How Data Speaks uses impact-based attribution to measure and improve advertising ROI

Transcript

Jessica Gondolfo:
Hey there. Welcome back to the Marketing Intelligence Show. My name is Jess and I'm the head of US marketing here at Supermetrics. Today's episode is all about impact based attribution. That's why I'm so excited to have Zeke Muccio here to tell us all about what that even means. Zeke is the founder and CEO of data speaks, an all-in-one analytics and attribution platform helping e-commerce brands to improve their advertising ROI. So without further ado, Zeke, welcome. I know our audience is excited to learn about this and so am I. So yeah, welcome.

Zeke Camusio:
Thank you, Jess. It's great to be here. Thank you for having me.

Jessica Gondolfo:
Yeah, great. So before we kick into the good stuff, you are a marketing expert whose formal background is data science. So I want to kind of jump in that and how did you even get here? What did that journey look like?

Zeke Camusio:
Yeah, so I never had a job you could say. So I've been an entrepreneur my whole life. I started my first business at 15. I started eight businesses in my life. Three of those were acquired and they were all in the digital space. So from 2007 to 2015, I had a marketing agency that got acquired in 2015, and then I started doing some consulting for some high-end clients. And with my background in economics and statistics, I naturally gravitated towards analytics and helping brands make data-driven decisions.

Jessica Gondolfo:
That's amazing. I love this. I keep hearing this blend between data and marketing and how a lot of data folks are really helping in the marketing department. So that's really super interesting and I definitely want to get into the how of data speaks, but maybe first, do you want to talk about who Data speaks is and what you guys do?

Zeke Camusio:
Yeah, of course. So essentially we're an all in one analytics platform with a focus on attribution. Attribution is the biggest challenge that modern marketers have. And what I mean by attribution is how do you know the actual return on investment that you get from every channel in every campaign where you are present? Customer acquisition cost is the highest cost that most companies have. It's a lot easier to keep bringing back your returning customers. Like getting new customers for the first time is very costly, and that's where most of the advertising or the marketing budgets go to digital advertising. So if you can get more efficient in getting customers for lower cost, you can then transform your business. If you're able to get more revenue without spending more on advertising, just by properly allocating your budget, you're going to be much better off. That's the main problem that we solve.
We help you understand the ROI for every piece of your advertising, so then you can allocate your budget in the most efficient way possible to maximize return on investment. So think about it as the stock market. Maybe you're holding five different stock. Well, if you look at the average, maybe it has a 10% return annual return, but some of those might be losing and some of those might be doing 20% or 30%. So unless you understand how each stock is performing, you're not going to be able to reallocate your portfolio and buy the best stock ever. And that's the challenge marketers have. They're relying on platform reported revenue or raw, which is off by a significant amount.

Jessica Gondolfo:
Yeah, that makes sense. And I was just talking to my CMO about this literally yesterday, and she was saying the funnels are just broken. They don't even exist anymore, especially in e-comm, which is a big space that you play in, and the journey is just different, right? You stop, you start, you stop, you start, you have 12,000 different platforms online, offline, so it's how do you know what's driving impact? And I feel like that's a great segue into we most markets will understand the concept of attribution and roas and last click, but what is impact based attribution?

Zeke Camusio:
The best way I can explain that is with an example. So let's say you wake up one day and you decide to go to the movies and you drive your car to the movie theater on your way in, there's somebody holding a sign saying movie tickets, 10% off. Well, you were going to the movies anyway, right? So that didn't give you two purchase. Now, let's say that somebody is just walking along the street and they see the sign and based on that sign they decide to go to the movies. Well, what platforms will tell you is that they generated two sales, but in reality, one of those sales was going to happen anyway, and the other one was the incremental value that that person holding the design provided. So that's what impact based attribution means is when you look at platform reported roas, there's this gross assumption being made that if you saw an ad that cost a sale, in this case even worse when let's say you see two people holding a sign.
So those two people say, Hey, I sold one ticket, but you didn't have two tickets. You only have one ticket. So that's what happens online. You have multiple platforms claiming credit for the same sales. So when you add up the revenue that each of these platforms claim, it will never add up to your total revenue. So that's what impact based attribution means. It means that what we're measuring here is not whether or not somebody saw an ad, but what actually drives sales. So how can you measure that? One thing you could do, for example, is you could stand on that corner and see how many tickets are sold when that person is there versus when the promoter is not there. So let's say that when the promoter is there, you sell a hundred tickets a day when the person is not there, you sell 80 tickets a day. So the incremental value of that person being there is 20 tickets per day.

