Sep 2, 2022
As modern marketers, we are swamped by numbers coming from all directions. How do we manage all this data, demystify it, and use it to our best advantage? Host Will Francis chats with Chris Penn to look at data from different and interesting angles. They talk about Regression analysis, using R and Python, GA4 and GDPR, along with lots of interesting real world metaphors! Chris also offers 5 key takeaways that any marketer can start using today.
Chris is the co-founder and chief data scientist at Trust Insights and co-host of the super popular podcast Marketing Over Coffee. A recognized thought leader and speaker he is a best-selling author of a dozen books, and has shaped five key fields: Google Analytics adoption, data-driven marketing and PR, modern email marketing, marketing data science, and artificial intelligence/machine learning in marketing.
They mention the classic Bob Stone book Successful Direct Marketing Methods with its foundational principle of Audience, Offer, Creative.
And if you enjoyed this episode please leave a review so others can find us!
“If you rolled out a new website, and you forgot to install Tag Manager on it, and you have three weeks of just missing data, your data quality is off. So, your pantry may look full, but the ingredients may or may not be out of date, they may be spoiled” Chris Penn, Trust Insights
Will: Welcome to "Ahead of the Game," a podcast brought to you by the Digital Marketing Institute. I'm your host, Will Francis, and today I'll be talking to Chris Penn, all about data and insights in digital marketing. How we can use data more effectively, generate insights that are truly actionable, and ultimately, turn data into better marketing. Listen to the end to get Chris's top five tips on becoming a more data-powered marketer. Chris is Chief Data Scientist of Trust Insights, and co-host of the "Marketing Over Coffee Marketing Podcast." He's a 2019 IBM champion in IBM Analytics, a Brand24 top 100 digital marketer, and author of over two dozen books around marketing, technology, and business. Chris, welcome to the podcast.
Chris: Thank you for having me.
Will: It is a pleasure, and I think that it's a very useful topic that I'm about to interrogate you about, if that's okay.
Chris: Of course.
Will: Gonna wring you dry with insights and knowledge, because I think we all struggle with this. I've struggled with this plenty in the past. We have so many tools. We have things like Google Analytics. We have an insights dashboard for every other platform, whether it's Google Ads, Facebook, you know, there's just numbers everywhere. There are some tools that can help you put that together. Things like Databox, you know, Supermetrics I think it's called, the one for Google Sheets, etc., etc., but there still isn't a silver bullet. There's a lot of work required to turn those numbers into something useful. Where are we with data and its usefulness in marketing? Is it all just a waste of time anyway, or should we be persevering to try and make sense of it all?
Chris: So, here's where I would say a lot of people get lost. We think of data as the thing we need to be, you know, creating, producing, and stuff like that. And it's not. Data is an ingredient. The outcome we're after is actionable decisions. We need to make a decision. Should we invest more in this? Should we use this? Is this channel where the audience is moving? And so on and so forth. We really want decisions. And a lot of people have made a lot of money in consulting talking about data-driven, you know, this data-driven marketing and things, but we're not there yet because we think either data is the thing, or, even worse, a lot of folks in the last five-ish years have really gone whole hog into this field of marketing technology.
And MarTech is important, but that's kind of like appliances, right. So, like, a blender. Well, if you are making steak for dinner, a blender is probably the wrong tool to be using, unless you're making some really interesting steak. And so, I like to think of all of this in terms of a kitchen. We have an outcome we want, which is a decision. That decision is made of things. We have an ingredient, data. We have tools. Blenders, stoves, coffee makers, etc. What marketers are lacking are two important pieces. One, the chef, because you gotta have somebody who can cook the thing, and two, a recipe that says here's what you're trying to make. If you don't know what you're trying to make, then whatever you produce is fine, I guess. But if your aim is a steak dinner, and all you've got in the fridge is fish, and all you know how to cook is omelets, you're not gonna make steak. It's just not gonna happen.
So, we talk about Google Analytics, and Data Studio, and BigQuery, and things. So, those are great tools. They're fantastic tools. We talk about the data that those tools generate, you know, Facebook data and so on and so forth. But we lack the recipe, and we, ourselves, for the most part, don't have the skills. We're not good chefs. So, without those two components, we're kinda hosed. And that's where the problem is in marketing analytics today. It's not the tools and it's not the data. I mean, it is in some cases, but the majority of the time, it's no recipe, no chef.
