Summary

This was an interview with Brook Shepard, founder of the digital marketing agency Mason Interactive, about the relationship between data and creativity in marketing. Shepard argued that data does not hinder creativity but rather improves and drives it. He emphasised the critical importance of starting with specific, measurable goals and understanding the predictable conversion rates throughout a customer's journey, which he termed the "thermodynamic laws of business." He used case studies from e-commerce and higher education to illustrate how to scientifically test creative elements, measure the impact of brand awareness campaigns, and use data to inform creative decisions. He also discussed demographic trends affecting the education sector and offered advice on how to introduce a data-driven culture within an organisation by focusing on solutions and incremental goals.

 

Key Points

  • The most critical starting point for any marketing strategy is to have a specific, measurable goal, as a vague goal like "more" is not actionable.
  • Businesses had predictable conversion funnels, which Shepard called "thermodynamic laws," that could be mapped and understood, from initial impression to final sale or enrolment.
  • It was essential to designate a single "source of truth" (e.g., Shopify, Salesforce) for data, as metrics would always differ across various platforms like Meta and Google.
  • Top-of-funnel brand awareness activities, which might not show immediate sales, could be measured scientifically by their downstream impact on conversions in other channels.
  • The goal of creative testing was not just to identify the best-performing ad, but to deconstruct it by isolating variables (e.g., background colour, call to action) to understand why it was successful.
  • Objections to data, such as seasonality, could be overcome by analysing very large, long-term datasets or by conducting controlled geographic holdout tests.
  • To effectively introduce a data-driven approach in an organisation, one should connect the data to solutions and incremental progress toward goals, rather than simply presenting problems.

 

Transcript

Transcripts are auto-generated.

 

Kiran Kapur, Host (00:01):
Hello and welcome. Now there's a view in marketing that data and analytics don't gel with creativity. My guest today argues that data can definitely drive and improve creativity.

 Brook Shepard, Founder of Mason Interactive  (00:13):
If you're judging a Meta advertisement or a TV advertisement by how many sales it got you, that's like judging a fish for how far it climbed up the tree.

Kiran Kapur, Host (00:22):
And I'm delighted to welcome Brook Shepherd, founder of Mason, which is a digital marketing agency. Brook, could you tell us a little bit about Mason, and then we'll get into sort of data and why data and analytics helps creativity.

 Brook Shepard, Founder of Mason Interactive  (00:37):
Sure. Thank you. Mason is a 17-year-old company privately held in the United States. We have offices in Charlotte and here in Brooklyn and in LA. And we are a full service firm. So we are not just performance, we're not just creative. We're trying to marry data design and performance marketing together to drive exceptional results for our clients. So we're full service and that means that we're seeing we're buying media on Google and Meta and TV and we're doing SEO and AIO and we're doing lifecycle marketing and we're doing affiliate marketing. And we're really leading with creative, which is informed by the data to drive these results for our clients. And I think that kind of summarises us.

Kiran Kapur, Host (01:19):
That's brilliant. And actually, yes, thank you for a complete description of full service agency. So I think one of the questions I get most often with this is one, does being driven by data stop you being creative. And the other question is, what data should I be looking at? Where do I even start? So can I get you to tell me what data you think is important? I'm

 Brook Shepard, Founder of Mason Interactive  (01:43):
Going to answer that narrowly from my perspective. So my perspective is that the number one thing that clients that are successful have in common, and there's a lot of things. There's product market fit, there's institutional rigour, there's branding. But the number one thing is having a goal. And more is not a goal, to my mind. A goal to me is 30% increase in turnover year over year. Or I don't care how much we grow, I need the efficiency to get 10% better. Or I want to sell 10,000 units by this quarter. Those are goals. And those are data points that you can use to work from. And so the client gets to set their own goals and then we come up with a data plan that works for that. One of the ways I think about it, if that onset makes sense, is I sort of think about it like there's some thermodynamic laws here that we cannot escape.

(02:42):
If we're dealing with an institution of school, we're dealing with a law school. The number of enrollments is a function of how many people apply. And the number of people that applied is a function of how many people raised their hand and requested information saying, "Yeah, I'm interested in a campus tour or a brochure." And the number, we'll call that an RFI. The number of RFIs is a function of how many people are on the website. And the number of people on the website is a function of how many impressions are in markets. Some impressions are better than others. A guidance counselor's impression for a college talking to my son, we mentioned, we've doing our time before is applying to college. That's a better impression than an advertising impression on TV. But these are knowable numbers. The number of impressions, we know what percentage of impressions turn into sessions on a website.

