How to Increase Sales 20% Using Predictive Analytics with John Wall of Trust Insights
Overview: John Wall of Trust Insights goes over how predictive analytics can help you optimize your marketing strategy. Learn about attribution modeling, data cleanliness, influencer marketing, COVID effects, and how to get started.
Guest: John Wall is a Co-host of Marketing Over Coffee and Partner at Trust Insights. He’s a marketing and sales professional with management experience with teams from 1 (with contractors) up to 120. He loves using technology to improve sales and marketing productivity.
All right, everyone. Welcome to Digital Conversations. I’m your host, Billy Bateman. And today, I am joined by a podcasting legend, John Wall, partner at Trust Insights, and Co-Host of Marketing Over Coffee. John, thanks for joining me.
Oh, thanks for having me here, Billy, I appreciate it.
Yeah, I’m really excited for this man. So, before we get into it, we’re going to talk a lot about predictive analytics, attribution, you know, just, you know, your data cleanliness within marketing. But before we get into that, let’s learn a little bit about you and what you do. So tell us about yourself and a little bit about your journey.
Yeah, sure. So I’m a partner at Trust Insights, a marketing analytics firm. We say that we light up dark data. We’re kind of data detectives, we help people figure out what’s working in their marketing, what’s not working, where to go next.
But my career path has been kind of crazy. I mean, as far as marketing tech, I’ve been in the startup world since about 97. And have gone through what this is like my seventh startup.
We’re actually three years in I mean, we’re well beyond startup phase, but I’ve kind of cycled through a number of times. And then before that, my background was actually economics. I graduated with a degree in econ. So I’ve always come at the marketing thing from the quantitative analysis side and tried to automate as much as possible. So yeah, it’s been a long, crazy path and a bunch of wild pitstops along the way, but everything with trust insights is going well.
Marketing Over Coffee
And then most of what we do at Trust Insights originally started, when I was working with Christopher Pen, we started this podcast Marketing Over Coffee about, it’s going on, like 15 years now.
But every week we talk about marketing and tech, and that has just kind of finally gotten to a point where it brings in clients, for us. For Trust Insights. We have a reputation as knowing about and keeping an eye on what’s going on in what’s changing, because this space is so crazy and dynamic. That has helped us build the community that we’re able to kind of trade ideas with and talk about what’s going on with Martec, and it works well.
Awesome. Okay, so you guys, you do a lot of work with predictive analytics. And I mean, it’s a nice buzzword. I don’t think a lot of people really know like, what is that? What does that amount to in the real world? So I think let’s, let’s just start there. So when you guys are talking about predictive analytics, like what are you looking at? And what can you help people forecast?
What is Predictive Analytics?
Yeah, the most common thing that we do for predictive is topic analysis. We’ll go take a look at grab a library of terms. Our chief data scientist, Christopher Pen will run that against a number of models that he has.
The most common example that we use, we have a blog post, we update every year called the cheese report, where we look at all the cheeses in the market, and we come up with the calendar for the next 12 months that says, Okay, these are when specific cheeses are going to be hot and moving. And so as somebody who’s creating content for a website, if you were in the cheese industry, you know that come May and June, you better be teeing up all your content and videos about halloumi. And I didn’t even know halloumi was a thing until I read the cheese report.
Predictive Analytics Example
As we dug in, it’s a grillable cheese. So it always peaks in the summers, when people are looking for halloumi recipes or want to buy Hulu me, it’s because they’re getting ready to throw it on the grill in July and August. And so and then, you know, as you keep digging into the data, you’ll see mozzarella is on fire or on Christmas time, cheddar right around New Year’s as everybody’s doing parties, monterey jack comes in around the Superbowl when people are making nachos.
The idea is that by looking at all this data, and looking ahead, you can predict the future and say, Hey, we want to drop our content, you know, on these weeks, because we know there’s going to be demand the week following that. We will have had content in place for a couple of weeks before the surge hits.
You know, these models can be applied to anything. I mean, if you have enough sales data, you can look and say, get an idea for maybe what your seasonal sales cycle will be. But we often find in b2b that there’s just not enough data to really get effective models going.
