Let’s connect
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Evan Harp sat down with Ozzie Solares, Global Head of Media and Analytics at J.P. Morgan Asset & Wealth Management, to talk about attribution.
Evan Harp: Attribution is critical for measuring success as it can tell the story of how media or marketing spend can directly affect sales. Do you feel that asset manager firms, by and large, are good at attribution measurement?
Ozzie Solares: I will say, based on my experiences, that attribution is a challenging space for the industry. Asset managers want the confidence to link media spend directly to sales, but the reality of the tech and data landscape and what's actually possible presents significant hurdles. Every solution requires some level of compromise and by definition that means accepting something less than perfect. Marketers need to have a clear understanding of what their organization’s needs are to know where and what to compromise.
Harp: What are the problems attribution solves?
Solares: Ultimately, the problem attribution solves is an understanding of the client or audience journey and how those touchpoints influence revenue. As marketers we use that information to make decisions and to justify marketing investment. Attaching ourselves to driving real, measurable revenue makes for an easier narrative in C-suite forums.
Without it, decision making, marketing or finance-related becomes too subjective. Our challenge as marketers is driving the best possible outcome within a set of constraints. Budgets are one of the biggest constraints we as marketers contend with. The second, is an appreciation for what marketing delivers as a compliment to sales (especially in sales-led organizations).
When you start to delve into those details, you realize that while the technology landscape isn't always doing us any favors and continues to evolve, staying on top of attribution is key to successful revenue generation and efficient spending.
Harp: Attribution is about data, and something like 80% of the data we use today wasn’t available even a few years ago. I would think that more data would make attribution easier, but it looks like it doesn’t translate when it comes to marketing attribution.
Solares: More data doesn't necessarily mean easier attribution, I actually think it has created a different issue which is the ability for our organizations to deal with the onslaught of this new data. Ironically, even though the ability to track user behaviors has gotten better, data privacy laws have become more stringent and that limits the ability to even collect it. Walled-gardens like Amazon, Meta, or Google only further complicate things. That is forcing us to evolve toward new measurement solutions that rely more heavily on aggregated data, but again , there are trade-offs. The good news is that with so much money on the line for advertisers of all kinds, there no shortage of vendors trying solve and fill this gap.
In the end, it's not about gathering all the data; it's about identifying the right data to solve a specific business problem first and then building a data model that compliments that need.
Harp: Let's talk a little bit through the customer journey. There are many types of attribution methods, between last touch and multi-touch. Can you talk about when to use a specific attribution approach?
Solares: Yeah. I think it's important to note that, it's probably not an either-or conversation. As you mature, you're going to find yourself using a combination of models. It's really important to understand what questions you're specifically trying to answer because each one of those models lends itself to a different use case, and potentially a different opportunity for you as a marketer. Generally speaking I recommend using market mix models (MMM) to guide investment allocations across channels and to measure ROI, while leaning on MTA solutions to understand the customer journey (and the friction points therein) to improve your ROI. Last touch isn’t bad so long as you understand what it does and doesn’t tell you, it’s probably best used when optimizing within channels versus across channels.
Harp: You’ve also had experience working with marketing mix models - how is that similar or different from an attribution model?
Solares: First off, they are both attribution models. MMM relies on the use of aggregated metrics in some kind of time-series. The advantages it provides is that it doesn’t rely on PII, it can account for 3rd party variables (like market conditions), and it accounts for the halo effects that direct attribution simply can’t. With multi-touch attribution, what you're usually trying to figure out is the fractional value of an individual touch point through direct attribution. DOne correctly it explains the sequence of customer interactions and the fractional value of each step in that journey. The challenge with MTA models is that you need proper data collection on both impression and post-click side. This is where data collection and data privacy laws have the biggest negative impact. This is also what separates good MTA solutions from bad ones. If they can’t account for offline interactions or limit data collection to owned-channels, look elsewhere because you aren’t getting an accurate picture.
Harp: Is it common for companies to use MTA and MMM in tandem, or do they tend to use just one or the other?
Solares: I can't speak about common, but I can tell you, from my experience on the media agency side serving multiple clients that there is value in having both available is ideal. If I did have to choose, I would go with MMM. Ironically, I was having a conversation with someone at a very prominent social media platform recently and they mentioned that they have pivoted to MMM more and more. When you consider how much data this platform collects and has access to, that says a lot.
Harp: How do you get everyone in an organization onboard? Where do you start? How can you navigate these leadership conversations across an organization?
Solares: As marketers, we all know that we have to get in the minds of our audiences to make an impact.