Will: Well, I suppose also it's... And I love the analogy, and I'm gonna push this analogy. I'm gonna keep it rolling for as far as possible until it becomes absurd. But in this imaginary kitchen, the cupboards are full. I mean, the pantry is like Willy Wonka's pantry kind of thing. There's a huge amount of stuff, and it's like, "Oh, wow. I could be a, you know, I could make anything. I could match Gordon Ramsay or anyone in the world because..." And I think that's the illusion we all have. We all have, well, we can do anything. It's all at our fingertips, you know? So, I think that's part of the problem as well.
Chris: Well, yes and no. So, there are cases where, for example, let's use Google Analytics. If you rolled out a new website, and you forgot to install Tag Manager on it, and you have three weeks of just missing data, your data quality is off. So, your pantry may look full, but the ingredients may or may not be out of date, they may be spoiled, etc. And so, there's the opportunity for your data to also go bad. The other thing is you may not recognize or be able to even understand what it is that you're looking at if you don't have experience. To an untrained person in the kitchen, a box of wheat germ and a box of sand look like the same thing. One is edible, one is not. And if you don't know what you're doing, you might try to bake a loaf of bread with sand. And then no matter how good your appliances are, no matter how skilled you are, you're still making a block of sand at the end of the day. It's not edible. I mean, it is, technically, if you wanna get rid of intestinal parasites, but in terms of nutrition, not so much.
So, the same thing is true with our data. Yes, we have a lot of data, but a lot of it is of questionable quality. Particularly when you start talking about third-party data, like data coming out of AdTech systems. Anyone who's ever run a Facebook ad or a Google ad or something like that, and you looked at the leads, and gone, "What happened here?" You know that the data can be iffy. For anyone who's run a CRM, we have all found email@example.com, you know, in our CRM. So, the data may or may not be great quality, even if we have a lot of it. Having a lot of it doesn't necessarily mean it's good. And, it doesn't... You know, to your comment about Gordon Ramsay, he's got a lot of skill. You could hand him five mediocre ingredients, and he'll be able to make something from it. As long as the ingredients are good, he'll be able to make something. You hand 5 or 10 or 50 or 100 fabulous, top-shelf ingredients to an idiot, and they'll just throw it all in the blender and you're like, "This is compost. You made me compost."
Will: Yeah. And I suppose... So, in practice, you know, if I've got an e-commerce website and selling 20, 30 different products, I run Google Ads, some social ads, TikTok, Meta, that kind of thing. Surely, I suppose, you know, your average marketer doesn't have an off-the-shelf, to continue with the analogy, just doesn't have an off-the-shelf recipe, an off-the-shelf process, for turning that into something. Where is that recipe? Like, why don't we have these kind of very standard practices in marketing? And at some point, obviously, I'd like to talk about what you think they are, and what we should be doing, how we should be putting this stuff together. But am I missing something? Are these recipes somewhere, or what?
Chris: The recipes absolutely do exist, it's just they're usually not in marketing. So, a lot of some of the best practices when it comes to data analytics, data science, etc., they exist, and they exist, and they've been battle-tested and time-tested for years, if not decades, in other professions. Financial services has a legion of best practices for analyzing financial data, time series data. There's books and books on stock market metrics, to understand, like, "Okay, well, should I buy, should I not buy?" Biostatistics, there is a huge discipline of techniques for doing things like biosurvivability, half-life, measuring the half-life, measuring the impact of medications and treatments.
So, all these practices exist. They just don't exist in marketing, because, in a lot of cases, marketers, for good or ill, many folks, particularly folks with more gray hair, like I have, got into marketing because they didn't want to do math. Otherwise they would've become mathematicians, or statisticians, or economists, or something like that. Instead, they're marketers. And this is not a slam at marketers. This is just the way people sort of work out. But, as a result, they never got exposure to a lot of these quantitative techniques that exist in other disciplines. So, let's take, for example, you've got your e-commerce store, like you said, and you sell Squishmallows, which are these plush, stuffed animals filled with a certain type of memory foam. You wanna know what's selling...what of my marketing is working. Well, there are a number of techniques that can help you understand that, but one of the simplest is multiple regression.
So, you basically make a big old table, and at the last column of the table is sales that day. And then every other column in that table is things like, you know, Facebook ad impressions, Facebook ad clicks, Twitter ad clicks, YouTube video views, website traffic, searches, and so on and so forth. Everything you could possibly put together, day by day by day. Put together a couple years' worth of this, and then run a regression algorithm of your choice. You could use linear regression, you could use, you know, for certain types, you could maybe do lasso or ridge, or gradient boosting. You take your pick of the regression algorithm of your choice, and you say, "Tell me which of all of these variables, alone or in combination, has the highest mathematical relationship, the highest correlation to the sales of my Squishmallows?" When you do that, you end up with essentially what's just a bar chart of, here, these are the three or four or five variables that really matter.