Jessica Gondolfo:
This is probably a complex question that I'm going to have following up to this, but how is data speaks solving that for clients for more of a technical side? And I'm a marketer, not a data scientist, so

Zeke Camusio:
Absolutely bring it down. Yeah, no, and I think it's important to understand it in a very simple way because going back to this example, you send a private investigator to stand on that corner to see what happens. Well, you can learn something from that, but what if you had five 50 private investigators in at different theaters? Well, that's what we do. There's 50 states in the us. So assuming you sell in the United States, what we do is we say, okay, we're going to have one measuring device in each of these states, and then we're going to look at the advertising spend and the revenue for that stage for that day. So not only do you have different channels, but within channels you have different media types. So for example, Google, you could do brand search, you can do non-brand search, you can do discovery, you can do YouTube, you can do Performance max.
So all these are inputs in statistics we call these independent variables. And then the dependent variable is the output is in this case, your sales. So your sales need to come from your source of truth. So let's say for example, Colorado, as you increase and decrease advertising spend for all these campaigns, all these channels, what happens to your sales in that state? So let's say that one day you decrease your Google performance Max spend by 50% and your sales stay the same. Well, that's an indication to the machine learning that that's not really doing much because if you remove it and your sales are not affected by it, that's an indication that it's not doing much. Now if the spend does go down and then the revenue goes down, then that's an indication that that's working, right? So that's an oversimplification of this. There's 50 different states, so it wouldn't make sense to just look at one state and see what happens there. But when you have 50 daily observations and you are analyzing this every single day, it's something that no human can possibly do. But with machine learning, we have this ability to really have all these private investigators all over the country measuring this and understanding the actual impact that all these campaigns are having. On your bottom line,

Jessica Gondolfo:
A lot of marketing is always has to have some kind of call to action because that's when you have a last touch model. If you didn't have a buy now or learn more or something leading you into leading you down a path that's going to be a purchase, then it's hard for you to essentially measure because you don't really, your brand impressions are hard to attribute back to revenue output. So does this model help brands to be more creative in how they are, what campaigns they're posting? Because now they're not necessarily looking at a UTM of conversion, but they're looking at an overall impacted spend. Am I right in that?

Zeke Camusio:
Yes. When you talk about calls to action, when you talk about last click, I think it's important to say two things. One is that last click is not useless. There's some value in that data. The second thing is what is that value? Well, that value is understanding what that last interaction was. Right now, we can't say that the person, if we think about soccer, we can't say that the person scoring a goal had a hundred percent of the credit for that goal. There's a lot that happened leading to that person scoring the goal. So that's really important for marketers to understand. You can't just ignore everything that leads to something happening and assume that because the last click came from a certain email, that's what got you a sale. Sure, that played a role and that should have some contribution to the overall path. But there's a lot that happened before that, and by properly measuring the weight that each step had, then you're able to understand the percentage impact that each touch point had in the overall sale.

Jessica Gondolfo:
Yeah, that's super interesting. When you are working, there's probably multiple contacts that you're working with in an e-comm brand, but who would you say you are working with most inside these e-comm companies? Who's your main point of contact for here?

Zeke Camusio:
We work with CMOs, sometimes CEOs, sometimes we work with VPs of marketing, essentially the person deciding how to allocate the marketing budget, and we help them. We give you strategic guidance in terms of you have this many dollars to invest in marketing. What we normally find is once we start tracking a performance is that for most brands, they're 20 to 40% suboptimal. What that means is that by taking the, properly allocating the budget in the most optimal way, they're able to get 20 to 40% more revenue without increasing that spend just by putting their money where it's yielding the best results.

