Select Page

So What? Marketing Analytics and Insights Live

airs every Thursday at 1 pm EST.

You can watch on Facebook Live or YouTube Live. Be sure to subscribe and follow so you never miss an episode!

 

In this week’s episode of So What? we focus on using predictive forecasting to enhance your marketing. We walk through what it is, how to pull the data together and some use cases for your forecasting. Catch the replay here:

In this episode you’ll learn: 

  • What forecasting techniques are available
  • What can and cannot be forecasted
  • How to make use of a content forecast once you have one

Upcoming Episodes:

  • Video SEO – TBD
  • How do you benchmark a website’s performance? – TBD

 

Have a question or topic you’d like to see us cover? Reach out here: https://www.trustinsights.ai/resources/so-what-the-marketing-analytics-and-insights-show/

AI-Generated Transcript:

Katie Robbert 0:21
Well, hi friends, Happy Thursday, happy rainy Thursday if you’re in New England, welcome to so what the marketing analytics and insights live show, I am joined by Chris and John, as always, today we’re talking about marketing forecasting with AI. And so I have a bone to pick with marketers in general, about this topic. So one of the things that marketers are always asking is, how can I introduce artificial intelligence, you know, into my team into my business? What are some ways to do that, and one of the things we talk about, one of the easiest ways to introduce it is to get your hands on a predictive analysis, like the ones that we’re going to show today, because there are numerous ways that you can use one report. It’s very scalable. And it’s something that you can use, you know, for a whole entire year, because it’s basically projecting search trends. And so that’s sort of that’s my beef with marketers right now, as we always talk about how easy it is to use these reports. And then they’re not they don’t get used. So that’s my little rant for the day. So Chris, take it away.

Christopher Penn 1:32
All right. I have no beefs yet, today, although the day is still young. Yeah, it’s early, we’re gonna do exactly do two things shaffers. Gonna go through very, very quickly. I’m warning you right now. But very, very quickly the process of the creation of the report, and then spend the lion’s share the time on. So what what do you do with thing? Because obviously, the technical stuff is a lot less interesting. So how do we think about building a predictive report? Well, this is what we would call in AI, a both a unsupervised and supervised learning problem. When you go into say, I’m going to go pull up an SEO tool, he’ll pull up h refs, and I want to look up a search term to try and understand like, what is something that we would want to forecast that we want to know, why does Google show some things more than others? The challenge is that we don’t know how Google works, just, Google doesn’t know how Google works. But what we can do is we can reverse engineer it to some degree to understand not only the core, but also the outliers. So I put in my search term here, I’m gonna put it in data science course. Right? Because, of course, we are marketing, our Trust Insights, data science course, which shameless plug if you go to Trust insights.ai, slash data science 101, you will find the course Okay, commercial over. We have all these pages, I’m going to limit this in my search tool to this year, because again, search algorithms are changing all the time using the data from five years ago, probably not super helpful. And what we see is we get also let’s filter explicit results, filter, live stuff filter for the English language, because that’s what I can write. And we want to filter pages that have gotten search traffic, because there’s no point in looking at data that doesn’t perform. And we get is you get 180 some odd pages in here that all have gotten search traffic for the search term. Now, here’s the thing that we know about Google, it’s not about the keyword. It’s not even about the phrase, it is about the document as hold a page as a whole that says Google thinks this thing is important. And when you look at all of Google’s data, from the you know, in things like Google Trends and stuff, it’s all about the topic. So what we want to do is understand what is in this topic. So we have some code here. And this is like a cooking show, all these examples are already baked. And I’ve written some code that takes those 187 results, and gives me a basic keyword list of the terms that all appear on those pages across the board with let’s pull up what some of the results look like here. Where did they go? We see terms like data science, machine learning, data, science course data, scientists, data analysis, big data. These are all a lot. These are things you would expect. But then there’s also other things like data mining, for example. We have business analytics, these are all terms that are not necessarily in a data science course term. But they are semantically they’re logically related, right? If somebody is doing or learning, learning data science, they probably also want to learn about data mining, right? So this first step of what do You need to know is is essential. Now once we know that, we then go and we build ourselves a keyword list, keyword lists, in this case, I’m taking all those terms that we just learned about, plus all the different logical variations like, you know, training video class. And that’s, I’m going to have this particular piece of software just make a big list of hundreds and hundreds and hundreds of these search terms. Not all of them are relevant, right? Not all of them get good volume. So what we do is we take this, we put this back into our SEO tool, and we say, okay, all of these 1000 terms, only 179 of them actually get searched. Right. So now we’ve we’ve sort of filtered out the stuff that we know is just silly combinations, we’re down to the good stuff.