And then you say, "Okay, well, let's see. Google Ads... Actually, no. YouTube ads from Google. That's one of the things, and our email newsletter, that's one of the things, and maybe, you know, that influencer campaign we did on Instagram, that's one of those things." That tells you these are the things that worked. These are the things that had that relationship. And then you build a test-it plan. Say, "Okay, well, our YouTube ads were identified as the top thing in this analysis, so this coming month, let's up our spend 25%, and let's see if our sales also increase commensurately. If that's the case, then we can now prove causation, and not just correlation." So, that's a standard best practice, a process, a recipe, for taking all of your data and slicing and dicing, and understanding it. But again, that type of analysis typically is not something marketers know to do.
Will: No. And even, really, senior, experienced marketers, you know, to be fair. That's really interesting. So, just for the benefit of an idiot, i.e., me, give me a really simple definition of a regressional algorithm.
Chris: A regression is simply reducing all the data to patterns, and looking for correlations. It's a series of correlations. That's all regression analysis really is. It literally means "to reduce." So, you have all this data, and you just reduce it down, and say, "Okay, which of these things matches closest to the outcome we care about?"
Will: Is that something you can just do in Excel?
Chris: You can. You can do it in Excel. It is exceptionally painful to do so, but you absolutely can do that. I typically do most of my work in a programming language called R. It's a statistics language that, you know, folks in academia are probably very familiar with it. The two major data science languages that the industry uses are R and Python. Both are great. Both are about, I think, equal in terms of capability. Both can interoperate with each other. And once you learn the techniques, it becomes very straightforward to do these things. There are off-the-shelf tools that can do this stuff as well. For example, IBM Watson Studio is one, where you can just upload your dataset, and then Watson will essentially do this thing called AutoAI, and do the regression for you, and pick which of the algorithms to use and stuff. There's actually a lot of products like that out in the market. They are all reassuringly expensive. The nice thing I like about R and Python is that they are free of financial costs. It's just the skill you need to make them do what you want them to do.
Will: Yeah, I suppose, for some people, it would be more cost-effective to just pay someone on Upwork or Fiverr or something to just do a one-off job or something like that, perhaps. So, I dunno.
Chris: You can. I mean, that's part of what my company does. I will say that it's a profession. Data science is a profession, you know, and my friend Tom Webster says it great. There are some things in life that should be reassuringly expensive, like sushi, and surgery, and science. Finding someone on Fiverr it's, be like, you know, "I'd like to find someone on Fiverr to remove my appendix." Like, "You know, I think I'm gonna go to the doctor for that."
Will: Look, I'm totally guilty. I've worked in marketing for years, in agencies, and me and my colleagues, we just never really needed to know much about what data science was or what it did. We just could ask certain questions, and they'd get answered, and we didn't really have to concern ourselves with how. But actually, probably guilty of doing a lot on intuition. You know, just making a lot of decisions on intuition, on gut feeling, or on looking at very basic correlations, you know, statistics that might have been in some ways misleading or incorrect, but massive assumptions, basically. And that's definitely widespread across marketing.
But I think it was really interesting what you say. It's very true, isn't it? We, go in to that as a profession. We don't go in as trained mathematicians or statisticians, so I think it's on all marketers to sort of train up to some degree in that area, even if you're gonna be briefing other people to do it. It's handy to get a bit of a handle on it. Now, funnily enough, the Digital Marketing Institute does have some courses about that. There's a web data and analytics short course. We'll put a link to that in the show notes. But anywhere that you can get some training on that, I would recommend that to our listeners.
Hello. A quick reminder from me that if you're enjoying our podcast series, why not become a member of the DMI, so that you can enjoy loads more content from webinars, and case studies, to toolkits, and more real-life insights from the world of digital marketing? Head to digitalmarketinginstitute.com/aheadofthegame, sign up for free. Now back to the podcast.
What are your kind of go-to tools that aren't horrifically expensive?
Chris: R. The R programming language. Because it is free of financial cost, and it is... I would say I probably spend 60% to 70% of my working day in RStudio, which is the free development environment for it, because I can connect it to my Google Analytics account, I can connect it to my Facebook ads, I can connect it to this, that, and the other thing, there's all these connectors, pull in all my data, and do all my analysis. And because I can pull in things like a biostats library, or a quant library from the stock market, there are pre-built packages, you know, tools that I can use in the R environment, to say, "Okay, well, I wanna run, say, a moving average convergence-divergence indicator on my website traffic," which, by the way, is one of the most powerful underused algorithms in marketing.