Jessica Gondolfo:
Yeah, that's sure. That's music to their ears. How much buy-in actually, when you are talking to these companies, do you still need to start from that education stage? Do you think the market is still in, you have to help them with understanding what these attribution models do, or do you feel like the market is somewhat understanding that they have to move into models like this?

Zeke Camusio:
Yeah, I love that question because there's one question that I asked during the sales process that essentially tells me whether that's an opportunity worth pursuing or not. What I ask, where does your return on investment data come from? And most of the time they say from the platforms or a variation of that, which could be my agencies provide that from what the platforms are saying. So my question is how much do you trust that if they say, oh yeah, we trust it, it's going to be very difficult for me to change their minds at that point. Because let's say that your bank account says that you have $10,000. If it's off by 40%, the way we normally find these raw numbers to be, it could be 6,000, it could be 14,000. Most people will not be okay with knowing that your balance is between six and 14,000.
You want to know exactly how much money you have in the account. Now, because these data counts directly from platforms, some people, I would say very few at this point, they get a face value. So if they say, no, we don't trust that we know it's wrong, then they're aware of the problem. So then what I ask at that point is, have you ever considered how much it's costing you to not solve this problem? Because we can do the math, we can run our attribution over their data and really show 'em how much they're wasting. So the lowest amount of waste we found in any program was 16%. So if 16% of your dollars are just going down the drain, that should be enough for most brands to say, that's not acceptable, because that would not be acceptable in any other area of the business.

Jessica Gondolfo:
That is such a great point. A lot of times it's not like, Hey, by proving value to you, I can show you what you could have that potential of, we could get you to be optimized or we could get you more efficient. It's that what is the cost of your business if you don't do this today? I feel like that is such a spot on thing. I feel it's very overlooked.

Zeke Camusio:
Absolutely. Yeah. It's in a lot of areas of life, and I'm going to get a little philosophical here, but you really need to consider what's the cost of not doing it? So yeah, maybe you don't want to exercise, but what's the cost of not exercising? So I think it's the same here where you sometimes, a lot of the clients that we come across, they have such great business models that in spite of 20% of their marketing budget going down the drain, they're still doing okay. So there's no immediate need to change that, but when we show 'em, Hey, you could be reinvesting this money elsewhere and getting this much additional revenue, it just makes sense, right? So there's more aware these days that this is a problem, and there's a lot of reasons for that. Google switching to Google Analytics for, as you mentioned, the landscape right now is really complex with a lot of different channels and campaigns and media types and users that can be tracked because of ad blockers and third party cookies being deprecated, G-D-P-R-C-C-P-A, iOS, I mean, I can go on and on, but the way I see it is that as a marketing leader, the most important thing you can do is make the right decisions and to make the right decisions, you need the right insights, and that's what we do.
We're providing those insights so they can be superstars and massively improved results without the need for additional budget.

Jessica Gondolfo:
That's super fair that it kind of leads me to this question about cookies going away at the end of the day, we've heard about it for years now, but they're at their end. So what does that mean for the models that you build? Is this a kind of, if you haven't started down this journey, you better start or how do you see that?