The last part of the process, then, is to use time series forecasting, to build was the past five years of data look like or the past, however, at whatever period time period you want to look at. And then it was time period for was the forecast of that data look like. So let’s go ahead and actually build that part right now. So I’ve got my data, I’m gonna put in my search volumes. I’m going to do this by week, because I like weeks. So we can be more precise in our timing. And I want to do this by term. And bring this in here. And I want to show just going forward, I don’t need to know the historical data, we’re using Tableau software here. You can, you can use any, any software you like. And so now, starting this week, these are the terms that are most searched for. Now, if I want to look at a specific term, like the big one, like bait data science course, on trying to sell data science courses, I probably want to focus in on that term. So let’s go ahead and duplicate this and change this instead to a line graph and move our term into colors, you’re going to see something horrendous. This looks like spaghetti, right? This is no fun to look at. But then I can go in just choose the individual term that I want. I want data science course slap some labels on that. And now I have a much better idea, oh, these the periods of time like September 5, people are gonna be searching for data science course more. September 20 1120 foot 22 people searching for it in the next is right next year right now is a really good time to be doing this. Obviously, in the early summer, not a great time. In the beginning of the third quarter, when a lot of people get budgets renewed on the new fiscal year, we see some some searching about that. So there’s some cyclicality and seasonality to that. But that’s the process of creating these things. Now, is there off the shelf software that does this exact process? No. But can you get at least a tiny bit of the way there even with thing basic things like Google Trends? Yes, to a degree. So Katie, here we are, we’ve got our forecast. We know what happened that Now what?

Katie Robbert 8:11
Now? What? Well, and so what we often see is, you know, teams are self included, you know, trying to figure out what content should I write? what keywords should I be optimizing for? What should I be promoting? You know, what keywords? Should I be, you know, putting into my ads, there’s so many different ways to be using just this one report. And so, that’s sort of the so what so if you take it, you know, if you start with just Well, you know, I have to write a blog post this week, what the heck should I be writing about? You know, what oftentimes happens is that people are just like, um, I don’t know, why don’t you write about this thing over here, or they actually do have a content calendar planned out. But it’s not aligned with what the audience is searching for. And when and that’s how this type of reporting can be super useful is you can customize it so that it’s your keywords, the keywords that you care about. So obviously, we’re focusing on our data science, one on one course. But we’ve also built this type of analysis for our company as a whole for all of our services. And so that way we know Okay, here’s what we should be talking about on our, you know, podcast, here’s what we should be writing about in the newsletter. Here’s what we should be blogging about on the website. Here’s what we should be posting about on social and really trying to hit every single one of those digital channels with the same set of keywords during the timing when it’s going to peak the highest.

Christopher Penn 9:44
One of the things that you can do is is also greatly simplify. This is a big, this is a lot to digest. You can just digest this down into a four week content plan. So to take a look and say like this week, we should be promoting data science online next week. We should have already created content for data science online course next week is published to make sure it gets indexed by Google. We should be getting content prepared, written right now today to be released in three weeks on machine learning training. And right now today, we should be looking at about four weeks out, what kinds of stuff can we be putting together things like a cloud computing certification or deep learning at deep learning training. And to your point, this is stuff that crosses channels. So it’s easy to say, Oh, this is an SEO thing? Well, yes, it is. But it’s also a content marketing thing. It’s a live show thing. It’s an email marketing thing. It’s a social media thing. Because when the audience is thinking about and looking for this content, I don’t know that they care, what channel it’s odd, they just want help with the thing.