All it is, is you take a 7-day moving average and a 30-day moving average of a number, like website traffic. And when the short-term average is above the long-term average, it means you're growing. Things are going good. When the short-term average converges with the long-term average, and is about to go below, it means your traffic's going down. It's going the wrong way. And that means that you've got to do something at that moment to goose your marketing, send out another email, spin up another campaign, something like that, to reverse that trend that you're experiencing before it becomes too serious. A lot of the problem that marketers have is actually they don't look at their data often enough. They do an end-of-month report or an end-of-quarter report, and then they're surprised, like, "Holy crap," you know, "we're down 40% for this month."
Well, if you had had a tool like a moving average convergence-divergence indicator, you would've known on the 9th day of the month, "Oh, hey, look, our traffic is about to go the wrong way. Our averages are about to flip. Let's do something about it then." That technique is used by folks in the stock market to say, "Okay, this stock is now a buy," or a sell, right? Okay, when the average flips the wrong way, sell it before it tanks too far. Again, this is an example of taking something that's been battle-tested in another discipline, bringing it into marketing, and say, "Okay, this is a case where it's gonna work really well for us as an early warning system on some of our data."
Will: That's great. I'm just looking at their website for the project. It's r-project.org. I mean, looking at it, actually, when you first mentioned it, I thought, "Oh, this is beyond most of us to use." But I think there are some, I think to use it, like, in a simple way that you just mentioned, I think if you're used to kind of muddling through other bits of marketing technology, I think you'd get through it. It seems realistic to use it on a basic level, I think.
Chris: And again, there are tons of courses, free courses, that you can take on it. IBM has a whole series of courses you can take, that are totally free. You can just search that on their website. You'll find courses on, you know, Coursera, Udemy. You'll find so much, even just on YouTube, of just people teaching all these different techniques, how to get started, and it's all totally free. So, there's no financial cost. It is just an investment of your time.
Will: Yeah. Well, I've just installed it as we speak. So, I look forward to playing with that, because I love stuff like that, and I suppose I'd been scared off that, and certainly Python, because, you know, I'm not a really heavy-duty coder, so, yeah. Right, that's interesting. So, okay. We've mentioned a few different things, a few different types of reports, looking at correlations, how marketing is affecting results and sales, also moving averages.
Chris: A report is a tool. A report is just a tool. It's an output of data, processed into some form. And so it's like a frying pan. So, you know, which pans are the most important in the kitchen? Well, it depends. Am I making steak? Am I making a bechamel? Am I making, you know, soup? If the answer is soup, none of my frying pans is the right choice. It's gonna be a very frustrating day, for everybody involved. So, the real question is what are the key outcomes that you as a decision maker need to be making decisions on? And that's where you start.
That, again, this is a part where most marketers fall down hard. They're like, "Okay, I need the right tools. What are the right tools? What's the best report? What's the best practice for this?" Like, "No. What are the questions you have to answer? What does the CMO or the CEO or the board want to know?" Probably they wanna know what's making money. Probably they wanna know what their margins are. Like, which products have the best margin? Probably they wanna know what are we wasting money on? What's just not working?
Will: And I think we're scared of asking as well. You probably should ask them, because I think we're just a bit scared of asking, like, what we should know. Do you know what I mean?
Chris: I do know what you mean. And, you know, one of the questions that I ask, that clarifies this immediately, it's my favorite question to ask of a stakeholder, is what do you get your bonus on? When I ask them, like, what do you get your bonus on? They're like, "Well, I get my bonus on revenue," right? "I get my bonus on closed won deals." "I get my bonus on, you know, sales-qualified leads." Once you know that, then, like, okay, now we can talk about tools and reports. So, "Where do sales-qualified leads live in your organization?" And some people, it's a CRM, and some people it's in your marketing automation software. In a few places, it might be in Google Analytics. But where does that number live?
Once you have that, then you can start doing, again, a regression analysis, to see, "Okay, well, what are the numbers that lead to a sales-qualified lead?" And now we can figure out which reports you need, from which systems, that answer that question. But if you don't start with "what do you get your bonus on?" then you're just kind of fishing around hoping that you pick the right answer, and it's a waste of everybody's time. It gets people really frustrated. When you hear someone say "I don't know," you say, "Update your LinkedIn profile, because you're about to be out of a job. If you don't know what you get your bonus on, you are hosed."
Will: A lot of people in that job feel like they're expected to know what they're doing, and feel like they're expected to know what questions to ask. And they feel like they would probably invite in a lot of criticism and what have you, and a lot of people poking their oar in if they start asking people, "What is important in this organization? What do we wanna do? What do we wanna achieve here?" And it's like, I think people are reluctant to be that person, weirdly. So, yeah. Okay, well, that's good to know.
