Dan Moyle 0:08
Ready to spend 15 minutes with the experts you admire need strategy sessions from thought leaders brought directly to your ears. Welcome to the sprocket talk 15 minute strategy podcast where every week George B Thomas uncovers the challenges that sales, marketing and service professionals face and of course, the strategies to help them overcome their biggest hurdles. So sit back and set your sights on growth with these bite sized conversations build your strategy goals. Ladies and gentlemen, welcome to the 15 minute strategy podcast. Hey, sprockets yours. It's your boy George B. Thomas. And yes, we're back for another episode of the 15 minute strategy podcast where we try to give you a strategy that you can think of moving forward in your business well in 15 minutes, but let's be honest, sometimes it's 18. Sometimes it's 20. But we're doing the best we can for you. today. I'm super excited as always, because we have a friend, even a mentor and somebody who I look at and go go I'm stupid well not stupid, but he's definitely smarter than I am. And so Christopher, why don't you let the sprocket tears The viewers, the listeners know who you are, what you do and where you do it?
Christopher Penn 1:11
Sure. My name is Christopher upenn, chief data scientist of trust insights.ai. We are a marketing data detective agency. So you got data mysteries in your marketing, we come and help you solve it.
George Thomas 1:22
By the way, let me just say that Santa's in the background is super dove. I'm loving it. I can't stop looking at him. It's just making me jolly on this episode. If you're listening to this episode, you should definitely go watch the video version over at sprocket talk calm but let's go ahead and dive into the deep end. Christopher when we want to have this conversation about what is predictive analytics, what are the foundational pieces that the sprockets ears, the viewers, the listeners even I need to know so we can carry on the conversation today.
Christopher Penn 1:54
So here's what predictive analytics is a very popular topic. It's sort of part and parcel of it. in data science in machine learning and AI, and fundamentally, it's about taking the data that you have and trying to forecast forward what's likely to happen. And the reason that's important, especially this time of year, you know, got the holiday trappings out is people are thinking about, what should I be doing in the year to come, whether it's fiscal year, calendar, year, whatever, what should I be doing? What's going to be important? And a lot of the times I we've all sat in those meetings, we're in a conference room, and there's stickies all over the walls. And what always ends up happening is that you get the hippo problem, right, the highest individually paid person's opinion about what the company should do not based in data not founded in reality, it's just opinion. And sometimes it works. Sometimes it doesn't. That's a risk you take when you go by opinion, when you use data, particularly external third party data, you can develop a better idea of what's likely to happen again, based on what's already happened. Now there's some things you can't predict right by definition, anomaly You are unpredictable. If on September 10 2001, you had made a forecast, you had no idea what's gonna happen the next day, unpredictable that can't be predicted. But there are plenty of things. And most of us work in businesses where there is a seasonality, a cyclicality that is predictable. So, let's talk to you an example of this. And if you're listening or watching, you'll want to go over to the show notes. We'll have a link in the show notes to a public dashboard that you can follow along with this is called the cheese OF THE WEEK forecast. We do these silly little forecasts because almost all of our work is under NDA. So we can't show client stuff. This is taking the names of about 50 different cheeses and looking at five years of search data when people Google for you know provolone cheese, and forecasted forward to try and predict when in the next 52 weeks, any one cheese is going to be searched for. So I'm going to scroll to the middle of the chart here, and I'm going to click on the January column the first week of January. According to this forecast the most popular cheese as well. Be jack cheese, American cheese. pepper jack cheese I think is kind of conflated and provolone. Why? I thought you serve it like holiday parties, right? inoffensive, they're fun. They're fun cheeses when you go. If you go either back in the past or forward to the summer, guess what mozzarella cheese is super popular. Why? Because of course you make pizza and, and other summery dishes. If you were a cheese shop or cheese vendor or a grocery store, you would want to use this forecast week by week to decide what to put on your end cap displays or your dairy section or whatever, to focus on what people are likely to be interested in based on what they search for. And that's a data driven way of thinking about your analytics. What can we forecast so that we can see here's what's likely to happen and then because we we all say and most of us want to be customer centric, we can use this data to think about what the customer wants, not what we want. We have to ask Obviously working our business strategy as best as possible, like if a customer searching for something we don't sell. That's a problem. But within the bounds of what we can do, we should be obeying what the customers are telling us, frankly, what they're typing into Google, they say they want. So that's predictive analytics in a nutshell, take the data you have, push it forward, and see what's likely to happen. Now, one of the most powerful applications of this for a lot of us who are digital marketers, is taking Google Analytics data like your traffic data by channel, organic search, email, pay per click, referral, traffic, social media, take your last however many years of data you have, and forecast that forward. If you can, you can then get a sense of here's how much traffic my website, my podcast, my whatever is likely to happen in 2020. And if you like the answer, great, keep doing what you're doing. You don't like the answer. Like if you see that looks like emails on the decline. That's a chance for you change your strategy now. So that we're not Talking about this, you know, this time next year like, Yeah, I just got my budget cut 40% because I didn't hit my goals like know in advance using predictive analytics, what's likely to happen and you can make changes to pivot towards the outcome you want.