Katie Robbert 10:59
Well, and it’s also a paid ads thing. So think about those, you know, search ads, or, you know, those keyword lists that you’re creating, trying to make sure that you’re hitting people at the right time, or even the remarketing ads where people have visited a page, and then they’re shown an ad, because of some sort of a pixel, if you have an idea of the keyword and the content, and the topic and all of those things, and it all aligns, then it helps people find you, it helps you answer the question that people are asking. And they’ll keep coming back to your website, to your content to your company, by your services. That’s really the goal of all of this is to help people find you, when they’re looking for something. I mean, Chris, you’ve shown statistics around how much content is created and then published on the internet. And that’s what you’re fighting against is not just your competitors, but the entire, you know, body of content that is created on the internet, because it may not just be your competitors that are talking about, you know, data science, I mean, maybe you have universities talking about it, you’re not necessarily competing with them, you have, you know, other statistical programs that are talking about, you’re not necessarily competing with them. But now you are, you’re competing for that attention span on the internet.

Christopher Penn 12:19
And it’s really important that time be a factor because when somebody is not primed in the in terms of interests, they’re dedicating less time in their day to that thing, and you’re competing against the Pope, and the Falcon and the Winter Soldier, right, and, and, and all these other things that soak up time during the day. But when somebody is looking for machine learning training, they are carving out time in their day during that period of forecast time, where they’re saying, hey, if you’ve got a solution, I’m listening now, now is the time that I’m listening. And so the understanding and Bo forecasts, those periods of time, do really does help us say, okay, we can earn some time in that person’s head. But if it’s for, you know, seven weeks out, and that window closes, it’s gonna be a lot harder to get messaging through that person, because they’re not looking for it. They’re not receptive for you ever. That one thing where you learn about something, and suddenly you see it everywhere, because your brain is awakened to that, that possibility you’re seeking out, you know, what hand people write with, or you know, which thumb is on top when they when they close their hands. And suddenly, like, Oh, I know exactly what hand this person is based on which thumb is on top. And that ability for the brain to spotlight, the things looking for is so important, because when we have timing down, the spotlight is briefly on us on the products and services we have to sell. And then the spotlight moves on to the next, you know, show on Netflix or whatever. And we’ve lost our window. Now, here’s the challenge. Not only are we competing against time and people’s attention, we are also unable to forecast some things. There’s some stuff that just cannot be forecasted. If you were to ask me, you know, two years ago, hey, when’s the next pandemic? I was like? No. Right. And then one year goes like, it’s right now, right now is the next pandemic. And now a year and change into the current one. People are rightfully asking, what’s the next one? Right, what do we have to be concerned about? And we’re back to? I don’t know, because you can’t predict what has never happened.

John Wall 14:43
Right? Simple one literally.

Christopher Penn 14:45
Exactly. You can’t you know, it’s it’s one of the reasons why our friend colleague at Edison research Tom Webster is often saying, You can’t forecast present presidential elections. You can’t because no election is ever the same. It’s not the same candidates. It’s not the same Same voting base. It’s not the same political environment, it’s not the same budgetary things. Everyone has a unique Black Swan event. And the likelihood of you be able to forecast something like that is pretty darn low. So one of the limitations of predictive analytics is, you can’t forecast it. If you were to go to it. Here’s an example. You could see just how easy it is to forecast something in your industry. If you go into Google Trends, and you type in let’s put in data science course. We’re going to move out here to a five year timespan you can gently see there is there is some level of cyclicality here. There’s a decent ups and downs. And it’s you know, it’s moving along. If I put in, for example, SARS, SEO, SEO, v2. This is a one time thing, there literally is no data. So this is not something that I could have forecast because there’s literally nothing to work with here. On the other hand, if I put in holiday gift guide, like clockwork, this is a highly seasonal time, you almost don’t need predictive analytics, because all you got to do is look and see it is basically you know, in this case is the third week of October 3 week of October 3 week of October 2, the third week of October, 3 week of October, October, right.

