? Because to your point, there is a lot of activity that cannot be tracked. There's a lot of activity that's happening that we can't Connect as knowing it is the same person. And so what we've seen become acceptable in the marketplace is, you know, we have the first touch, last touch, you know, are like the most common things and we're just going to give air quotes all the credit to the first or the last trackable thing. And there's, you know, some platforms that get a little more sophisticated in, you know, the multi touch, but to a certain degree they are all flawed. And at the end of the day, you know, what we're all trying to accomplish is exactly what you said in figuring out where do we continue to put our resources, where do we continue to to invest and how do we make those investments work harder for us so that we are able to, to scale it and increase the amount of revenue.

Brandi Starr [00:12:49]:
And so like what do we do about that? Like this is one of those things where it, I feel like it's a conversation we have over and over about how it's a problem and that's where it ends is just kind of like, yep, this is a big issue. And you know, it's one of those like we just walk away and agree to allow it to be an issue.

Zeke Camusio [00:13:09]:
Absolutely. No, I love your question. Right. Because that's really what we should be asking ourselves. You know, like how do we get that those answers and, and the, you know, you even talked about one of the two ways that you can use which is MTA or multi touch attribution. There are a lot of limitations with that. As you said, it's arbitrary, right? So it's if you use last click, why would you discount everything that happened prior to that last click? If you use the first click, how do you know that that's what gets the, should get the credit rather than all the supporting channels that happen after that? Even if you decided to split it evenly across all of them, how do you even know that all those touch points had the same weight or the same influence on the purchase decision? So while multi touch attribution is useful for seeing people went from this ad to this email to this and this in terms of qualitative data is not useful for actually measuring attribution. What's useful rather than MTA is impact based attribution, it's not based on the, the fact that we have all these touch points, but you have to measure the individual impact of each.

Zeke Camusio [00:14:38]:
So how do we do that? Think about it. In the most simple example that I can think about, you're investing $10,000 a month on Google and you double it and what happens to your Sales, when you double it, do they increase? Do they stay the same? What happens when you, if you turn it off, you just stop investing in Google. What happens to your sales then? Do they increase, decrease, stay the same? Right. And if they decrease, by how much? And that's how we measure impact. The problem with that, of course, is that we can't just double our investment arbitrarily or just pause campaigns. I mean, that would have a pretty big negative impact on our business. So what we do is we use incrementality. Incrementality is basically the marketing lingo for a randomized control trial, which is you have a control group and a treatment group.

Zeke Camusio [00:15:38]:
And in this case, you have to find a random variable that has nothing to do with performance. So what I mean by that is, in this case, the most common one that we use is geographical segmentation. So it could be by zip code, by state, by city. So if you increase the advertising spend for, say, Alabama, and you see your leads increase by a certain amount in Alabama, that would be a very strong signal that of the impact of that specific channel. Now, the way to design a system to be as useful as possible is you want to maximize the statistical relevance of the test while minimizing the negative impact it's going to have in your business. So if most of your sales are from California, you don't want to be turning off on California. You know, you want to work with states that represent a small percentage of your overall market, but are good samples of what happens in the nation if you sell in the United States. So that's essentially the closest way to measure actual impact.

Zeke Camusio [00:17:01]:
And nothing is perfect, even sometimes. We're usually in the 90 to 95% accuracy in terms of being able to predict what's going to happen if you invest 10,000 here or here or here, but nothing is 100%. I think what's important to keep in mind is that right now, most companies work with data that is not even 60% accurate, and that's really, really bad. So if you can go from 60 to 95%, that's a significant increase in your ability to allocate your budget optimally, especially if you have really large budgets.

Brandi Starr [00:17:41]:
So I think the, you know, listening to that definitely feels a little overwhelming. And my first thought is we need a data scientist on the marketing team. And we're, you know, just thinking about Most of the CMOs that I talk to and just the amount of pressure that is on everyone to get so many of the things done. How are you seeing this play out in companies effectively, because it's one of those things that it's like, this sounds great in theory, but will I ever have the resources and or bandwidth to be able to execute on that level of testing and analysis?

Zeke Camusio [00:18:27]:
Yeah, so I think it's the. How much the importance of doing this is proportionally to the amount of dollars that go into your advertising. You know, so what I normally say is, like, if you're investing, you know, under $20,000 a month, you can get away with a very low level of sophistication. You know, just see what the platforms report, run some tests here and there, you know, look for the best creatives, best audiences. But you don't really need this level of data science. But when you're investing, you know, hundreds of hundreds of thousands or millions per month, then the. What we found on average is that we're able to get 22% better return on ad spend just by reallocating your existing budget. So if you're spending a hundred bucks, we'll show you a better way to spend a hundred bucks.

