In this week’s In-Ear Insights, Katie and Chris talk about AI applications and distractions. How important is it that you know the technology? What level of depth is necessary for marketers to make marketing technology work for them? Tune in to find out!
Subscribe To This Show!
If you're not already subscribed to In-Ear Insights, get set up now!
- In-Ear Insights on Apple Podcasts
- In-Ear Insights on Google Podcasts
- In-Ear Insights on all other podcasting software
Advertisement: Data Science 101 for Marketers
Do you want to understand data science better as a marketer? Would you like to learn whether it’s the right choice for your career? Do you need to know how to manage data science employees and vendors? Take the Data Science 101 workshop from Trust Insights.
In this 90-minute on-demand workshop, learn what data science is, why it matters to marketers, and how to embark on your marketing data science journey. You’ll learn:
- How to build a KPI map
- How to analyze and explore Google Analytics data
- How to construct a valid hypothesis
- Basics of centrality, distribution, regression, and clustering
- Essential soft skills
- How to hire data science professionals or agencies
The course comes with the video, audio recording, PDF of the slides, automated transcript, example KPI map, and sample workbook with data.
Sponsor This Show!Are you struggling to reach the right audiences? Trust Insights offers sponsorships in our newsletters, podcasts, and media properties to help your brand be seen and heard by the right people. Our media properties reach almost 100,000 people every week, from the In Ear Insights podcast to the Almost Timely and In the Headlights newsletters. Reach out to us today to learn more.
Watch the video here:
Can’t see anything? Watch it on YouTube here.
Listen to the audio here:
- Need help with your company’s data and analytics? Let us know!
- Join our free Slack group for marketers interested in analytics!
What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for listening to the episode.
Christopher Penn 0:02
This is In-Ear Insights, the Trust Insights podcast.
In this week’s In-Ear Insights we’re talking about, again, artificial intelligence and machine learning the fun things you can do with it, the focus that everybody has on it, and how folks might not necessarily be focused on what actually matters.
So we were talking with some friends last week about the different types of AI technologies and things.
And Katie, one of the points you made was the average business person, the average marketer doesn’t actually care how many hyper parameters a model has, or, you know, the different types of Transformers that are that are connecting things, all they care about is, is the thing gonna save me time, save me money or make me money at the end of the day, and what we see happening a lot in the press, the media, the attention at conferences and things, as people are still very focused on the technology itself.
And I’m reminded Clay Shirky had a famous expression.
technology becomes suicidally interesting when it becomes technologically boring when we’re no longer obsessed about the bells and whistles.
And now we’re like, Okay, well, what do we do with thing? So from your point of view, why, why is that we’re still focused on the bells and whistles, you know, that being part of the being human nature? And where should we be paying attention?
Katie Robbert 1:23
Well, I think we’re so focused on the bells and whistles, because it’s still something new people want to understand how it works.
It’s not like buying a new vacuum, like a new vacuum, even though it’s new to you.
And it might be a robotic vacuum, it’s still essentially at the end of the day a vacuum, it’s going to suck up the dirt off of your floor and deposit it in sort of a canister or a bag like the fundamentals are pretty much the same.
AI, in general, is still one of those like, Well, what does it do it? Is it a human? Is it my like, did you insert my brain into a computer? Like, what does it do? Is it thinking for me, and I think because there’s still so many questions, and we haven’t done a stellar job of demystifying what AI is, and a lot of companies don’t want to they want to leave that like Mystique around it to make it more interesting and attractive, which I think is a mistake, by the way.
People have a lot of questions.
So they’re obsessed with the technology.
So there’s the this is really cool, what does it do? And then there’s the Oh my God, is it going to take my job? Is it going to create a lawsuit for me, because we’ve seen AI go really, really poorly.
And so I think that’s why people are still focused on the technology itself with like, what does it do? Will it say something or do something or make some sort of a decision that will put my company in jeopardy? Or will it replace my job.
And so I think that’s why that’s part of the reason why not the only reason, it’s part of the reason why people are so obsessed with the technology itself is they want to understand how it personally impacts them.