Katie Robbert 16:29
Chris, you bring up an interesting point. So you just said you almost don’t need predictive analytics. That is some of the pushback that we get from the market is well, I know what the seasonality is, why do I need to predict the forecast to tell me? And I think one of the things that’s, you know, missing from this is the nuance. So yes, you know, when holiday gift guides are going to be going out there. But that’s such a broad term. Do you really want to be competing with every other, you know, consumer company in the world, publishing their holiday gift guides on the third week of October?

Christopher Penn 17:05
There’s there’s definitely that nuance, there’s also granularity. So we can tell based on the data here, literally just eyeballing for this particular very, very, very seasonal thing, exactly what’s going to happen, right. And again, this is a case where you probably don’t need a whole lot, if I shorten this just to gift guide. Now, we’ve got, again, some seasonality here, but then there’s other periods of time where there are there is interest, then, yeah, it’s great. If you’re a retailer, and you only plan on selling things, two months of the year, this is a great forecast. If however, your business demands that you make money the other 10 months of the year, this forecast does not help, right? Because, yes, you needed to get the holidays, right. But you also need to get everything else, right? Because you’ve got to make money in time. And you could not say to a retail CEO, yeah, you don’t need to do anything January through September is just take that time off. Right? That’s, that’s unrealistic. When you start to look at other things like a fusion detection. Now you’re back to Well, is there a seasonality? Yes, there probably is. But it’s, it is so mixed up in all kinds of other stuff that you actually do need technology to find it, you cannot eyeball it, there is no one obvious event. So yes, if you’re lucky enough to be in an industry where you only sell things once a year, at a very specific time of year, great, you don’t need predictive analytics. If on the other hand, you need to make money the other 11 months of the year, and you have a search trend like this, then you probably want some machine help.

Katie Robbert 18:47
Now, you know, if I were someone who specialized in intrusion detection, I’d be like, Huh, this doesn’t tell me when I should be creating content. And then of course, you always want to make sure that you’re getting all those variations. So you’re not just creating the same intrusion detection content over and over and over again, trying to optimize for the same keyword or promote against the same keyword over and over and over again. And I think that that is definitely where you start to get into that more advanced, you know, the search trends, because Google Trends is a really good tool to sort of like do that gut check. But I think you can only compare up to five keywords. And then your graph is going to be a bit of a mess. And it’s only historical.

Christopher Penn 19:32
Exactly. And this is also why you’re really hitting on the importance of that very first step in the process, which is, yeah, you think intrusion detection is the thing, but then when you look at the top 50 or 100 pieces of content, they get traffic for that. And you run that unsupervised machine learning analysis, you may find that actually, it’s not the thing, right. So in the case of intrusion detection, you may find that IDs might be The term, you probably also see things like firewall in there that is not semantically the term and this is one of the bones I have to pick out, I do have a bone to pick.

Katie Robbert 20:09
I knew in search

Christopher Penn 20:10
with search tools, which is they’re very narrow in their linguistic reach, right? If I go into a search tool, I put it intrusion detection. And I say you’ll give me some keyword ideas. It’ll give me intrusion detection software, intrusion detection system, intrusion detection, service, and so on and so forth. It won’t say firewall, right? It just the word firewall, even though if you have domain knowledge, the two pieces of hardware are very closely related. And in a conversation with a a cybersecurity professional, they will say intrusion detection and firewall, probably in the same breath, a fair amount of time. You’re not going to see it unless you do that unsupervised learning to say, Okay, what in that top content, what else is being said that I need to take advantage of. And so it’s really important that you not skip that step. One of the challenges that we have with with machine learning is that the machines will take the data they’re given and learn from it. Which means that the data that goes in has to be good. If you are dealing with Hippo problems. Hippo is highest individually paid person’s opinion. And they say, Oh, we only need to focus on intrusion detection, well, then you’re going to create a forecast that has a bias. And that bias is the opinion of that person, not what the data actually says. So part of using forecasts smartly, and well is being actually data driven. Which means you look at what the data says not what the boss’s boss, the CEO says, because they, they’ve been doing this forever, and it’s the way we’ve always done it.