Chris: I've not that had be my experience, though. My experience is that when you ask somebody, "What do you get your bonus for? What are you held accountable for?" that actually seems to provoke a sense of reassurance, because when you ask someone that question, they go, "Oh, you're looking out for me. You're asking me what's important to me." And that, you know, really de-fuses that sense of, "Well, do you know what you're doing?" It's like, "No, no. You're looking out for me. You're asking me what I'm gonna get my bonus for, so I'm gonna tell you, and you're gonna help me get a bigger bonus."
Will: Oh, I agree with that. I mean, one of the best pieces of kind of work advice, career advice, I heard well over 10 years ago, because I can't remember where I heard it was, you know, "Don't be afraid to be the person...in fact, try and be the person in the room who asks those really dumb questions." Like, the really fundamental questions like, "Sorry, what are we doing here? What's the reason for all this, and what's that? And how does that work?" Because so many people sit in meetings afraid to ask those questions, and they just sort of ask their kind of, you know, the little tactical questions about their little bubble and their little job, but, you know, I think being the person that's really opening that box and asking those big, fundamental, supposedly silly questions, shows a lot of confidence, and shows you're actually properly thinking about the challenge at hand, rather than just turning up for work every day and putting your time in so you can go home and spend your salary, you know.
Chris: No, that's absolutely right. It's one of the reasons why, you know, when people ask me what I do, I often say I'm a data detective. And so, if you take a detective out of fiction, like Sherlock Holmes, for example, a lot of what Arthur Conan Doyle wrote about was essentially just inductive reasoning... Deductive reasoning to a degree, too, but inductive reasoning. Here's a bunch of data... In fact, there's even a quote, I can't remember from which book it is, but, you know, Holmes is quoted as saying, "Data! I cannot make bricks without clay." We're the same thing. We're doing the same thing. We've got data. We don't necessarily know what condition it's in. And we've gotta turn it into our final outcomes. And to do that, we've gotta get all of it. And so, asking a question like "So, what exactly is the point of this campaign?" Or, "Why are we focusing so much on this thing?" are reasonable questions to ask, because you need those answers to understand what the outcome is. And if no one can clearly articulate the outcome, like "we wanna boost sales 20%," then you can't do your job.
Will: Absolutely. What do you make of GA4?
Chris: Google Analytics 4 is a reskin of a tool called Firebase. Firebase was Google's tool, is Google's tool for doing mobile app analytics. And GA4 is essentially a web analytics-Firebase hybrid. Because Google realized, very early on, that these lovely smartphone devices we're all carrying around, these supercomputers in our pockets, are where people do business. They're where people make decisions, they're where people read, and there's a lot of cross-device stuff happening. You know this yourself. You will look at something on your phone, then you'll go to your computer to, you know, order or whatever, because typing in, you know, filling out a form on your phone is never fun.
And so, GA4 was built with that in mind. It was also built in mind to reduce everything down to a granular measurement called an event. An event can be a tap, a swipe, a scroll, etc. Again, from the mobile world. Why does this matter? Because the old Google Analytics, which has had the same basic data model since 2005, when Google acquired it, really was poorly matched to Google's ad system. Google Ads is all event-driven. So, now you have two systems, GA4 and Google Ads, which are at the same level of granularity, so they can share data more easily, they can match data more easily, and you can pass data between the systems more easily.
So, Google Analytics 4 really is a tool that makes Google's life easier. It makes it easier, it makes their ads more effective, it allows them to use the same machine learning tools and the same algorithms for attribution across both systems. So, it's all about making Google's life easier. You will note that at no point have I said it makes marketers' lives easier. It does not. It is a different system. It is a very, very different system, and it takes some getting used to. It is not unknowable. We've been using it for almost two years now. It is a different way of thinking, it is a different way of doing things, and it is still a work in progress. Like, just last month, Google added, say, "Hey, there's 16 new metrics that you can use in the product." We're like, "They should have been in here on launch day on October 2020. You're just getting to them now." But that's, you know, that's the nature of it, and I can't complain. Most people can't complain, because we ain't paying money for it, right? So, you get what you pay for.
That said, there are a lot of people I think who have, giving some consideration to other systems. I'm a big fan of recommending Matomo as a backup system to Google Analytics. It's a free, open-source product. I would describe Matomo as where Google Analytics was five years ago. It's pretty good, it's pretty simple, it's very free. You could pay for hosting for it and stuff. But if you're used to the Google Analytics architecture of old, it's not bad. Google Analytics 4, again, it's really powerful. It is a much more powerful system, and once you learn... It's like we were all riding bicycles, and suddenly Google dropped off a Porsche. Like, "Okay. Well, this is a bit different. I don't have to pedal anymore. But holy crap, I can run into a wall really fast with this thing," and tires flying through the air and stuff like that.