But you know, as far as like applying that same model to other stuff. Women’s shoes, we have a client that uses the women’s shoes fashion report, so they know six to three months out from Black Friday, what’s gonna be hot in specific kinds of shoes and where to go.
Effects of Covid on Marketing
So it’s gotten a little bit rough. You know, it used to be rock solid, we could kind of generate models and they would just always be you know, 95% greater confidence interval. We would just know that pretty much certain. But COVID has kind of thrown a wrench in things. There’s, been a massive change in search behavior. And so a lot of markets have been messed up so it’s not as easy as it used to be. We used to kind of be able to say yeah, we can definitely do that for everything. And now for a lot of projects, we say hello, we’ll look, we’ll go in there, we’ll run some models, and we’ll get back to if we think we can do some predictive because it’s a lot more difficult than it used to be.
Yeah. So with COVID, what, what are you seeing is the changes and just consumer behavior? Since COVID?
Covid Effects on Consumer Behavior
Yeah, you know, in one way, it’s not radically different. Really what people have been saying some of the studies we’ve seen is that, it’s, it’s as if we just jumped five years ahead. We’ve kind of been on this ramp of e commerce is going in this direction, and eating up more and more brick and mortar. And it’s suddenly as if we just jumped ahead five years into the future. Because everybody now was forced to go online for purchases, where there’s this group of people who still like to go to, you know, the local big box store to buy stuff. And now they have no choice because of, you know, the toilet paper is gone, or they don’t want to go outside because their risk or whatever. So that’s one thing.
Covid Effects on Web Behavior
The other one that’s kind of interesting is we’ve seen a lot of stuff that web behavior, it’s almost as if it’s the weekend all the time. We used to see very specific patterns in like Monday through Friday. There was all this b2b commercial traffic. Then on the weekend, that the commercial stuff was only 10 or 20% of the traffic. And then it was sports and Facebook and social and you know, kind of messing around. And now it sort of looks like the weekend all the time, because people don’t go into the office or they can’t even go to certain places or whatever. But people are now free to kind of roam where they want when they want to roam. And so it’s yeah, it’s changed the open rates in the rules a little bit on a lot of stuff.
Interesting. Okay. All right. To go along with predictive analytics, when you guys were talking about attribution and like, Okay, this is something our marketing team, I’ll hear them debate it. And sometimes they’ll get in, and sometimes I’m like, I’m not gonna step in that today.
But how do we decide where somebody came from? Was it the LinkedIn touch? Or an email? Was it a display ad? How do we decide if this is the straw that broke the camel’s back and got them to purchase? I’m interested to hear what your opinions are and what you’re seeing there.
Attribution Modeling: Where is my Traffic Coming from?
Yeah, so attribution modeling. I mean, you’ve kind of nailed it right at the heart. And you’re still tracking human behavior. And everybody has these situations where they’re like, okay, yeah, it came from the white paper. We saw and it came in from the white paper. And then the person comes in a month later for a customer orientation. And they’re like, Oh, yeah, you know, I learned about you because George in accounting used to work for me six years ago. It has nothing to do, everything gets ruined. And so it’s basically a best fit model. You’re kind of just always coming up with mathematical models or a procedure and seeing how close you can get to reality knowing that you’re just never going to nail it 100%. You can’t get it right.
And so, a big push, as this initially grew was, you know, first it just started with source codes. It was just like, what was the last thing they did, like, okay, we’ll give him credit for that. And then, of course, you saw that it was like, it’s always website. Generic website is where they came from, because nobody just comes to the website, they come from somewhere else to the website.
Multi touch attribution, you know, we finally got enough data to be able to handle that. So we were able to say things like, well, we emailed them this white paper, and then they came to the website. We can give, you know, 25%, credit to email 25% credit to the website. Then the math gets more advanced, maybe you’re saying, well, the last touch will give 80% to and the first touch only gets 25. Some people would flip that they’re like, well, we’re interested in acquiring new business. So we’re gonna put 80% of the first touch. Then maybe you start to get really fancy, and you do time decay, you know. So it’s like, well, anything over a year, we don’t really care about like if it was in the last nine months we care about. And so there’s a bunch of different models.