Katie Robbert 21:48
Well, and I think that that’s definitely one of the challenges with introducing, you know, a new type of tactic into your marketing. Because you need to demonstrate that it’s working in order to get people to buy into it so that you don’t have that Hippo problem, where someone says, Well, no, I want us to be writing about this. And my thought leadership should be about this. And so one of the ways that we would suggest trying to introduce this is to do a small proof of concept. And so if you have maybe a month’s worth, so four weeks worth of that analysis. So Chris, if you can go back to that Tableau sheet. You know, if you cut this down to just look at the month of May, for example, then you have all of the different terms that you should be using just in the month of May, you can focus your content and your social posts and your email and all of the other digital channels mentioned, using this map, you would want to take your, you know, baseline data right before you start and right after you start just to see how your metrics have improved. And so Chris, we haven’t even talked about what kinds of metrics you should be looking for after you start using a predictive forecast. So what would a content marketer be looking for to demonstrate success?

Christopher Penn 23:10
It depends on your goals. If you’re doing search, right, and you want to measure organic search, traffic increases, we’re doing social media, you want to look, you know, clicks from social media, it always depends on what the mechanism of distribution is. But the point is that you should be seeing increases, if, if and when you use the technology and the end the data properly.

Katie Robbert 23:34
So we do have a comment that I think is interesting. So Tiago says some bosses won’t trust the analyst plotting the data is if that’s personal, and the analyst is trying to diminish them. And I do think that that is a very good observation. And that comes into, you know, some of the change management and how you, you know, introduce this idea. And that’s why we do suggest starting with a very small, you know, proof of concept to demonstrate that you’re getting better metrics. I mean, at the end of the day, if you can show your boss that you’re getting more traffic, which leads to more conversions, which leads to more money, they’re gonna have a really hard time saying no, and if they still say, no, then that’s a whole different issue. You know, but it does happen. And but I think that does a really good point that it is, there’s a lot of ego involved with using the data versus I just want to do it this way. You know, John, you spend a lot of time talking with our prospects in the community, what, what some of the feedback you get about, you know, using a predictive forecast, or what are some of the questions in terms of what the heck is this thing? Yeah,

John Wall 24:40
you know, a great point that you brought up is, you know, these managers that are like we already know, the seasonal cycle, you know, we’ve done this for 20 years, and I have seen that over and over again, where they can pick the cycle, but the one thing they can’t do is get, you know, 20 deep on the term list, like they usually have four or five of the terms that are dead on and know what they’re doing. But there’s always that Three or four terms that are not on their radar, you know, are not on their website whatsoever. And they’re just missing opportunity, you know, it’s related terms that could be driving traffic. And it’s just completely invisible to them. Because it’s just it gets so abstracted. There’s so many layers down in a lot of companies, as far as you know, what are the blog posts we’re going to do this week. So having this list of the terms and the times to do them, that’s the invaluable thing to, you know, to get to the client is that they’re able to know what to write this week, because they know what’s going to drive traffic over the next week. And then, like you said, closing that loop of being able to demonstrate that, yeah, here’s the five articles we did, you know, and here was how much traffic they generated over that period of time. And then two, I think this is beyond what we’re talking about today. But and then getting into a cyclical review cycle, you know, you’re gonna find that there’s content like that, that every year, you want to go and refresh before the cycle comes back, because you want Google to come back and reindex it as fresh content again. But yeah, it’s just a great way to keep driving traffic,