George Thomas 6:14
Yeah, we all want that outcome that we want, we definitely not want our budgets cut. And we all like cheese. But I'm super curious if we want to do what you said and take our Google Analytics or take other data that we can get, how do we actually get started or set up a system where we can pay attention to this predictive analytics moving forward?
Christopher Penn 6:36
So that's a good question. And it can it's a very challenging question there because there are there is a very, very, very steep learning curve, the most basic form of predictive analytics. It not particularly reliable but better than you know, just guessing wildly, is to actually take your existing data, take it and put it into the spreadsheet software is choice. Make a chart right and men Any software packages and things allow you to put on a simple trendline a simple linear trend line and, and essentially just sort of drag that forward and say, Okay, this is the general trend or direction that we're going. Like when you look, for example, Instagram engagement rates over the last year, that trend line is going down into the right like it's a, you don't need to be a real sophisticated data scientist to see that that's going to zero. That's the basic version. It's okay. There's that now a quantum leap to the next level, which is where you start using statistical software. So IBM SPSS, tableau, our programming language, Python, any of the sophisticated tools where you can do a much more rigorous mathematical forecast using what's called time series forecasting and the generally accepted technique for this and the statisticians will know this. Everyone else will be googling is called autoregressive, integrated moving averages. It is essentially saying What happened in the past account for seasonality account for the for basically moving averages over time, project that forward. And that's really the gold standard today for doing this type of time series forecasting is to say, look, I want to use this algorithm to push this data forward. And then there's one more quantum leap, which is, most people are not going to have access to which is to build what's called a deep neural network model of it you can use in this case would be a recursive or convolutional neural network to have a machine fully build its own predictive model. Those may sound fancy, they sound complicated, they don't perform any better than the the time the the, the statistical model. So stick with the statistical model. If you don't have that technology or that knowledge. You have two choices. You can either hire it either in house or at an agency or you can learn it and there are plenty of great courses out there. I recommend cognitive class from IBM to go to college. In a div class.ai, totally free 100% free of financial cost, it will just cost you time to take the prerequisite courses to develop those skills. So if you have time and no money, take the course if you have money and you don't have the time hire.
George Thomas 9:15
So I've got two questions burning in my brain. I'll ask the first one because it came to me first, but then when you were just talking, I was like, Oh, I really want to ask this thing. As far as like painting the picture of understanding your year in success. I'm going to come back to that that I'm going to want you to be like, what does it mean from a revenue standpoint, when you know your year in success, like the year of cheese, right? If you're a cheese vendor, and you knew the year of cheese, what would that mean for your bottom line? But here's the question that I want to ask before we get there. And that is just the fact of is this a one year strategy meeting? You could do this once and you'd have it for the next 365 days. Or Christopher, do you see this as something that if if you're really a savvy company, you might be doing this once a quarter? or once a month? Like how how micro Can you make this data? Or is this more of a macro plan when you talk about predictive analytics,
Christopher Penn 10:17
it depends on how agile your organization is and how skilled you are with the tools. With Google Analytics data personally from myself, I run this monthly from my just from my own personal website, because the thing about any kind of trend to software is the more data you have, the better the model gets. And it's like the weather right? You can forecast out a month in advance but the forecast accuracy tends to decline the further forward you get. So as as with weather so with your anything marketing related, the more data you have, the better your near term forecast gets. So I would rerun it as often as you have the resources and capabilities to do so. And whatever your change cycle is, so if you have Monthly marketing calendar. There's no point in writing this weekly because you've already set the month, right? If you have a quarterly marketing calendar, run it quarterly because that's the way you're going to be able to pivot. Don't run it more frequently than you're willing to make changes.
George Thomas 11:13
Yeah. So let's go back to this other question, because I always try to put my mind in the mind of the consumer, the listener, the sprockets here, and in that section before the one we just did, My mind went to Oh, my gosh, you mean, I gotta learn something. You mean, I gotta pay somebody something, what the heck. And I really want you to take a couple minutes here and paint the picture of the importance of this isn't just some fancy, you know, term predictive analytics that we're saying it's not something that's like just because it's cool and nerdy that we're talking about, but but the amount of impact that it could have on a business if you knew what was going to happen in these micro moments throughout the year, and why it might make sense for you to invest You're time to learn it or invest your money to get it.