One thing that we’ve done, that’s unique to our attribution models, is we use machine learning to just grab your entire data set. Instead of having a formula, we just grab everything, and we test every single touch point and option. If it’s, you know, a single white paper, we look at the, you know, even if you only have 30,000 contacts, you might have 200,000 interactions. And so we test that white paper against all the interactions, and our statistical models can come back and say, Okay, well, you’ve got this white paper. But we know that if we pull that white paper out of the model, statistically, it’s probably not going to change anything. So even though a lot of people hit this white paper, we know that it’s not really worth doing.
And at the other end of the spectrum, you may have, well, only 10% of the group watches this video or receives this email. But like those folks always close. So that is a legitimate point. And so we create the model, monthly or quarterly, we just run the data and you basically get a map back of like, here’s the marketing programs where they fall in the funnel to which is cool. Instead of just saying, you know, it’s binary. Yes/no, win/loss, it’s, well, the new business is all coming from your social media presence. And the deals that are closing tend to be getting email newsletters and things like that, like we see where things fall.
And another thing, we get to a point where you see how much of a crutch is organic search. Because we see some customers where, like 80% of their traffic is coming from organic search. And well, that’s great, because that’s almost free inbound traffic. But the thing with that is, if something goes wrong with your pages, or you start, you know, you get hit by an algorithm change, your business could be wiped out.
So yeah, that’s I’m kind of rambling on, because there’s so much going on with that. But that’s our view of the world.
Importance of Human Connection
Awesome. So I want to circle back to a few things that you you hit on. So one was just that example of, hey, we thought they came in through the website. But it really was like, somebody talked to me and told me, I should try your product.
I actually, I had a conversation with Dr. James Oldroyd at BYU a couple of weeks ago, and we were looking at some data around what technologies people are using, and we’ve mapped it geographically. And we saw there’s a huge trend towards if a company is located in a certain region, the adoption is probably going to be higher for them in that region. So somebody here, we’re located in Utah, call it the Mountain West region. We’re going to have higher adoption, because we can just go down the street and talk to somebody. Or we may already know, like, hey, my buddy works over here. I’m gonna call them up, see if it’s a fit. Or somebody leaves us and goes to another place, or they use us at somewhere and go somewhere else.
He’s seen that as a huge trend and other research he’s doing as well, that just the idea of being close and human connection, is still very real in the sales and marketing world.
Oh, yeah. And that’s it. So we see that now in this huge wave of influencer marketing. And there’s a lot of stuff going on, on that front. In fact, we have a lot of research that we do on that front of, you kind of hit these certain groups of people that are super spreaders. There are certain folks that, you know, like a chansons Meister and out of San Francisco, he’s not a San Francisco anymore, but he was out of the Bay Area. And he was like, this cluster of martec people around him. And if you try to martec tool, and it worked well, like suddenly, you’d see a pop up in 25 other places over the next six months.
There’s a lot of that, that, yeah, we’re not really to that point. You know, we have it people kind of underestimate. In the past three years, there’s been this huge boom in accuracy, because now we’re tracking mobile, and, you know, desktop. Yep. Which we couldn’t be able to do. And so now being able to get that back together, there was another huge pile of data into it.
The Influencer Effect
But yeah, then there is always the, the influencer effect. And so that’s how we suggest to go that route is making influencers one of your marketing programs, and you know, then it will statistically show up if they’re driving traffic, because usually the influencers will be, you know, pushing people to pricing and download pages. You’ll see them real late to the end of the cycle. And so we see conversions. Oh, and depending on the climate to have affiliate might be something you might even want to look into, you know, there’s sometimes you have people that are such influencers, that you’re willing to give them a cut of the sale, because they can bring in so much traffic and so much action.
No for sure. My brother works for a real estate software company, and they drive I think he said 60 to 70% of their leads through affiliates. So he’s like, yeah, that’s, what works, man. So.