Christopher Penn 26:01
you bring up really, really important point there, you can use any time series data to forecast, right, you can use any data that is numerical in nature, and has a time element can be forecast. And the more granular it is, the better the forecast, which means that if you’ve already got really good data, say in your Google Analytics software, you can forecast Google Analytics traffic by channel, for example, by source medium, if you know, if you have a year’s worth of organic search data, or two years or three years or whatever, you can credibly forecast the next year for that channel. If you have your email opens and clicks for your marketing automation software, and you’ve got, you know, day or week level data, you can then use that to forecast for to say here’s what’s likely to happen for our email. One of the things that we did for one of our larger clients is we forecast the traffic that they’re likely to get from each channel at the beginning of the year, and then we help them plan to say, look, look, we’re pretty sure you’re gonna have a traffic drop in April, right? The forecast shows this, what are you going to do about it to keep that from happening? Right? What will you what actually take us? Okay, maybe instead of traffic dropping 30% in April of 2021, maybe we can get dropped 20%. Just trim off some of the impact. That makes a big difference. But again, people don’t think that of the data you’ve already got, how do you blunt the impact of things that aren’t working? And then on those periods of time, you know, if your Google Analytics status shows that you’re likely to have you know, more email opens in June 2021. Great, send more email that month, right? Because if people are opening it, why would send you know, send good email, maybe that month, that’s the month you run an extra sales promotion in your email because people are opening it. But people aren’t using the data you already have.

Katie Robbert 27:59
One of the, you know, very brief case studies that we talked about in our sessions when we talk about predictive analytics is the best and the worst time to send email based on the search trends for people searching for how to set up their out of office. And so we’ve used that data because when we found a Trust Insights, only just a couple of years ago, we didn’t have enough of our own historical data for email opens. And so now we do but at the time, we didn’t. So we would use search terms such as how do I set up my out of office and setting up out of office and out of office settings to determine the best and the worst times to be sending email because we want to be getting it right. And so we actually use that as a subject line, you know, the worst performing email, you know, this year, and unsurprisingly, it was around July 4, but sometimes it wasn’t exactly the week of July 4. And it really depended on when the hope that holiday fell of when people were going to be taking time off. And so I felt I felt like that was a really interesting nuance, because we could guess that people will probably out of the out of the office on July 4, but to really be able to nail down that timing, especially when you’re limited on resources, and you’re limited on budget is so essential. Chris, we have another question, which I think is a very good one. I know you’ve talked about this is how valid are the data from 2020 for forecasting? You know, because of the pandemic, everything changed.

Christopher Penn 29:38
Exactly. And so the answer is it depends. Right? It depends on what industry what sector and what the impact was to your business, right. Some industries and businesses and and consumer interests didn’t change all that much. Right. Some had massive spikes. We have one client that isn’t that in touch. travel industry, and they saw road trips take off like crazy while airplane, you know, air travel cratered. And so for them because we knew the volatility in that market, we said you cannot use your previous data for forecasting, it’s simply can’t be used, when instead you have to do is very short term forecasting video, narrow windows a time for weeks of back data to project to just a week forward, keep that window nice and tight, until you start to see some more regularity, you are starting to see that now. Others industries, you know, for example, a lot of the work that we’ve been doing in machine learning and AI, not a lot of change, right? There’s the interest in that stayed relatively consistent, because it was not something was directly impacted by the pandemic. And the big thing you got to keep an eye on and this is an important point about artificial intelligence is that you, the person have to be providing some of the domain expertise upfront, and then in the QA portion to say like, yes, this makes logical sense, or no, something went wrong in the data or something went wrong in our analysis, or to our last question. Yeah, the data simply is not reliable there. You know, I was looking the other day at mass shootings in the United States that there’s a data set for that, of course, and I was surprised that actually the numbers didn’t change all that much where they occurred changed, but not the overall numbers. It’s like, okay, what’s, that’s disappointing. But the day, it all comes down to looking at the data, so part of your work upfront is to do some exploratory data analysis and get a sense of ease candidated use for forecasting. And here’s one of the the catches and tricks. If you use the data, you actually got to break it. for forecasting in the future, you actually got to cause yourself problems. I’ll show you a funny example. This past week, I was at a Twitter chat. For CME world, the content marketing folks, a lot of fun, a good group of people. And I pulled in advanced the data for engagement. So engagements, number of followers. Obviously, there were favorites and retweets for the past few weeks, and then the total addressable audience. And you can see, again, you don’t need necessarily a degree in mathematics to look at this and go Hmm, things that, you know, kind of a downward slide here. So it’s okay, let’s interrupt this. There’s a blog post over on the Trust Insights website, if you want to read how we disrupted this. But the following, you know, this week’s data, look what happened, like all the metrics, opposite across the board, the experiment, we did work, now we have a problem. The problem is, if we are then using this data to forecast next month, next year and things and we don’t, and we’ll keep repeating what we did, from a tactical perspective, we’ve now introduced an anomaly into our forecasting data that we have to compensate for, or acknowledge that a forecast is going to have some accuracy problems. So it’s interesting, the so what actually disrupts the What happened? And what’s likely to happen. So just be aware of that when you’re doing forecasting that if you’re doing forecasting, right, you may have a harder time with it, because you’re changing the future as it occurs.