Will: Yeah. For a lot of marketer users that aren't necessarily very data...they're not really pros in the way that you are, I think, yeah, at first glance, it's like, "Oh, no. I don't like this." But, I mean, clearly, we've been struggling with event-based tracking for years, and this, I think, helps.
Chris: I will deliver this warning. There's a warning about it. When you move to Google Analytics 4, your data does not go with you. When you turn on Google Analytics 4, it is brand new from the day you turn it on. The old Google Analytics, Universal Analytics, or Google Analytics 3, comes to end of life July 1, 2023. On that day, it will stop collecting data, and GA4 will be the only system that will collect data. Which means that if you have been waiting to switch to Google Analytics 4, it is now, the day we're recording this is August 16th, 2022. That means that August 16th, 2023 will be the first day you will have year-over-year data, because you've missed July, you've missed the first half of August. And so, the longer you wait to get Google Analytics 4 up and running, the less data you're gonna have once the cutover happens. It doesn't have to be perfect. Just get it installed, get it running. You can run both systems side by side, but get it up and running as soon as you possibly can, because right now, you're losing year-over-year data every day you wait.
Will: Yeah, true, true. Yeah, we've been pushing people to do that in various places, webinars, and podcasts, and resources. And there's lots of resources on the Digital Marketing Institute library. I should mention there's an e-book, there's webinars, loads of stuff there. Okay, that's interesting. I mean, it's interesting that you mentioned having other analytics systems. Again, don't wanna get bogged down too much in tools, but you use Matomo, which I think used to be called Piwik, didn't it? And I think I may have had a play with it some years ago. But are there any other tools like that, that you think are worthy of mention or recommending?
Chris: Again, tools are like appliances. It depends on what your budget is. You know, you may or may not be able to afford the Vitamix or the Blendtec, you know, high-end blender. You may, you know, be using the Hamilton Beach with the Walmart special or whatever, the cheap blender. But there's a bunch of different tools out there. For example, for my email newsletter list, because my list is so large, it's 235,000 subscribers, I have to use an open-source tool called Mautic, because all of the commercial services, I could not afford to send my own list because of the size. It would cost me thousands, if not tens of thousands of dollars a month to send out my weekly newsletter to a quarter million people.
But using Mautic and Amazon's Simple Email Service, it's about $100 a month, which I can manage. That's a lot more manageable. Mautic is also open-source software. It is open-source marketing automation software. If you're familiar with HubSpot or any of the other marketing automation tools like Pardot and stuff, it's very similar. Here's the thing about open source tools, and I'm sure you're seeing a trend, because we've been talking about R, and Matomo, and Mautic, and stuff.
Will: Yeah, just gonna ask you about that, yeah.
Chris: Open source tools trade time and expertise for money. So, you're not paying money, because my Mautic instance costs me $5 a month. Show me another marketing automation system that costs you $5 a month. But, you've gotta know how to administer a Linux server and run a VPS, in order to have it up and running, and configure DNS, and do all the things, you know, sysadmin work, and then install and run the software and keep it up to date. So, you take on the administrative responsibilities, but in exchange, you're not paying $200, $300, $400, $500 a month. Which, again, we're a small business, too. You know, we're three people. A system that costs us $5 a month or a system that costs us $500 a month, that's a pretty big difference. Because I have the expertise, I'd rather spend the $5 a month, and keep the other money to build the business with. So, if you have people who have good technical skills, or you have a good IT department, look into some of the open source options. They are really good, very mature, and super low-cost.
Will: Yes. And just to one more thing about tools. Like I said, I've talked about things like, you know, I mentioned in the beginning, things like Data Studio and Databox, these kind of amalgamation tools. Are you a fan of that? Are you a fan of these dashboards where you can feed in data from different sources, or are they just a bit of a distraction because they sound great, or what?
Chris: Again, it's a tool. It depends. It depends on how you use it. If you're putting everything in the blender, some things it works well for, some things it doesn't work so well for. So, a tool like Data Studio, it's like a spreadsheet or a PowerPoint deck. It is a tool. And you can do great things with it, you can do stupid things with it. Dashboarding and reporting in particular is something that marketers don't have a lot of training to do well. One of the things we say all the time, I say it in my reporting keynote, is, you know, a dashboard without a decision is a decoration. So, if you've got a dashboard with all these different widgets and dials and knobs and whatever, that looks great. It's like art on the wall. But if you're not using it to make any decisions, it's really not useful.