Zeke Camusio:
Yeah, so for context, for those people who haven't heard about this third party cookies are being deprecated this year, most browsers are just not going to support 'em anymore. So Safari already doesn't support him, Firefox doesn't support him, and Google Chrome, which is the number one browser in most parts of the world, it's going to stop supporting them. So what are third party cookies? So anytime you install a pixel or a snippet of code from a third party tool onto your own website, that's considered a third party cookie. So this could be the Google Analytics tracking code, it could be the Facebook pixel. So essentially there's all these technologies, ad blockers and privacy settings where you just disabled that and they're disabling that by default. So what it means is that any third party code that you have on your website is not going to be able to run.
So that means no information will be able to be sent to Google Analytics, Facebook ads, and so on and so forth. So the way to do it is to use first party cookies. So the difference between a third party cookie and a first party cookie is that a third party cookie is code that references a script on some other website or server. So it could be a Google Analytics, for example. A first party cookie is runs something in your own domain, and that's how they're being identified as first party or third party. So the right way to do it is for you to track everything as a first party cookie. First party cookies are never going to go away, or at least not for the foreseeable future because that's how the web works. For example, if you go to Gmail and you log in, you don't have to log in every time.
It just remembers that that's who you are. So the web relies on first party cookies and those are not going to go away. So there are two benefits of having first party cookies. The first one is you'll be able to track significantly more, and then you can stream that data to, in this case, Facebook ads and Google Analytics. That's the first value that they provide. You can track the data yourself and then stream it to whatever it needs to go. The second one is that it allows you to identify users and build a customer journey. So it helps you understand that you just saw this, clicked on this ad on Facebook, and then came to our website and sign up for emails from your iPad, and then the next day you opened an email from your iPhone signed up for SMS, the next day you came back and purchased something on the website.
We're talking about one person, three devices, five channels, but we're able to build a customer journey and see exactly what you've done, and it usually starts as a anonymous profile. So we know that it's a certain device ID and so far we don't know who you are, but we know that you've seen this page and you've clicked here and you've done this after all that at some point there's going to be a personable identifiable event I that's going to be you sign up for email or you buy something, you provide information about who you are. So based on that, then we're able to identify that profile that we had already been. So yeah, first party cookies are the way to go. If you're not already relying on first party cookies, you don't really have much time. You have a couple of months before all your tracking is useless, and I'm not being apocalyptic now, but you're going to notice that most of the data is not going to be able to be tracked anymore.

Jessica Gondolfo:
Yeah, I mean, you're talking about a millennial marketer here whose entire marketing journey was like cookies and cookie remarketing and all of this. So it's been a really interesting to talk to a lot of folks and understand what they're doing to deal with that challenge and what they're putting in place so that they can still have really strong attribution, really storing a lot of that historical data so that they can use it. So it's going to be interesting for sure, but I feel like having some kind of impact based attribution strategy will help you once these other channels really take away the cookie tracking. It will help you.

Zeke Camusio:
Yeah, absolutely. And I think it's important to explain that the impact based attribution doesn't rely on first party cookies either. So the reason for that is that even if we're able to track that you've clicked here and you've seen this ad and you've done this and then you purchased, we still don't know. We can't infer that if there were three touch points in that journey, we can't infer that each of those had an impact in your purchasing and equal impact. So first party tracking is really useful for the day-to-day operation, while impact based attribution is useful for allocating your marketing dollars the most optimal way possible.

Jessica Gondolfo:
That's awesome. That's awesome. You've been a super customer for four to five years. Yes. Get me honest on that one. And so you started with Google Sheets. Yeah, the Google Sheets, and then you moved into our BigQuery product. So I'd love to understand how does Supermetrics help?

Zeke Camusio:
Absolutely. So the foundation of our technology relies on getting data to us consistently and reliably. I have not found any ETL tool that has done that to the same extent that you guys do it. So we've tried maybe six or seven tools. We tried five Train, we've tried Air by Supermetrics. There's a couple more obscure ones that I can mention as well. But with Supermetrics, I was sold right away with, I started with Google Sheets and eventually that became two too constrained for what I needed. We needed to pull way more data than a spreadsheet could handle, and that's when we migrated to the Google BigQuery solution. And yeah, it's been, we have not looked back since then.

Jessica Gondolfo:
That's amazing. Is it fair to say that Supermetrics or enabling Supermetrics to run on the backend of your product and push all of that data in has enabled you to scale faster or be able to handle more customers?

Zeke Camusio:
Yes. We could not do this manually. There's just no way. In the same way that the way our machine learning works is every day. So we talked about the 50 different movie theaters across the us. So not only is it measuring what's happened in the past, but it's also making predictions about what's going to happen the next day. So that relies on data coming in constantly, so the algorithms can then measure the prediction against the actuals. That's only possible if you have tables that are updating automatically. There's no way we could have a person downloading data, exporting it, building our data. Pipelines are a hundred percent automated. We couldn't do it any other way.

Jessica Gondolfo:
That's interesting. And I want to touch back on the ETL thing because this is actually something we come up with a lot and it is people who have your level of data expertise. When they say, why supermetrics over another ETL provider, what would you tell them?