When engaging with sales, they're self-interested in the sense that they want to be able to drive revenue because usually they're compensated based on that. But the key to delivering your narrative is remembering that marketing is a partner to sales. When you're talking about things like ROI or contribution, just remember that. Because it can result in sometimes a defensive reaction if you suggest that marketing alone is the one driving revenue. Best to keep that in mind as you bundle your narrative to them.
I think, on the CEO and leadership side, at the end of the day, they want what's best for the business. And the decisions they're trying to make are, if I give you resources, it means I can't give somebody else resources. So they're trying to decide “where am I going to get the most bang for my buck or where am I going to get the best return within a certain timeframe?”
For the CFOs particularly, by nature, they're more analytically inclined when it comes to investment decisions. When you start discussing ROI models, expect more detailed questions. They are the people who I have quite literally had to open up the hood and show them all my spreadsheets. They're asking about what coefficient I use, what's the statistical significance of that variable? What are all the components going into your models? It's all coming from a good place when we interact because I think they're genuinely interested - especially when you show them advertising and marketing is both an art and science. I think some of them don't always realize that at first.
If you can feed into that curiosity by coming prepared to show them the details, I think you'll be better off.
Harp: Attribution models are rarely a “set it and forget it” type of thing. It seems they evolve with new data, business goals, and organizational demands. How do you manage that constant evaluation?
Solares: In my own experience, spending a lot of time on the foundation pays dividends down the road. It's probably not the sexiest part of the exercise, and at times it can feel laborious, but taking time to really set up your infrastructure based on the customer journey is critical.
It’s important to be honest about where your data foundations are strong, what resources you actually have internally or externally, and your own comfort level with the subject matter. Approaching attribution with humility and realism sets you up to build practices that are not only credible, but able to evolve meaningfully over time.
Harp: Before we wrap this up, is there anything else people should know as they approach attribution?
Solares: It's a natural desire to have a one-size-fits-all approach. But it might not be possible. For example, when you look at financial data, that varies greatly from market to market. The US has certain laws around privacy and transparency, and the EU has different laws. Whether or not your business is B2B or B2C also changes the nature of what you need to measure and the time cycle you are looking at.
I talked about the importance of building a strong foundation before, and I want to underline that. The more you invest in having a solid foundation with alignment between departments on what data matters, the better you will be able to build your attribution model to tell your story accurately.
Evan Harp sat down with Ozzie Solares, Global Head of Media and Analytics at J.P. Morgan Asset & Wealth Management, to talk about attribution.
Evan Harp: Attribution is critical for measuring success as it can tell the story of how media or marketing spend can directly affect sales. Do you feel that asset manager firms, by and large, are good at attribution measurement?
Ozzie Solares: I will say, based on my experiences, that attribution is a challenging space for the industry. Asset managers want the confidence to link media spend directly to sales, but the reality of the tech and data landscape and what's actually possible presents significant hurdles. Every solution requires some level of compromise and by definition that means accepting something less than perfect. Marketers need to have a clear understanding of what their organization’s needs are to know where and what to compromise.
Harp: What are the problems attribution solves?
Solares: Ultimately, the problem attribution solves is an understanding of the client or audience journey and how those touchpoints influence revenue. As marketers we use that information to make decisions and to justify marketing investment. Attaching ourselves to driving real, measurable revenue makes for an easier narrative in C-suite forums.
Without it, decision making, marketing or finance-related becomes too subjective. Our challenge as marketers is driving the best possible outcome within a set of constraints. Budgets are one of the biggest constraints we as marketers contend with. The second, is an appreciation for what marketing delivers as a compliment to sales (especially in sales-led organizations).
When you start to delve into those details, you realize that while the technology landscape isn't always doing us any favors and continues to evolve, staying on top of attribution is key to successful revenue generation and efficient spending.
Harp: Attribution is about data, and something like 80% of the data we use today wasn’t available even a few years ago. I would think that more data would make attribution easier, but it looks like it doesn’t translate when it comes to marketing attribution.
Solares: More data doesn't necessarily mean easier attribution, I actually think it has created a different issue which is the ability for our organizations to deal with the onslaught of this new data. Ironically, even though the ability to track user behaviors has gotten better, data privacy laws have become more stringent and that limits the ability to even collect it. Walled-gardens like Amazon, Meta, or Google only further complicate things. That is forcing us to evolve toward new measurement solutions that rely more heavily on aggregated data, but again , there are trade-offs. The good news is that with so much money on the line for advertisers of all kinds, there no shortage of vendors trying solve and fill this gap.
In the end, it's not about gathering all the data; it's about identifying the right data to solve a specific business problem first and then building a data model that compliments that need.