And so what they should be focusing on is not so much the nuts and bolts of the technology, but yes, how it impacts them, but in a positive way, which is where my brain is at.
And so it’s like great, the more stuff that we can automate, the more stuff that becomes repetitive, the less work that is for me, and I can focus on what’s next with the business, I can focus on servicing my clients, you know, 100%, I can focus on, you know, making sure that you know, everything else that goes on in the business is neat and clean and tidy and buttoned up really nicely.
And I don’t have to worry about, you know, being in the trenches, you know, creating reports and doing things that AI could be doing for me, so I am team take my job AI because I have other stuff I need to do that you can’t do for me.
And I don’t care how you get it done.
Just don’t you know, create a lawsuit for me.
And you know, don’t spit out incorrect data.
Christopher Penn 3:59
When you you mentioned something interesting in there, that the folks that, you know, sort of have that Mystique around it, I would imagine a lot of the people who are promulgating that sense of Mystique are vendors.
Because if you figured out exactly how simple a lot of algorithms and models where you’d like, I don’t need to pay you a lot of money.
You know, you’re just doing linear regressions over and over again.
So there’s that aspect.
But the other aspect is the technology itself, is it that people don’t understand the outputs that it’s so despite whether they don’t understand the method by which it gets to those outputs or a combination of both.
Katie Robbert 4:37
I think it’s a combination of both.
So if we think about, you know, the work that we do and some of the reporting that we pull together, we still have to go into a long explanation around what the report itself is saying.
And that’s not necessarily AI based reporting.
It’s putting numbers on a slide.
And so I think there’s that piece of the education When you slap the term AI on top of basic numbers on a slide, it’s that increased level of anxiety of now I really don’t understand it because it must be really advanced.
But at the end of the day, it’s still a bar chart.
Christopher Penn 5:16
Well, yeah, that’s the one of the things I’ve struggled with is that the methodology is not transparent to the user, nor does it need to be.
When you look at an attribution model, that’s just a big bar chart.
Recalling the words of a past colleague, it doesn’t look any more expensive than, you know, a bar chart that was made by a person, right? It’s essentially the same thing.
Even though the underlying technique is different.
It’s still just a bar chart that says, you know, you’re really good organic search.
And so I’ve had the opposite problem of this thing doesn’t look expensive enough, right? It doesn’t look like even though the answers are more correct and more thorough to the average person, it’s not clear that it’s better.
And so you know, should someone could someone justify paying more for something, when it’s not clear to them that, alas, such attribution model bar chart looks like this, you know, at a Markov chain model bartra looks like this, hey, look, they’re exactly the same bar chart, like, yes, the numbers are a little different, but they’re not so wildly different, that you’d be like, Oh, my God, this is so much better than than what I’ve been getting.
So from that perspective, you know, sort of the opposite.
Like, how would you communicate to somebody? Yes, the thing that is more complicated under the hood, you don’t see that we’ve saved you that complexity from how do we still communicate that it is better, and that it is saving you it’s not saving you time, because you’re still just getting a bar chart, but it’s helping you make better decisions?
Katie Robbert 6:46
Think about all of it.
So when we onboard a client, a lot of what we asked for is, can we see samples of reports that you’re currently using, so that we understand sort of where your level of understanding is, and the types of reports you’re used to getting? And I would wager that 10 times out of 10, the reports that we get from them, created by either their internal teams or by other agencies, the first thing we look at a goat, well, this doesn’t help anything, because it’s, you know, it might be slick and polished, and have like, you know, the font is embossed, and has a drop shadow.
And you know, that’s great.
But at the end of the day, the data doesn’t tell me anything, and I can’t make a decision with it.
And that’s always been the soapbox, that we will get on and die on the hill that we will die on to say, I don’t care.
If it’s an ugly PowerPoint, like, we obviously don’t go out of our way to make them ugly, sometimes that just happens.
But if the data isn’t useful, if you can’t make a decision with it, then no, you just wasted a lot of money.