Christopher Penn 12:03
So our best case study was for another agency, and I can't that's unfortunately under NDA. We did the forecast for them, we refreshed it quarterly for them, they took the forecast and used it. That's the key part. Most people don't do that part, like I bought the thing, everything's gonna magically get better right now, you got to use the data, but they used it. And they saw within a six month period, a 45% increase in lead generation. So their lead generation went up 45%, just by creating content. They knew what they had to create, but they didn't know when to create it. And by having by adjusting the time, they were able to meet the customer where the customer was where the customer was what they were thinking about. And they got that result for ourselves. You know, it fluctuates month to month for us, but it's typically I would attribute 20 to 25% of of our successes accompany our lead generation to this version. is what we see other competing companies doing. For example, we did a webinar back in August on social media 2020. We wrote a whole white paper with our partners to talk Walker. And when we were scheduling the webinar, one of the things we did was we pulled up a predictive calendar when people search for out of office out one, when you search for that, you are looking for your autoresponder so you can go on vacation, right, so when that search volume is lowest people are in the office. So we scheduled our webinar based on that timing. And it was wildly successful without as our second most successful lead generation campaign for the year for ourselves and our partner, we generated 1200 leads for that. And the total ad spent for that campaign for 1200 leads of which our partners that have 1000 were net new to their database was $50. That's it.
George Thomas 13:53
I mean, so it's it's mind blowing because the impact could be substantial. Just amazing and And what I love to sprockets ears, I hope that you rewind that section and I hope that you heard that it is a matter of you can generate leads with it, you could do certain ads and generate revenue with it. Like it's across the board, just a business strategy in general. And I wouldn't even say you could use it, because you would know what to share on social to even attract traffic. So I mean, the matter of a tract convert and close, like those three things could be impacted by predictive analytics. So here's the thing, Christopher, if they're wanting to Well, no, actually, let's do this. I want to do this before we before I ask the last question. I want to ask a question before that and, and I hope that we're getting into the minds of the sprockets yours might be thinking, Oh my gosh, this is magic. So what I want to do right now, yeah, exactly. What I want to do right here is I want to I want to ask the question, are there Any myths about predictive analytics about using this data in the way that we've been talking about that we need to like debunk right now before anybody gets started on this journey,
Christopher Penn 15:12
two big ones. Number one, we just said this is all math. It's not magic. There's zero magic to this and you cannot predict what is unpredictable right? So that's number one. Number two, and I think is a greater myth and one that I've talked about recently on another show is the myth that this takes away all your creativity This is the compass This tells you the direction you need to go you still gotta walk you still got to find your way you know no map no compass is going to tell you where not to step. Oh, look, there's a pile of horse crap I don't step in that right. So there's still a tremendous amount of opportunity for creativity within the direction that you're being set but this is telling you go north instead of West go west you'll end up in a lake you go north you get back on the road, how you get there is still up to you. So disabuse the the notion that this takes away creativity that makes You're slave to the machine it doesn't it just tells you the direction you should be going.
George Thomas 16:04
So as we close out the 15 minutes strategy podcast because they're trying to stay to that 15 minutes, what are some of the hurdles or potholes that you have seen people make when they start down this journey of predictive analytics?
Christopher Penn 16:15
So a couple of big ones. Number one, again, thinking it's magic, it is not and don't let the data override your common sense right. So if you know for a fact that something is happening in your industry, because you are out talking to customers all the time, the data is a guide, it is not a an ironclad rule. And to and this is the biggest one of all this is true of all marketing data, period. You have to use it. You just having the forecast is meaningless. If you don't use it's like having a recipe that you never cook, right. Oh, great. I've got this amazing recipe for brief Beef Wellington, but I'm still going to McDonald's for lunch. Well, you gotta cook. The recipe doesn't cook itself, right. And so the same thing is true for all the any kind of animals especially predictive analytics, you will pay. You know, if you hire an agent, you'll pay a reassuringly expensive amount of money for this forecast. If you just let it sit on a shelf, you've essentially bought nothing right? You will get no value out you must take action, take bold action on it in order to reap the value.
George Thomas 17:21
Christopher if this Rocketeers have questions as they take this bold action, where can they reach out? Where can they get ahold of you online?
The easiest place to go is our company website. Trust insights.ai.
So sprockets here's here's the thing, you don't have to be blind going into 2020. As a matter of fact, you should have 2020 vision with predictive analytics moving forward. Make sure you pay attention to that and we'll see you on the next episode.
Dan Moyle 17:52
Did you enjoy this episode of the 15 minute strategy podcast? We'd love to know. leave a rating and written review wherever you listen to your favorite shows and keep that level Going by visiting sprocket talk comm sign up for your free membership and in that membership area you can find bundled episodes where we combine like strategies to help you grow better make the world better and share this episode with your friends and co workers who may be battling this same obstacle. You can always reach out to George B. Thomas on Twitter with questions or guest suggestions or just to talk about your favorite Marvel superhero and go out into the world and leverage this strategy for your success. And we'll see you on the next episode of the 15 minute strategy podcast.
Where should I email this content to?
About the Expert
Christopher S. Penn is an authority on digital marketing and marketing technology. A recognized thought leader, author, and speaker, he has shaped four key fields in the marketing industry: Google Analytics adoption, data-driven marketing and PR, modern email marketing, and artificial intelligence/machine learning in marketing. Known for his high-octane, here’s how to get it done approach, his expertise benefits companies such as Citrix Systems, McDonald’s, GoDaddy, McKesson, Toyota, and many others.
Dan Moyle 0:08