Zeke Camusio [00:19:24]:
You don't have to spend any more. You'll get 22% more conversions or more leads. That's an average. You know, sometimes it's 40%, sometimes it's 18, but that's significant. If you are investing a million dollars a month, that's $220,000 that you're leaving on the table by buying the wrong media. Right? So as your budget gets larger and larger, it's more and more important to do that. Now how do you do it? You can do what you mentioned, which is you hired a team of data scientists that you invest in infrastructure like a data warehouse, a customer data platform, analytics, machine learning infrastructure, and so on. Or you can hire a company that helps you do that.

Zeke Camusio [00:20:12]:
We do that, of course. You know, like, that's we, you know, the need that I saw in the market, you know, when I sold my last marketing agency in 2015, I started doing consulting for a few clients. And I. What I realized was one of my clients made an investment. I remember the ROAS that the platform reported was around 5. So I don't remember the exact amount he invested, but let's say that he invested 10,000, and rather than getting 50,000 in return, he sold, you know, 20,000. He's like, what? But this is a Ross of two. You know, it should be, you know, it should be $50,000.

Zeke Camusio [00:20:48]:
It should be a Rosa 5. And I was like, no, that's not what it is. You know, and we had this Conversation where I was like, that's what it's supposed to be, but it's not right. And you're supposed to make this investment without the right data. Right. So that was really the, the aha moment for me as an entrepreneur where I saw, you know, it's not just this guy, it's everybody else. And you know, we are relying more and more on pixels and pixels are working worse and worse every year. Right.

Zeke Camusio [00:21:22]:
So yeah, I mean, I, you know, because this is what, what we do, you know, you know, it might seem like it's really the, the only path, but it's not. You can hire a data scientist, you can do this yourself. And again, I mean, if you're small enough, you might just get away with just doing what the platforms are telling you until you have the funds to get serious about how to properly allocate your budget.

Brandi Starr [00:21:54]:
Okay. And I think the question that I would have is we've talked a lot about how we are able to look at this in order to understand attribution and optimize ad spend, but I think that that's only one component of it. When you're looking at this for clients, how are you thinking about. Because you know, I think about like ad spend is primarily like top of funnel attract people to you. How are you accounting for the things that happen more middle of the funnel where they're being nurtured or they're being touched by an SDR or you know, if it's software, there's a trial. Like how do you factor in all of those things as well so that, you know. Because I think one of the drawbacks I've seen of attribution is making it seem like one channel is really performing well or not performing well when it's really the influence of other things. So how do you account for all of those activities that are leading to revenue when you're, you know, for clients that are working with you guys trying to solve this?

Zeke Camusio [00:23:05]:
Yeah, and that's where our, you know, our job as data scientists, you know, gets really exciting because we get to talk to, to each individual client we work with and understand their particular business model. So it's important to know that most machine learning algorithms optimize for a certain outcome. So if you're optimizing for new client acquisition, that's gonna be, you know, you're gonna do a set of things to get the, you know, the, the, the most customers for the lowest cost possible. We feel like that's very short sighted. You know, there's a lot more to the business than just acquiring customers. For example, understanding the lifetime value of individual clients is really important because you could have campaign A, campaign B, campaign A. You know, you get customers for 80 bucks. Campaign B is $100.

Zeke Camusio [00:24:01]:
But those customers are, you know, stay with you for longer, spend more, you know, just better customers, you know, so understanding that is really in terms of what it is that you're optimizing for lifetime value. Customer retention, new customer acquisition, how to balance out all those things to have the optimal way of allocating resources in your business. And there's also the more qualitative data that needs to be analyzed. It's not just taking dollars from here to, to there, but it's also like, what, what are the. The activities that prospects engage in before becoming clients? You know, what percentage of those need to talk to sdr? What percentage of those, for example, watch a webinar or, you know, download downloads, an ebook before converting what percentage of those? Like, look at a pricing page. So that's in machine learning, that's called feature selection and feature engineering, where we say, out of all the possible things that could cause this, what are the things? Right? Like, let's start with a list of, you know, with our domain knowledge of some of the usual suspects. We interview our clients, we try to get some ideas for what activities they're doing that maybe we're not measuring, maybe we're not looking at. And then a great team of machine learning engineers will be able to take all those things and look at how much signal you have for each, discard things that don't really explain your data very well and keep those factors that are relevant and actually have an impact and not only correlation, but a causal relationship and can explain your data as well as possible.