And so that’s really what it comes down to is, if the AI isn’t helping you get to data that you can make a decision with, it doesn’t matter, AI, no AI the data at the end of the day, you need to be able to make a decision with it.
And that’s, again, that’s sort of where I’m at with in terms of using AI not using AI.
It’s not that I don’t care how we get from A to B.
But I don’t need to be in the weeds with the details of how the AI pulls the data together versus how a person pulls the data together.
At the end of the day, can I make a decision with whatever’s being pulled together? Yes or no? That’s what I care about.
And that’s what people who are in a similar position should be caring about versus, you know, well, you know, when it does when it spotted fifth line of code, what is it actually doing? Like, there’s other people who care about that thing.
And that’s Chris, where you and I come together in our partnership, you care about what the code is doing.
And I care about what I can do with the data.
And so because I already know that you’re worried about what’s in the code, I don’t need to worry about it.
Because I trust your judgment, and whatever you’re doing.
Therefore, I could just focus on the output.
Christopher Penn 9:03
That really is an interesting question, getting back to, you know, kind of the vendor side of things in that.
A lot of what is out there, a lot of the different tools and things are built on a lot of assumptions, right? When we look at the average report that a client gets, or even reports we get from our tools, they’re built on assumptions.
And those assumptions are the developer said, hey, these are the things that we’re building into this product that we think you’ll need.
And a lot of times to your point, it’s like barfing data, right? Like, here’s a whole bunch of data that nobody asked for, you know, here’s your impressions for this and that, and so on and so forth.
It almost sounds like you have a lot of cases, the vendors aren’t necessarily sure what the client wants to they kind of try and build everything.
And then you end up with a system like Google Analytics, which has 510 dimensions and metrics because there’s always some obscure dimension that you’ll need.
For some use case, so should companies be thinking about building more of their own stuff that is suited to their to their actual needs, because again, a lot of these tools, they don’t present data in such a way that you can make a decision from out of the box, right, you have to do a lot of massaging of the data, you know, even to straight Google Analytics data, it takes a lot of massaging to get to the point where you can say, okay, the decision we want to make is, do we spend more money on Facebook ads or not? Do we spend more money on organic search or not? And it’s not something that the data tells you out of the box.
I almost wonder if, when we talk about AI machine learning, if companies really should be in the the planning to build their own models and stuff that’s customized to the decisions they specifically need to need to make, because concrete manufacturing company is going to make some very different decisions than, you know, a retail convenience store chain.
They just have to, and we’ve both been in situations where, you know, someone’s tried to shoehorn like a manufacturing payroll system into a into an agency.
It’s just a disaster.
So what do you think about that of companies saying, Yeah, we actually probably should invest in some of this stuff.
And make sure that it helps us make the decisions we need to make.
Katie Robbert 11:18
I don’t know that you need to immediately leap to we need to build our own because the decisions we need to make, we can’t get the data for what I would actually say in that scenario is companies.
I know this has kind of come as a big shock, Chris, you can probably see where I’m going to go with this.
But first of all, companies need to sit down and make a plan they need to do their business requirements gathering to figure out what the heck it is that they need.
So that’s step number one.
Now, what strikes me is sort of there’s this whole sort of movement.
But there’s this idea right now that content needs to be personalized.
Websites need to be personalized, you get needs to be personalized.
Every single person on the face of the planet is unique snowflake and has to have their personalized experience.
Why is that not true? For what you get from a vendor? Why aren’t vendors hopping on that bandwagon to say, you know, what, we offer all of these different things, we’ve created it so that you can mix and match and do all this stuff, so that you then the customer, the B2B company, buying my software, can then personalize your experience with what I’ve created, because you’ve handed me your business requirements of what you need, and therefore I can accommodate that.
So that to me, that would be a better solution.
If the things already built, if I as the customer can tell you, the vendor what I need, you, the vendor should be able to then personalize to me and my business, the things that I’m after.
So it’s a two way street, I need to tell you what I need, you need to accommodate that.
And then if you can’t, then I need to look at building my own.