Okay, now I want to ask you another question about attributions. So if somebody is just getting started, like, you guys have a very complex model sounds like that you’re working with which I love. But let’s just say new startup, trying to figure out okay, we want to track some attribution. Where do you what do you think is the best way to start? Like, is it a last touch first touch? A little bit of a combination?
Getting Started with Attribution
Yeah, for starters, you know, Google Analytics has got some pretty good stuff out of the box. I mean, you can set up goals in there. It does a 90 day window, decay model. So you know, it will track everything that happens in 90 days, but it’ll discount the older stuff. So that can get you in the door and running. And it’s not even terribly expensive. I mean, we can do a ga model and a basic start for like, 10 grand. If you’re a startup and you have no money that’s painful, but if you’re considering, you know, hiring your first marketing hire versus having somebody just build that model for you and be able to point you which way to go. It’s actually a lot cheaper to go that route. It is.
But yeah, it’s, you know, pretty much setups Google Analytics, set up a set of goals, you know, you can get, I think it’s 20 goals in ga out of the box with the basic addition that’s free. And so that’s enough to at least point you in the right direction. I think a huge lift with that is that as long as you set up your channels, right, you know, you can see what your ads and social are doing. You can kind of break out that traffic and get a feel for who’s coming in the door and when they’re coming.
And then I’ve been out of the affiliate game too long as a real player. We have an affiliate program, but it’s more of a partnership thing. But there’s software on that front too where you can set people up as affiliates, and then you just give them a custom landing page for their purchases. And that’s, so that’ll show up over in ga also. So you can measure it all. Over there. Yeah, that’s a good way to get started. Awesome. I love it. I love it.
So let’s get hit the last point, which I think is really the foundation for all of this: data cleanliness. Anyone who’s tried to figure out any data related answer for marketing and sales, has probably had the pleasure of realizing man, the exports, I’m getting from, you know, whatever program I’m pulling it out of, it’s not been set up very well, I’m gonna have to clean all of this up. What do you guys seen as problems that people can avoid and best practices with keeping your data clean?
We see in some situations that we’re trying to get to a report or get to a model, that 90% of the work is the building the data structure, cleaning the data, getting it to a point. Especially when it’s cross platform, you know, that’s when you start to run into problems. If you’re trying to do an export from your email system, and from your CRM system and get that stuff to reconcile. I mean, it’s, it can be a ton of work. So, yeah, there’s just a ton of labor into it. And this is another one, where we kind of suggest that you should really check out getting some professional help. Because why would you want to have to build scripts, and all kinds of stuff to make three spreadsheets from three different systems work properly, when somebody else has already done that work, and can show up on day one, and load your data and get it to work.
Data Cleaning Challenges
So yeah, there’s a lot that goes into that. But yeah, it’s a huge challenge. I mean, documentation and compliance is a huge part of that, you know, as you’re building and answering questions, keeping detailed records of all the steps, and maybe even getting to the point where that gets scripted, so that in the future, you just drop the files into a folder, and they, you know, get spit out in the outputs in the right format. So you’re just dealing with exceptions, rather than, you know, cleaning out all the, the spammy or not, so hammy email domains and things like that, like, you can just have scripts throw that stuff away. But yeah, it’s a ton of labor. It’s, it’s something that you want to document and hopefully automate, to speed up in the future.
And then the other problem with that is, you know, the API’s change. or the the products change, or the output reports change. Any time, you could think you’re there and next month, you run the reports, and it breaks because the output has changed on one of the systems because there’s a new release of CRM, software, whatever.
So yeah, it’s ongoing and challenging. But the results are huge. I mean, we see across the board that it’s like, you can get a 20% lift for your whole enterprise, if you’ve got this thing built and tuned, right. Because it can just make such a huge difference and controlling your marketing spend and putting your money towards where it actually drives sales.
Yeah, yeah. So let me ask you this, John, then. What what platforms do you guys work with the that you find are the best to work with?