One of the Trust Insights dashboards that we built for our own company has just got two numbers on it. That's it. It's got form fills, and it's got website traffic. Two numbers. And basically, you know, green arrow up, red arrow down. And that's set as my browser homepage. So, every time I open a new tab on my browser, I see just that dashboard. And that's all I need on a day-to-day basis to look and go, "Huh. That's weird. That's, you know, why is that red all of a sudden?" I don't need it to be more complex. I just need to look at it and be able to make a decision, "Oh, I need to go see what's going on. Why is that number red? That red number was not red yesterday." That's what we mean.
And a dashboard has gotta help you make decisions. If it doesn't, it's just a waste of your time. And, you know, a Data Studio, Databox, Tableau server, all these tools are good tools. But the recipe and the chef are what matter more. A dashboard or a report, or data itself, doesn't have to be complicated. It shouldn't be complicated. It should be, you look at it and go, "I know what decision I need to make. If it's a red arrow on today's dashboard instead of a green arrow, I know what I need to do when that red arrow shows up."
Will: But even that's not always obvious. Do you know what? Like, going back to thinking of just a simple thing like an e-commerce site, you've got a few different ad platforms running ads. I don't know. Sometimes it's not even obvious what the important numbers are, and okay, if sales are down, or clicks are down through Google Ads, what do I do? You know, I think so many marketers are still caught with that, like, what does it mean? You know, what decision is this telling me to review?
Chris: Well, it depends. It depends, and this is something, you know, that's probably the phrase I say the most. It depends. It depends on the number. You know, we built this one dashboard that broke out a client's, all their data by channel. And again, it's just real simple. Green arrow, red arrow. And they can look at it every single day and go, "Huh. Display ads, the arrow's red today." Okay, great. Now you know what system you need to go dig into. They open up the display advertising system, and then there's all your component metrics, you know, impressions, clicks, click-through rate, etc. And they can look at those numbers and say, you know, green arrow, red arrow. Which of these things is the problem child? Is it that we're not getting enough impressions? Okay, well then we need to maybe either fix our creative or up our bids. Is it the clickthrough rate? Is there a discontinuity between, you know, what we're triggering for or where the placement is, and what the offer is?
Bob Stone's 1968 direct marketing framework is still one of the most valuable frameworks for diagnosing your marketing campaigns. Audience, offer, creative. Is it have I got the wrong audience? If you do, fix that, because nothing else matters. Have you got the wrong offer? Got the right audience, but you're offering them something they don't care about or they don't want, that's your next step. And then the third step is creative, is the right creative to highlight the offer to the right audience.
Marketers screw this up constantly. They go the wrong way. They say, "It must be the creative." And they would do 500 variations of a creative, not realizing they're showing ads for a, you know, a barbecue widget to people who don't have homes. If they don't own a home, they don't have a barbecue outside. So, they may look at the ad, and they're like, "Oh, that's cool. That'll be a great, you know, great Father's Day gift for my dad. But I'm not gonna buy it because I don't own a house, and I don't own a grill." You've gotta get the audience right first. And so, a lot of the time when we're teaching people how to diagnose their marketing, it's using that framework, and say, "Okay, you've got green arrow, red arrow. Now, tell me, have you got the right audience?" And that's when they go, "I don't know if I got the right audience." Okay, well now we have a line of inquiry. Now we can start digging into the data and say, "Do you have the right audience?"
Will: Yes. That's an interesting framework. We'll definitely put a link to that in the show notes. I think that's what it is, isn't it? It's about diagnosis, diagnosing marketing, and working out what to change, what to alter, try and work out where we're getting it wrong. We're always getting something wrong, we just have to find out what. And like you say, you're the detective that's trying to hunt down those issues, right, and optimize.
Okay. In terms of just getting away from the quantitative side of data, there's obviously a lot of talk about third-party, first-party data at the moment. I believe Google's delayed the change to Chrome now, till 2024?
Chris: Yep. They have, but it's kind of a non-issue in a lot of ways, because, particularly in the EU, you are seeing a lot more action against companies for violating GDPR, the General Data Protection Regulation. China rolled out its PIPL Act two years ago now. And GDPR has substantial penalties, like, financial penalties. China's penalties for violating their privacy act means that if your executives show up in China, they go to jail. You get locked up. So, you know, and we see this patchwork quilt in the United States of different states having different regulations. California, Virginia, etc., all have different data protection regulations. As of right now, GDPR is sort of the gold standard across the board. So, for marketers, you wanna make sure that you are fully GDPR compliant, regardless of where you live, where you do business, because it is the gold standard, and your mission as a marketer over the next two years is to focus like a laser on acquiring first-party data.