But I think the immediate leap to build your own because I can’t find it in a vendor that could potentially waste a lot of time and money and resources, when really we need to be pushing back on all of the software vendors to say, cool, so you need to be more adaptable, and customize it and personalize it to what I need.
It’s data data, you can personalize, they just need you need you as the customer just need to tell them what it is that you’re after.
Christopher Penn 13:23
It’s interesting, because that is essentially what like Google Data Studio is Google’s like, we don’t know what you’re going to need.
So here you go, here’s a canvas for for building it.
And you don’t have to write the code to generate all the stuff you just need to, to be able to work with the interface as it is.
And yet, the number of people that express abject terror at having to use Data Studio or get, you know, do anything useful with it is Legion when you talk to people, you know, for example, with Google Analytics for it’s a BI tool is no longer a reporting tool.
Google’s made it quite clear you should be using Data Studio for recording and the number of people who are so angry.
So so angry at Google about this design, change it there’s a whole Twitter chain of people just ranting about the fact that ga for no longer doesn’t do what Jay did.
And yet what they’re saying Google and what we’re saying is, we don’t know what you want.
So here’s the ability for you to create it.
So it’s kind of this weird schism of on the one hand, the tools and the vendors and stuff that are out there are not giving people think that they need to make decisions.
On the other hand, when you give people an environment where they can make the things that are customized them, they don’t.
So what do we do?
Katie Robbert 14:39
That’s where companies like Trust Insights come into play.
And I know you’re laughing, but that’s really, at the end of the day, why a company like ours exists.
And so a lot of businesses, they’re either too busy or don’t have the skill set or don’t have the motivation and they just want someone to do it for them.
They’ll tell you what they need, but they don’t have have the capability to set up the system to do it.
So companies like ours, we can come in and set up your systems to be customized to do the thing, so that you can make the decision with the information so that, that at the end of the day is what we do is we help you make decisions with your existing data, we’re not manufacturing data out of nowhere, it’s your data, you already have it, you just need to put it together in such a way that you can do something with it.
And that’s where we can help you.
But you also need to be using a system, a software vendor, like a Google Analytics, or, you know, one of the many social media monitoring tools or you know, whatever the system is where you keep your data, you need to be using that and putting your information in it in the first place.
So there’s, there’s three parts to it, one, you need to know what you want.
So that’s your business requirements.
Two, you need to have a system that’s collecting the data, you know, so that’s your software vendor.
And three, you need to have a way to put together that information to make a decision about and that’s either your bi team or your analytics team or a consultancy, such as Trust Insights, who can put it together in such a way that it comes full circle to say, this is what I want, this is where it went.
And then this is the output.
So I’ve now gotten what I want, based on my requirements, and now I can see it visualized.
And now I can make a decision.
So it’s a full cycle of things.
And if you can’t hit those three marks, it doesn’t matter if you’re using AI or not.
Christopher Penn 16:33
So it sounds like at the end of the day, it’s actually a people problem, even though we’re talking about technology.
And you know, whether it’s it’s how fancy it is, or what it does.
It sounds like you’ve got a human being a problem again.
Katie Robbert 16:45
Yes, well, and so that, again, sort of to demystify, AI.
AI is a person built thing we built that we created it, it doesn’t it didn’t suddenly get dropped on the planet by aliens, like, we built it, we created it, we manage it, we run it, we review it, we monitor it, we tweak it, like whatever it is, humans are doing the thing.
So if you’re concerned about the methodology of AI, some person believe that AI can’t make decisions for data it doesn’t have it can’t make decisions that is not programmed to make.
If you’re concerned about the output, that’s a person problem, you need to go back and get in there and fix it.
If you’re worried that it’s going to cause you legal issues.
That’s a person problem, because somebody’s programmed it to be racist.
So like, it’s all it always comes back to it’s a person problem.
If you can’t adopt AI in your organization, it’s likely a person problem, because it stems from, you know, lack of skills to be able to maintain it, or it stems from insecurity of people not wanting AI to take the job, or it stems from lack of understanding of what the heck the thing does.
And that’s an education problem.