(03:28):
We know what percentage of sessions turn into RFIs. We know what percentage of RFIs turn into applications. We know a percentage of applications turn into enrollments. And so knowing that at a baseline before we even get to the creative part is a way to drive strategic decisions across an organisation. And you can do the same with e-commerce. Many people who buy a suit is a function of how many people who added to cart is a function. You get it. So I start with that data. And I think those things are absolutely key. We have something like 60 clients and you'd be surprised how many people don't start from those precepts of here's what we know, here's the data that we have, here's our goals. How many impressions do we need in market to drive the business outcome that we want downstream? That's before even talking about creativity, but you asked what are the most important things.

(04:22):
I think the most important thing is setting a goal and then knowing what your ecosystem is so you can follow those laws of thermodynamic business.

Kiran Kapur, Host (04:31):
You've made that sound very easy. I have to say. That's great.

 Brook Shepard, Founder of Mason Interactive  (04:34):
That's my job.

Kiran Kapur, Host (04:35):
Yeah. Okay. So I can come up with my goal and what I noticed about you saying more is not a goal. What you actually gave me was sort of smart objectives. They were goals that I want to be 30% increased over this period of time or whatever. So that's great. I'm now making very specific goals. And then you said cheerfully, there's thermodynamics of the business, which is a lovely phrase and I'm going to borrow that. Great. And then you said, "But we know this stuff." And I'm sitting here thinking, "I bet you have clients that don't even know where to get that information from before they've done anything else."

 Brook Shepard, Founder of Mason Interactive  (05:08):
That's true generally. However, it's different now than it was five years ago. People know more now. Generally, clients have to decide what their source of truth is. If you're an e-com business, that's almost always going to be Shopify. It could be your order fulfilment system. It could be Google Analytics, GA4. If you're an institution of higher education, it's probably going to be Slate or Salesforce. So you have to decide what your source of truth is. And you're right that those sources of truth will never agree with one another. What Facebook says you got in sales is not what Google says you got in sales, is not what Shopify says you got in sales.

Kiran Kapur, Host (05:44):
Cool.

 Brook Shepard, Founder of Mason Interactive  (05:44):
So somewhere you need to decide, and someone on the institution side needs to say, "This is our source of truth. This is our Bible and this is what we're going for.

Kiran Kapur, Host (05:53):
So I've selected one of those. Am I going to know all that data? Because one of the other questions I was going to come onto, but we can look at now, is there is a view in marketing. If I can't justify any other way, I say it's brand awareness. So brand awareness will cover a multitude of things if I want to do things. But you're actually telling me, no, my brand awareness is going to drive the number of people that do this, which is going to drive the number of people that do that.

 Brook Shepard, Founder of Mason Interactive  (06:17):
Oh yeah. Yes. So I had a client, and they know who they are, and if they're listening to this, it's more than one client, but I'm talking about a specific client now that said three years ago, I want to turn off all of our Facebook advertising because it's not driving enrollments. And I said, "Don't do that. " I probably said it like that too, "Don't do that. You can't Google something if you don't know what it is. " And they said, "No, we're going to turn it off. Our source of truth, Salesforce said it's not driving incremental enrollments. We're turning it off." And I said, "Okay." And then that institution, they had an enrollment challenge fairly quickly. And so they knew that and different people at the organisation knew that differently. Some people on the marketing team knew it intuitively, some of the finance people did not know it and did not understand it.

(07:07):
And so what we did is I designed a test for proving the efficacy of the lift of top of the funnel advertising brand awareness, specifically in the case of Meta for this school. And I'm aware of the fact that when I talk to a business or a prospect and I talk about spending at the top of the funnel, I'm aware of the fact that I sound like an ad salesman, which guilty is charged, but I also, I know what I'm doing and I've spent a billion dollars of client's money and I know how this stuff works. So I hired a data scientist from an Ivy League institution here in the United States who didn't rubber stamp, but improved and fixed and then agreed with my conclusions. And we're able to see that 30% incremental spend at the top of the funnel on Instagram does drive 40% increase in downstream conversions in Google.