So, let's take a quick detour here. First, second, third-party data. First-party data is that you, the user, give to me, the marketer. You volunteer. You say, "Here's my data." Second-party data is when you, the user, give your data to a partner of ours that you know, your informed consent. So, if Digital Marketing Institute and Trust Insights were doing a webinar together, and you fill out the form to attend the webinar, and you see both of our organization's privacy policies, but it's on the Digital Marketing Institute website. That's second-party data because you're giving the data, as a user, to DMI, but you know it's going to Trust Insights, because we've co-branded it.
Third-party data is when you, the user, are giving data, maybe voluntarily, maybe not, without consent necessarily, to somebody else. So, that can be everything from AdTech systems to buying an email list, etc. Third-party data is the thing that's in the crosshairs. It is going away. It doesn't matter, you know, what timetables or schedules or whatever. Because the legal liability, in all of Europe, and all of China, and the good chunks of the rest of the world, it is going away. So your mission as a marketer is to focus on first-party data. Get people to voluntarily, eagerly give you their data.
Will: Yes. And, I mean, to what extent, if I'm a shoe company, I'm trying to find out people's shoe size, and gender, and preferences, and activities, I mean, it's that kind of more textured stuff about them as a customer that can just help me be more relevant to them.
Chris: Exactly. And the thing that you have to switch up in your head is, A, what data do we actually need to make decisions? And B, what data helps my customers? Do I need to know your name? Maybe, maybe not if I'm a shoe company. Do I need to know your shoe size? Yes, that's useful data. I'm not gonna show you shoes that are in stock for things that are not your size, right. That's silly. So, that's a relevant piece of data to make a decision. Do I need to know your household income? No. I just need to know whether you can buy the shoe or not. Do I need to know what shoes you browsed on my website? Yes, because that could help me recommend other shoes that you might like.
So, there's a lot of that being very introspective and looking what data do we actually need, what data is just a distraction, and what data is radioactive? GDPR in Europe just reclassified and said that all PII, all Personally Identifiable Information, is also classified now as sensitive protected information. So, any personal identifying information in EU, because of a recent court case, is now treated with the same sensitivity as things like medical data. So, unless you need people's personal information, don't collect it. If you're just sending out emails to an email list, do you need the person's name and title and stuff? No, you don't, just to send out an email newsletter. When someone says, "Hey, contact me. I want a demo from Trust Insights," like, okay, then it's relevant. Yeah, what's your title? What company do you work for? But do I need that information for my emails? Probably not. And do I want to store that data and have it be a legal liability? Again, probably not.
Will: Interesting. That's interestingly increased stringency of GDPR, and being even more strict about, yeah, only kind of asking, you know, and gathering what you need. Fascinating. So, you spend, you know, your time telling people how to do this better. Some of your time anyway, telling people how to use data more constructively.
What five or so things would you recommend that people go away and do after listening to this, to start to become a genuinely more data-led and data-literate marketer?
Chris: What decision are you trying to make? And if you're not clear, hammer that out first with your stakeholders. What decisions do you need to make? Where does that data come from? What condition is the data in? And then, how can we display that data easier or better or more clearly? Who needs to be seeing your data? So, that's a big part of governance. And then, I would say the last thing for people to really give some thought to is how much data do you actually need? Like, what is essential for making decisions, versus what is just extraneous information. And, critically, look at what your partners collect. Look at what, you know, if you use a CRM like Salesforce or HubSpot or whatever, they're collecting data too. If you're doing business in locations where that data is restricted or prohibited, your partner is providing you that data, so the legal liability for that data is still on your organization. You've gotta know that. You've gotta know what's going on inside the systems that you own. If you don't, you are gonna be held liable for it one day, and you will be like, "I didn't remember doing this." And that doesn't fly in court.
Will: Well, look, I'm aware our time is kind of pretty much run out, Chris. Thanks so much for all your insight, your knowledge. I got a lot out of that. I do have one last question for you, actually. Just tell our listeners where they can find you, and connect with you online.
Chris: Easiest two places. Go to trustinsights.ai. That's my company. And christopherspenn.com. That's my personal blog. Those two places, you can find everything else, you know, YouTube and Twitter and my blah, blah, blah, blah. All those other places. But those are the two places that are easiest to go to find all the stuff that we create.
Will: Well, thanks a lot, Chris. Really appreciate that. It was very useful, all your insights there, and I hope to chat to you soon. Goodbye.
If you enjoyed this episode, subscribe wherever you get your podcasts. And for more information about transforming your marketing career through certified online training, head to digitalmarketinginstitute.com. Thanks for listening.