Brandi Starr [00:26:03]:
Okay, yeah, I think exactly what you said is where I always run into challenges, because that's the other thing I see is like attribution models looking at one particular outcome when you could be trying to optimize for deal size or, you know, all sorts of different things. And so hearing all of this, you know, I think about the hardest part is always getting started. And especially for someone like myself who, you know, is not a data scientist and is only more of a user of these, you know, user of the data as opposed to one that's interpreting it. It is, it's very much the where do we start? And so we say talking about our challenges is just the first step. And nothing changes if nothing changes. And so in traditional therapy, the therapist normally gives the client some homework, but here at Revenue rehab, we like to flip that on a TED and ask you to give us some homework. And so if we're, you know, if our, if our listeners are like me and sitting here, like, figuring out, like, how do I work the magic? You know, short of, obviously they can reach out and talk to you, but in general, what's going to be that first step? What's the one thing that we need to be thinking about identifying in order to figure out if this is a path we should even go down?

Zeke Camusio [00:27:35]:
Yeah, so I want to make a very important clarification, which is insights are not the ultimate truth. You know, what they give you is a very good approximation of what might be going on. Right. So the way I explain it is if you are in a foreign city that you've never seen before, you're going to get to your destination way faster. If you have a gps, it doesn't mean that it's not going to give you the wrong turn at some point, but it's going to make your job significantly easier. So I think it's important to adopt the scientific method of experimenting, you know, of taking insights to formulate a hypothesis and then putting it to the test. Right. So if the, the hypothesis that is that if I invest 10,000 in this, I'm going to get 50,000 revenue, I don't have to take that as face value.

Zeke Camusio [00:28:30]:
I can actually test it. You know, we believe that that should be part of any kind of attribution platform. You know, not only the ability to get insights, but also the ability to track the outcomes that you get once you implement them. So I think it's really important that applies to everybody across the board, regardless of your business size. Now, in terms of the homework, I'll say that if you're investing under 20,000, again, I think that's enough just knowing that. I think that the advice that I have at that stage would be even though you don't have an attribution platform or a team of data science scientists, creating the habit of sitting with your data once a week is really key. So you need a consistent schedule and you need to agree before going into it what kind of metrics you're going to be looking at. So you're not looking at different things every week.

Zeke Camusio [00:29:28]:
You decide, you know, these, we, we say one primary metric between three and five secondary metrics. So worst case scenario, you have six metrics to look at, you know, and that's intentional because you want to, you don't want to be overwhelmed with a lot of different things. You Know, you want to have. You want to be focused on what really matters. And. And then you're just putting your calendar, do it on a weekly basis. And you're going to learn a lot by following these trends, you know, so I think that's really key. Now if you are over 20,000, that's where it makes sense to start figuring out your attribution.

Zeke Camusio [00:30:06]:
And you really have two paths. You know, you could hire data scientists, you know, build this kind of infrastructure, or you can work with a team that does this for you. You know, Data Speaks, my company, is one of them. But there are a lot of companies out there that I can recommend very highly, you know, so Recast is great. I think it's like get recast.com northbeam IO also a really good service. So, you know, we're not the only one. But the way I normally look at it is most of these platforms will probably pay for themselves 20 to 100 times. You know, so you are leaving money on the table.

Zeke Camusio [00:30:55]:
On average, we found that it's 22% of the money you're investing in ads, and fixing that problem will cost you $1. You know, if you're, if you're leaving $22 on the table, it's going to cost you $1 to get those $22 back. So most of the time, it's going to be a fantastic investment. You just don't know. I think most marketers presume that there's. They're leaving some money on the table, but nobody really knows what. What. Where did those $22 are going.

Brandi Starr [00:31:28]:
Yeah, and I definitely think that that's a common challenge of where you know that you're not quite where you want to be, but it's always hard to figure out exactly what's not working. Well, Zeke, I have enjoyed our discussion, but that's our time for today. But before we go, definitely tell our audience how they can connect with you and also how they can. We talked a little bit about what Data Speaks does, but tell us how they can get more information there as well.

Zeke Camusio [00:31:58]:
Yeah, so if you want to learn more about Data Speaks, you can go to Data Speaks AI if you want to get in touch with me, that would be Zeke Z E K e data speaks AI or you can find me on LinkedIn. And, you know, I'm happy to answer questions. You know, I feel like I spend a lot of time talking to entrepreneurs, marketers, and a lot of what I do is really education and setting them up for success. Sometimes we end up working together, but even if it's not the right time. I do want to be a resource to you and a lot of people have helped me get to where I am today. So it's pleasurable for me to be able to pay it back whenever somebody needs help.

Brandi Starr [00:32:45]:
Awesome. Well, we will make sure to link to your LinkedIn as well as the Data Store Speaks website. So wherever you are listening or watching this podcast, check the show notes so that you can connect with Zeke. Zeke, again, thanks so so much for joining me. I have truly enjoyed the discussion and I've learned some things.

Zeke Camusio [00:33:03]:
Awesome, Brandy, thank you for having me. And yeah, have a fantastic day.

Brandi Starr [00:33:09]:
Awesome. I hope you all have enjoyed my conversation with Zeke. I can't believe we're at the end. Until next time, bye. Bye.