(07:56):
So knowing that data, we know this for scientific certainty. So knowing those data points and developing consensus across an institution on those data points and making sure that no one thinks you're talking about witchcraft is a big deal and a hard thing to do. But we know that because we're not just wantonly saying, "Let's spend money at the top of the funnel." And the way we talk about it now is we, and I've stolen this from someone and then changed it a little bit and hopefully Trendsmorgified it. But it's like if you're judging a Meta advertisement or a TV advertisement by how many sales it got you, that's like judging a fish for how far it climbed up the tree.That's not what the fish is supposed to do. Judge the fish by how far it slims downstream, judge the squirrel by how fast and far it climbs up the tree, but the Instagram's job is generally, the Instagram advertising's job is generally not to drive sales.

(08:46):
It's generally to drive awareness. And I know that people have built big businesses on Instagram sales, and I'm not saying it doesn't happen, but generally its job is to introduce people to things so that they can convert on it later through remarketing or through Google.

Kiran Kapur, Host (08:58):
And also it depends what you're selling. So if you're selling a chocolate bar, maybe, I don't know, that's probably a bad example, but Instagram might drive something. Whereas if you are selling an education and you're driving people in September or two years time, clearly Instagram, it's going to take a lot longer. You're trying to just raise awareness. So again, I see what you mean about horses for courses. Okay. We may come back a little bit to brand awareness. Could I ask you to take me through one of the case studies on your website, because I thought it really brought to life how data can help you with creativity. And that was your, I think it's a coffee roaster, which I will pronounce very Britishly as taste. Well, I don't know if that's correct.

 Brook Shepard, Founder of Mason Interactive  (09:42):
That's correct. Yes. Okay.

(09:46):
Yeah. Taste coffee roasters are a independently owned, sustainable coffee brand. So most of us have Kurag cups or K-cups. Those things go right in the trash or we have an espresso machines. Those things go right in the landfill. Taste pods are biodegradable. So they're roasting all their coffee and now for different tariff reasons they're doing it in the United States and their K-cups are compostable. So what we did with them was, and the market's changed as we did this case study, but what we did with them was, Meta will tell you what ad is the best. Meta will tell you that. You run an ad on Instagram and Facebook through Meta. It's going to tell you what ad. If 20 ads up, it'll tell you what ad is the best. It won't tell you why it's the best. So you therefore can't make more ads like that to drive future success.

(10:35):
You don't know what about it was what drove success. So if you have a client who's willing and has the intestinal fortitude like Taste does, it's a good idea to set aside a portion of the budget and say, "I'm just going to test creative." And so I'm going to take ... I know what the best ads are. We all know that. You can get them right out of Motion or Instagram. You can find out. I'll find out what the best ads are and then I'm not going to pretend that I know why they're the best ads and I'm going to deconstruct them. So I'm going to take the best ad, and I'm going to make three variations of it or 12 or two, but in this case it was three because of the budget. I'm going to make three variations that isolate variables. I'm going to do just the background colour because you have a brand guideline that says you can use green, black, and mocha.

(11:17):
Let's see if the ads are identical for everything except for the green, black, and the mocha backgrounds wish converts better. Okay, cool. It's the mocha. Great. Actually, it was the green. "Hey, this is the green. That's great. Let's move forward." So now you know which background colour works best. And then you can test calls to action. Is it add to cart? Is it save the planet? Is it drink better coffee? You're going to try all three of those with the best background colour and now you're going to know what the best call to action. So now you know what the best ... You already knew what the best ad was because Meta told you that, but now you know what background colour is the best, and you know what call to action is the best. And you can keep iterating in that and you can try things like, does a product shot work better than an action shot?

(11:57):
Does a light animation work better than static? And the answer is always going to be yes to that question. But you can go through these series of iterative tests and at the end of it, you'll know that it wasn't just ... Now you know why it's the best ad. It's because the background colour is this one, it's because the call to action is that one, it's because the product shot is this one and you mentioned sustainability this much times. Now that you know those things with data, the creative team is free to go back and iterate on those and be creative because their decisions are informed by that data. This is what actually works, do more of it. It doesn't mean we shouldn't keep testing and trying new things, but the new round of creative is informed by the data of the previous three or four rounds of testing.