Zeke Camusio:
I'd say reliability is number one. I want my data to be there and I want the system to be able to recover from errors, and that's very common. If you're dealing with, in our case, we have 20 different connectors, something is going to go wrong, some transfer is not going to work, and for a number of reasons. Now is the technology good enough to figure out that something went wrong and run it again? So if it wasn't able to, for example, get the data for Google Analytics for yesterday, when it runs the next time, will it backfill that? So you guys do that extremely well. That's number one. Number two, I really like the ability to build my own queries and decide what to get from the APIs. With most of the tools we've tried, essentially they just connect to the API and dump all the data they have onto your data warehouse.
The challenge with that is that then you have to build a query to filter the data that you actually need. So you're not pulling this massive data set that is going to slow down your applications or your reports. So with Supermetrics, I'm able to say I want my, for example, for Google ads, I want to say I want my spend, my clicks, my conversion rate by campaign, by ad group by date, and that's all the data that's going to go to that table. And then I can create multiple ones for every data source if needed. And then once I get everything I need on my data warehouse, then we can create all the cleanup and data prep that goes prior to our machine learning models running.

Jessica Gondolfo:
That's awesome. I'm so happy that we can do that for you. I'm so happy you've been a longstanding client. We have quite a few people who use Supermetrics to fuel their product growth and their client growth through that. So it's just awesome to see a really great success story. Data speaks is such a cool product, and it's just such a hot topic right now. So many people are trying to understand how to build models like this, so I love learning about love learning about it. I have one other question for you, which is, if there's one other thing, things that you would make a suggestion about on the marketing landscape, if there was one trend that was really sticking out to you, maybe something you've seen across the way that people are spending or where influence is coming from, what would you say that one thing that you're starting to see is

Zeke Camusio:
In terms of marketing tactics or what marketers should be thinking about right now?

Zeke Camusio:
I'll give you two things. One that might seem like but is really relevant, so I can't end the call without sharing this with you, and then one that is a little different. So it is really critical to allocate your marketing budget the way it's going to provide the best results. There's nothing like it. There's no other activity that will have a greater impact on your bottom line. If you're able to get to 20 to 40% more revenue without spending more, there's nothing that will top that, right? So that's the most important thing. You really need to understand the return on investment for every dollar spent across every channel, every campaign. So let's move on to the next one. You should, I call this converting paid traffic into organic traffic. So what I mean by this is with most advertising, you are paying per click. So every time somebody clicks, they have to pay.
So let's say there's a path that has 10 different touch points. If they have to click, you have to pay for each of those clicks, three bucks, the $30 right there just for the clicks. Now you have your own channels, you have email, you have SMS, you have social media. So it's really, really important that as you are acquiring traffic to your website, that you're able to get in to sign up for your channels. That way they don't have to pay for every interaction anymore. So it's critical that you are testing different calls to action that you have. In most cases, we suggest having a first time customer discount code, and when you see these popups test everything, lifestyle pictures versus product pictures, probably 20% off versus 15% off. Try free shipping, try if you're in B2B, maybe a white paper or some video or webinar. You really need to be testing this all the time, and at any given time, you want to be producing a new call to action that will beat the existing one you have and do this systematically until you find the right one, because that's going to have such a huge impact if you're able to cut customer acquisition cost by converting pay traffic to organic traffic.

Jessica Gondolfo:
The next time that I am in there buying something for that 10% pop off, I will think of this conversation and think of you because it's very true. It's very, very true. Zeke, thank you.

Zeke Camusio:
One thing that I usually say is, because this can be a little intrusive, especially when you first come to a website, what you could do is when you close it, make sure that once it collapses, it stays a little box at the bottom of the website. It's not interfering with your experience as a website visitor, but you can expand it anytime you need it.

Jessica Gondolfo:
That is great, because that weird, if this goes away, will I get it back kind of mentality. That's such a great addition to the user experience, which I think is super important for that. Good call out on that one. Zeke, thank you so much for your time today. Thank you for sharing. I've learned so much on this call and I think that data speaks is doing such incredible work, so it was really exciting to talk to you.

Zeke Camusio:
Appreciate that. It was a pleasure for me as well.

Jessica Gondolfo:
Awesome. Thank you.