Harp: Let's talk a little bit through the customer journey. There are many types of attribution methods, between last touch and multi-touch. Can you talk about when to use a specific attribution approach?
Solares: Yeah. I think it's important to note that, it's probably not an either-or conversation. As you mature, you're going to find yourself using a combination of models. It's really important to understand what questions you're specifically trying to answer because each one of those models lends itself to a different use case, and potentially a different opportunity for you as a marketer. Generally speaking I recommend using market mix models (MMM) to guide investment allocations across channels and to measure ROI, while leaning on MTA solutions to understand the customer journey (and the friction points therein) to improve your ROI. Last touch isn’t bad so long as you understand what it does and doesn’t tell you, it’s probably best used when optimizing within channels versus across channels.
Harp: You’ve also had experience working with marketing mix models - how is that similar or different from an attribution model?
Solares: First off, they are both attribution models. MMM relies on the use of aggregated metrics in some kind of time-series. The advantages it provides is that it doesn’t rely on PII, it can account for 3rd party variables (like market conditions), and it accounts for the halo effects that direct attribution simply can’t. With multi-touch attribution, what you're usually trying to figure out is the fractional value of an individual touch point through direct attribution. DOne correctly it explains the sequence of customer interactions and the fractional value of each step in that journey. The challenge with MTA models is that you need proper data collection on both impression and post-click side. This is where data collection and data privacy laws have the biggest negative impact. This is also what separates good MTA solutions from bad ones. If they can’t account for offline interactions or limit data collection to owned-channels, look elsewhere because you aren’t getting an accurate picture.
Harp: Is it common for companies to use MTA and MMM in tandem, or do they tend to use just one or the other?
Solares: I can't speak about common, but I can tell you, from my experience on the media agency side serving multiple clients that there is value in having both available is ideal. If I did have to choose, I would go with MMM. Ironically, I was having a conversation with someone at a very prominent social media platform recently and they mentioned that they have pivoted to MMM more and more. When you consider how much data this platform collects and has access to, that says a lot.
Harp: How do you get everyone in an organization onboard? Where do you start? How can you navigate these leadership conversations across an organization?
Solares: As marketers, we all know that we have to get in the minds of our audiences to make an impact.
When engaging with sales, they're self-interested in the sense that they want to be able to drive revenue because usually they're compensated based on that. But the key to delivering your narrative is remembering that marketing is a partner to sales. When you're talking about things like ROI or contribution, just remember that. Because it can result in sometimes a defensive reaction if you suggest that marketing alone is the one driving revenue. Best to keep that in mind as you bundle your narrative to them.
I think, on the CEO and leadership side, at the end of the day, they want what's best for the business. And the decisions they're trying to make are, if I give you resources, it means I can't give somebody else resources. So they're trying to decide “where am I going to get the most bang for my buck or where am I going to get the best return within a certain timeframe?”
For the CFOs particularly, by nature, they're more analytically inclined when it comes to investment decisions. When you start discussing ROI models, expect more detailed questions. They are the people who I have quite literally had to open up the hood and show them all my spreadsheets. They're asking about what coefficient I use, what's the statistical significance of that variable? What are all the components going into your models? It's all coming from a good place when we interact because I think they're genuinely interested - especially when you show them advertising and marketing is both an art and science. I think some of them don't always realize that at first.
If you can feed into that curiosity by coming prepared to show them the details, I think you'll be better off.
Harp: Attribution models are rarely a “set it and forget it” type of thing. It seems they evolve with new data, business goals, and organizational demands. How do you manage that constant evaluation?
Solares: In my own experience, spending a lot of time on the foundation pays dividends down the road. It's probably not the sexiest part of the exercise, and at times it can feel laborious, but taking time to really set up your infrastructure based on the customer journey is critical.
It’s important to be honest about where your data foundations are strong, what resources you actually have internally or externally, and your own comfort level with the subject matter. Approaching attribution with humility and realism sets you up to build practices that are not only credible, but able to evolve meaningfully over time.
Harp: Before we wrap this up, is there anything else people should know as they approach attribution?
Solares: It's a natural desire to have a one-size-fits-all approach. But it might not be possible. For example, when you look at financial data, that varies greatly from market to market. The US has certain laws around privacy and transparency, and the EU has different laws. Whether or not your business is B2B or B2C also changes the nature of what you need to measure and the time cycle you are looking at.
I talked about the importance of building a strong foundation before, and I want to underline that. The more you invest in having a solid foundation with alignment between departments on what data matters, the better you will be able to build your attribution model to tell your story accurately.