(12:34):
And that's changed. The methodology changes a little bit all the time with the way Meta changes, but the idea of isolating for variables to use data to decide what is the best piece of creative is still valid.

Kiran Kapur, Host (12:47):
And what happens when somebody says, "Oh, but it's seasonal." We get that one a lot. Well, the reason it didn't pull so well is it's a seasonal advert or there's something else that's happened or you can get ... It's very easy for people to then argue against it by saying, well, but there was another variable that affected things.

 Brook Shepard, Founder of Mason Interactive  (13:08):
Yeah, people love that. I have two things to that. One is if I ask any of my account managers here at Mason, there's like 35 of us. If I ask any of them what the best ad was for a client yesterday

(13:23):
Or this week, most of them are going to know if they're at all competent, the answer to that. Most of them here are going to be able to say, "Yes, it was the one with the blue widget." What they're not going to be able to do is tell you over the last three or four, there's two parts of this. The first part is they're not going to be able to tell you over a longer date range, what the best ad was. AI can help with that. AI can help with taking a look at thousands of ads at once over the course of years and saying that regardless of seasonality, regardless of sale period, regardless of skew, regardless of price point, what do the best ads have in common? And you'll find things like for a Saville Row food incline we work with, you'll find that it's three quarter length shot and you need to be able to see a little bit above the guy's head down to his knees and he's a three quarter profile and the background is light.

(14:10):
And you wouldn't know that if you're looking at one data set, you would know that if you look at years worth of data. S-o then I have erased seasonality and sales and promotions by looking at what worked well over the course of huge data sets over the course of many years. That takes some doing, but it's not rocket science. That's the first thing about dealing with seasonality. The second thing is, I mentioned the test we did to prove the efficacy of top of the funnel Meta spend for an educational institution. We did that with a holdout. Education is a seasonal business. There are usually enrollments twice a year. Some schools have rolling admissions. Some schools are once a year, but generally there's seasonality and it's always Q1 because you go home and your brother's a doctor and you're not and your mom gets mad at you and then you apply to college.

(14:50):
I've been doing this a long time. That's how it works.

Kiran Kapur, Host (14:52):
Brilliant.

 Brook Shepard, Founder of Mason Interactive  (14:54):
There's always an increase in applications to schools in Q1, in early Q1. And everyone looks really smart and higher ed in Q1 in the marketing teams. But so the way we designed the test for this other school, and we'll talk about it in more detail, not today, but as I get the permission, I released the case study, is we did a geographic holdout test. So we ran two cities in the United States, Charlotte and Austin, Texas, Charlotte, North Carolina and Austin, Texas are both about the same. They're both just over a million people. They're both what we call purple states. They're Republican and Democratic. The cities tend to be a little bluer than the red or surrounding areas. And we ran no top of the funnel advertising, just bottom of the funnel, high intent, keyword search demand on both of those markets for five months to see how they performed.

(15:42):
And then we added top of the funnel advertising to just one of them to see how they performed. So we erased seasonality by doing the geographic holdout because it wasn't seasonality. We know the other one went down and the one we introduced the top of the funnel to went up. So we were able to prove that it wasn't seasonality. Now, could someone say that the seasons are different in Charlotte versus Austin and Austin's a little hotter than Charlotte? Sure. But we did that test three different ways in six different markets and the data kept being the same. So eventually, that's how we dealt with the pushback of the seasonality. One is you can look back over a huge time using AI, and two is you can do a rigorously designed holdout test that will prove that things are not entirely beholden to seasonality.

Kiran Kapur, Host (16:28):
No, you grinned, as I said, people can challenge the data. I think the two elements you get with anything to do with data-driven and marketing is one, but I can't use data, I'm creative, or two, but we can always challenge the data. And sometimes it's like you can't win, but we do need data. Somewhere along the line, we have to come down to data. One of the things I love that you said was you need to know why something outperforms. So this is presumably where the judgement comes in because you need to know what to test next or instinctively you can look at the two and go, "This is why that one's outperforming."

 Brook Shepard, Founder of Mason Interactive  (17:09):
I think it depends on the client. They get to decide how instinctive they want to be. I think it's good to know why some performs so you can do more of it. And I think generally the creative team likes to know too. The creative team does not like to be told up front. No one likes to be told in life that this is what you're going to do next. No one wants to hear that in life. And the creative team especially is resistant to hearing up front. I have an inkling that dark and sexy is going to work better than cool and smooth. And I would too, if I had that Joe Jobs, but running it after the fact and saying, "This is the one that worked better." I mean, here's the commonalities it has across our entire client portfolio and this client's advertising over the last three years.