Katie Robbert 33:24
I don’t even know where to go with that.

John Wall 33:28
That’s right, where you’re on the beach, you’re charting your own future, as opposed to just sitting around waiting to happen.

Christopher Penn 33:34
Well, exactly.

Katie Robbert 33:36
That’s a really good point. And I think that one of the big takeaways, one of the big so what’s, you know, especially if this episode is that, you know, we as marketers, and as data analysts tend to get so focused on the what happened, the reactive that we forget about sort of that future planning of like, how do I get proactive with my data? And again, having a predictive forecast, like this is a very, I’m putting in big quotes, simple way to start to introduce that, you know, to your team, you know, obviously automating a lot of your reporting. So you’re not bogged down by that, you know, focusing less on the what happened, you can’t change what happened. And you know, but you can, as Chris, as you just mentioned, you can influence what is going to happen if you have some sort of an idea of what direction to move in.

Christopher Penn 34:27
Exactly right. And ultimately, if you do your job, right, you should not necessarily be able to predict all that, well, what’s going to happen because you should constantly be changing constantly be adapting to the circumstances, and making the best things. But what tends to happen then is that your marketing performs at such an optimum level, that the forces that will change and impact your marketing are not inside your company anymore, and you can start to forecast trends outside of your company. So Instead of looking at your Google Analytics data or looking at your Twitter chat data, you can now say, Okay, I got to look at consumer trends at the macro level, because I can’t influence the entire country, Alicia Taylor Swift, I can’t influence the entire country and what happens. So I can, I can rely on data further and further away from the data that I’m affecting, and still be able to build some senses of, of influences, right, you can’t change when the sun comes up, you can’t change the seasons. So if you’re in agriculture, you would still have some level of forecasting will be necessary outside of the things that you have control over. But you definitely want to make the stuff you have control over as optimum as possible. Even Rex you’re forecasting internally. To quote, The Terminator movies, there’s no fate but we make for ourselves.

Katie Robbert 35:51
This took a turn that I was not expecting.

Christopher Penn 35:56
So to recap, there are multiple forecasting techniques available and you have to combine a lot of different techniques together to get the most reliable forecast. There’s some things that can’t be forecasted. If it’s never happened before. Yeah, can’t forecast it. But the most important thing of all is you’ve got to do something with the forecast having it on a in a binder or on a PowerPoint does nothing. You’ve got to change what you do any parting thoughts, John, Katie?

Unknown Speaker 36:27
Come with me if you want to live.

Katie Robbert 36:29
I was gonna go with that. I’ll be back.

Christopher Penn 36:34
Extra to it yet. We’ll catch you next week. Thanks for watching today. Be sure to subscribe to our show wherever you’re watching it. For more resources. And to learn more. Check out the Trust Insights podcast at Trust insights.ai slash t AI podcast, and a weekly email newsletter at Trust insights.ai slash newsletter. got questions about what you saw in today’s episode. Join our free analytics for markers slack group at Trust insights.ai slash analytics for marketers. See you next time.

Transcribed by https://otter.ai

 


Need help with your marketing data and analytics?

You might also enjoy:

Get unique data, analysis, and perspectives on analytics, insights, machine learning, marketing, and AI in the weekly Trust Insights newsletter, Data in the Headlights. Subscribe now for free; new issues every Wednesday!

Click here to subscribe now »

Want to learn more about data, analytics, and insights? Subscribe to In-Ear Insights, the Trust Insights podcast, with new 10-minute or less episodes every week.

Pin It on Pinterest

Share This