Tech Stack Options
Um, you know, it’s funny, there’s, there’s kind of no best. There’s a bunch of players that are all good for the right situations. But the ones that we see. Google Analytics just dominates as far as web analytics right now, because it’s free to get started. So small arcs can do it. But it’s got enough enterprise juice that it can work from mid size. And then you can if you want to write a big, huge fat enterprise level check, you can get ga 360. And that’s also pretty hardcore.
We see Adobe all over the place, that’s another, you know, real player in the marketing arena. And of course, Oracle’s NetSuite is just great because it’s enterprise grade. You can get payroll and you know, all these other departments that normally marketing goes nowhere near.
other big players, you know, CRM is is there’s just such a battle going on there between like Salesforce and HubSpot, and a few other players. A lot of them even now are coming to the place where it’s kind of like, if you’re in that Salesforce ecosystem, it’s hard to make an argument to get out of it. Because it’s pretty much like anybody worth a dime is going to integrate to Salesforce. So there’s that.
But we see HubSpot all over the place too, because they’ve done a great job of making it affordable for small organizations. And they’ve got a lot of neat functions over on the marketing side, too, you know, that was kind of where they got their start. But they’ve really made a hard push to CRM.
So yeah, those are the the major players. We’ve got couple email service providers. We see Marketo all over the place. Eloqua has been eaten up by Oracle, but it’s still around. So that yeah, those are kind of the major folks we tend to see in the in the game.
Cool. Okay, man. Well, before I let you go, I want to ask you, is there anything that I should have asked you that I didn’t? You know, you’re thinking, Man, if Billy was smart, you would have asked me this question.
No, no, you hit most of the big, big guns. Because, you know, yeah, that’s it. It’s data cleanliness, attribution. And, and predictive are the big things.
The only other stuff that we see, the influencing thing we touched on for a minute is, is huge. Because we’ve done a lot of stuff, as far as not looking at who has the most likes, or has the most followers, but analyzing the actual traffic and see who’s being talked about. Because that’s where you really find the influencers.
And so we’ve seen a lot of opportunities for companies, instead of getting, you know, the person that has 4 million followers and wants 20 grand to do something, if you dig into the data, you’re gonna find these people that are down at like the 70% mark. Maybe they only have 1000, or 2000 followers. But for some reason, when these people post something, that’s when all the million follower, people jump on, like, they’re the ones that the trend makers are watching. And so trying to find those folks who can move the needle for you. There’s, there’s a lot of money there.
SEO and Content Marketing
And then SEO, we didn’t really talk much about that. But SEO is just, you know, money on the table for almost every company. I mean, you can, if you create a good blog, you can get some inbound SEO traffic, and it costs as close to zero as you’re going to get for marketing campaigns.
Awesome. We’ve got blog posts from two years ago, that still drive visits every week, and not just a handful, like a lot of visits. And I’m just amazed. I’m like, okay, like, this was this was a hit!
Yeah, it’s nuts. I mean, I, so I was working at a company that had a software development tool, and we had a white paper that ran for like, seven years. I mean, it was just, here’s how you handle this chunk of the software development process. It was pretty much the only paper out there on that. And so it was just every month, there’d be 30 leads coming in from that. And we spent like two grand to write it the first year, and we got seven years of leads out of it. So if you can find your nation and publish some great content, you’re not going to be able to beat that for return.
I agree. When I was working inside sales, they did a research study on lead response. And man, I saw it on LinkedIn yesterday. They were still pulling data out of this and and promoting it. And I was like, wow, okay, this one just keeps on giving. And that’s over 10 years old.
Okay, man. Well, with that, I think we’ll wrap it up. I would encourage everyone check out Marketing Over Coffee. It’s great. There’s a reason it’s been around forever. But if people want to contact you and continue the conversation, John, what’s the best way for them to reach out? Yeah, marketingovercoffee.com is fine. There’s links there to go to every place else or @johnjwall on Twitter, and I’m on LinkedIn, you can look them up over there. I’m easy to get ahold of. We’ll chat later, man. Thank you again. Cool. Thank you. I appreciate it.