(17:49):
They're generally receptive to that. And if your creative team isn't receptive to that, you should get a new creative team.

Kiran Kapur, Host (17:55):
Love it. One of the other things that, again, your website gives us as a case study is scalable growth for mission-driven organisations. And as you said earlier on, that mission-driven organisations was an area that you really enjoyed working in. So again, I think this is an area that most people don't think you have to worry about data and analytics because they're mission-driven.

 Brook Shepard, Founder of Mason Interactive  (18:19):
Yeah. By mission-driven, I mean schools mostly. It's about half of our business. There is something on our website about a client called Hate Ends Now, which is a Holocaust. We call it a school business because their job is to educate people and they do it in schools. They're not a school, but schools hire them to come in and educate children about the Holocaust, which is harder because when I was a kid growing up, we would have Holocaust survivors come into class and that was a long time ago. So it's interesting. One of the things that's happening in the space here, and I don't know if this is happening this way in the UK, is since 2008, we had a recession. There is something like a million fewer people born every year in the United States. Since 2008, it's been, I don't know, 16 years, something like that.

(19:05):
If you look at the preceding time period, equal time period going back before 2008, we average about a million people being born, fewer people being born each year since then.

Kiran Kapur, Host (19:14):
Gosh.

 Brook Shepard, Founder of Mason Interactive  (19:15):
So there are fewer people in, and that means the people that are applying to college now, 18 years ago, there's just fewer of them.

(19:22):
This leads to a problem. If you're an institution that has growth targets and your mission is to educate people to have a better society, because you want everyone around you to be as smart as possible to deal with the challenges of AI and geopolitics. The obvious solution to this, if there's fewer people being born, is to get more immigrants into your country that'll come to your colleges.This is the obvious logical solution to this, and it's historically what my country has done. In the current political climate, it seems like that's a little askew. And so you may have just detected some of my politics there, but knowing these things ... I was on stage at Google 10 years ago talking about this, and we call it the enrollment cliff. There's just a lot fewer people in market. So knowing that and understanding that is a key thing to understand if you're marketing any sort of mission-driven school that's trying to make people and society around you better.

(20:11):
The second thing is that behaviours have changed. And I don't know if it's societal or if it's COVID or what it is. I think it dovetails with COVID. Certainly, the way I built this business, and I've worked with growing a lot of schools, is to, I mentioned earlier, get more people to fill out an RFI so that they can enrol. Children are skipping the RFI part of it and going straight to enrol more than ever before. So the percentage, we still know the percentage at which requests for information, RFIs turn into apps, turn into enrollments. We still know those things and they still exist. The data exists. But it used to be this percentage of people went from an RFI to an app. It's now this percentage. Instead, people are going right to the app portal for institutions, which again, sort of makes sense when I think about my son's applying to college over the last couple of years, my son's application process, he was applying directly to the colleges through a common app or I forget what the British system is.

(21:04):
We talked off camera about he's applied to British schools. He went directly to that system and applied. He didn't do a campus tour of Goldsmiths or Queen Mary. He just applied directly. So knowing that data is key and understanding it and understanding where you're at, is key because if you're not informing your decisions on that data, you're going to be very confused. And by the way, I think a lot of people are really confused. I spoke to an institution in the middle of this country, a top ranked school. I won't tell you who they are. They came to me and they said, "We really like your portfolio. Our new marketing director worked with one of your senior people at a different institution." They said, "We need to increase our business school enrollment and we don't want individual people to apply." It's a certificate programme, I should say.

(21:51):
It's not a master's or a PhD. It's a certificate programme, a career enhancement programme, a prestigious university. And we don't want people like you to apply, bro. We want people like entire companies to apply because the average order value of getting a hundred people or 200 people or every mid-manager or Coca-Cola or every manager at Masonter active, the average order value of that is so much higher than just Brook Shepherd coming in and applying for his one class with $3,000. That's what we want. And they said, "And we need to show results immediately." And I said, "That doesn't make any sense." Because to sell into a larger organisation is going to take a multi-pronged approach and you're going to have to be thinking about your email marketing and your TV and your mailers and you go all that together. And then once you sell into the HR department of a big company, there's probably a six month sales cycle of finding out what they want to know, and then there's developing the cohort.

(22:43):
It's probably going to take a year to get your first cohort in. It might take six months, it might take year and a half. It's going to take about a year.

(22:50):
That does not square with showing instant results to please the board. My sales team and I knew this, we didn't contradict the client on the prospect on the call, we agreed to them. And then we had a separate plan approved. What would happen if they only went after individuals, which will show instant progress because I know how long it takes to enrol a mid-career professional in a career enhancement programme at a prestigious university. I've done it a million times. I know how long it takes. So we went in with, even though they only asked us for the first one, big cohort, we went in with big cohort and then also we had some hidden slides in the deck in case they brought it up about going after individuals. Wouldn't you know lo and behold, Kiran, halfway through the meeting, someone new pops into the meeting and says, "Hi, I'm the boss and the people you've been speaking to.

(23:33):
Sorry, I'm late to the Zoom. None of this makes any sense. We need to only talk. We need to show results now. I need individuals." We said, "Great. We've got that plan right here in our back pocket." We unhid the slides and we went through it. My point is that they at that institution were so conflicted about their own goals, which you came back to in the beginning, started in the beginning. They couldn't articulate their goals at their four-person marketing team for this part of the college of this university. And therefore they're going to have a hard time being successful because the data needed to do the creative to reach the mid-career professionals or the HR professionals that will get the mid-career professionals and mass, has to be driven by a specific goal. And if you can't at your organisation agree on what your goal is, I would submit to you that you're going to have a hard time driving effective advertising campaigns.

Kiran Kapur, Host (24:23):
Thank you. I think we're coming towards the end of the questions, but I just wanted to ask, if you were somebody sitting in an organisation now, and a lot of our learners are apprentices, a lot of our listeners are a sort of early stage career, how do you introduce this straight data-driven to your organisation? Because you can't go up to your boss going, "By the way, we don't seem to have any data and I think we should have. " Or maybe you can.

 Brook Shepard, Founder of Mason Interactive  (24:46):
My answer to that, I have a good friend, he was in charge of the data for this institution and he was very ... The sky has fallen. All of his data was, "We're not going to hit this enrollment target. We don't have enough leads." And I would say to him, "You can't do that. You can't go into the organisation. You can't go into your boss's office and just say, the sky is falling, even if you're right that it is and no one else knows it. " No one wants to hear that. You need to be giving solutions to fix it. And this person said, "Well, that's not my job to fix it. My job is to analyse the data." And I said, "They're going to fire you. " No one wants anyone an institution that is just complaining. It's not useful even if you're in command of the data.

(25:27):
And he did get fired, and he learned that lesson and now he has another job where he's in charge of a much more complex organisation, hitting their goals and crushing it. And that wasn't the only lesson he learned. We all get better and I've learned a lot of lessons too. But if I were going to introduce data to my company, I would be using it to talk about creating incremental revenue or turnover, as you guys say, about incremental revenue and success towards goals. I would not be getting lost in a spreadsheet forever because no one wants to look at spreadsheets forever. I would be connecting the data to reaching incremental goals. That could be, I've been taking a look at our cohort and it turns out that 30 to 35 year olds do better than 35 to 40 year olds. That's the data. I want the marketing to reflect those user personas.

(26:07):
Let's do creative like that. That's useful. I found out that our ads tend to do better on weekdays. We tend to sell a lot more on weekends than we do on weekdays, weekdays, our fallow periods. I want to try the idea of introducing a sale just during our very slowest period, which is Tuesday at three o'clock/sale. That's useful. That might not be a good idea, but that might be a useful piece of data. So connecting the data that you have, which by the way, we all have access to Claude and ChatGPT and Google's Notebook LM. We all have access to the same tools. We can all pop these things in depending on your company's privacy restrictions and take a look at this data and be in command of it.

Kiran Kapur, Host (26:44):
That is a brilliant place to end. Brook Shepherd, founder of Mason, thank you so much. That was really insightful and I thoroughly enjoyed it. And I feel like I've learned quite a lot as well. Thank you very much indeed.

 Brook Shepard, Founder of Mason Interactive  (26:57):
Thank you for having me.