At the end of the day, it’s always comes back to a person problem, because even when we have AI, running a lot of systems and processes and businesses and all of these things, at the end of it right at the bottom of that little AI Rainbow is a person programming, it is a person maintaining it is a person updating it.
And I might be on a little bit of a soapbox right now, but yes, it all comes back to people.
Christopher Penn 18:23
So what I’m hearing is we just need to eliminate the people and we’ll solve these problems with AI.
Katie Robbert 18:27
You are definitely not listening to a word that I’m saying.
Because you still need people to run AI.
Eliminating the people actually makes the problem worse.
Christopher Penn 18:38
So from a a next steps point of view, if you have people who are obsessed with the technology, and you know that they’re they’re gushing over the latest transformer model or whatever, and you need to redirect them.
How do you how do you reprogram the human to be focused on the right things?
Katie Robbert 19:01
Well, I think it’s contextual.
So you know, Chris, I know that you are genuinely interested in the technology.
And it’s something that you know, you’d mentioned like outside of business hours, you’re just curious, and you want to know what it does, what it can do.
And sort of that’s how you get your ideas to bring back to the business is it’s a hobby of yours.
It’s something that you’re genuinely passionate about, I think in terms of the boundaries of the business, then there needs to be a, you know, more clear cut.
Here’s where, here’s the amount of time you can do for professional development within the business, you know, construct.
Here’s where we need you to just get the work done.
And so I think it starts with a conversation.
So let’s say you have someone on your team who is not getting their work done because they’re really too focused on trying to find some sort of an AI solution and they’ve been going down this rabbit hole for a few weeks now when really they just need to get the report done.
But they’re convincing you that if I can just find the right AI, it can do the report for me.
It starts with the conversation of, you know, what is the problem you’re trying to solve? What do we have now? How can we build this an incremental step so that maybe you’re not starting on day one with AI doing the whole thing, but maybe AI does 10% of it.
And maybe the next time it goes 20%, and 30%, and so on, so forth, up until the point where you’ve built on it so much that you understand every piece of where AI fits into the reporting process, and then you’re comfortable with AI taking it on, and it gives that person a little bit more.
You know, it gives them the opportunity to really work with it.
But also, you’ve refocused on to say that you really just need to get the work done.
You know, so that’s one possible solution.
The other one is really sort of that sort of that company wide mentality of is AI the right step for us.
It might not be AI doesn’t work for everybody.
And it’s not the right move for everybody.
And it really depends on the type of business that you’re running, the type of employees that you have, the kind of output that you’re trying to generate AI might actually just get in the way, because you’re either too small, or it’s just not what you do.
Christopher Penn 21:12
That’s true, particularly once you start looking at, you know, the computational costs of some of the services, you know, the the new language models that we’ve been playing with, they require a lot of horsepower.
I mean, we’re talking probably a room full of horsepower.
In some cases, you can rent it from other companies.
But at the end of the day, it’s like, does this generate substantially better results than what we’re doing? presently? Sometimes it does.
Sometimes there is a, you’re at a point where it’s like, Okay, the first dress that this thing is spitting out are substantially better than what we had.
And in which case, like, okay, the you could make a case that there will be a positive return on investment there.
In other cases, like, not really.
You know, I was playing with Google’s Pegasus summarization model, it’s a incredibly processor intensive thing.
And at the end of the day, the results are pretty average.
Now, they’re, they’re not so much better.
They like, wow, I have to figure out a way to build this and everything we do, like nope, you actually would be, you could take the money you’re gonna spend on the processor farm and hire an intern to do exactly the same thing and probably get better results, even with someone who’s a minimum wage employee.
So Well, that was helpful, I think in terms of understanding where, where AI fits in how we should be thinking about it.
I think, if you’ve got questions about where it fits in, that’s something we do too is if you just want to have a chat about how does this is this even a good fit? easiest place is to pop on over to our free slack group go to Trust insights.ai slash analytics for markers got folks 1800 some odd folks talking about analytics and marketing and other things all day long.
And if you want to catch this show, on some other channel, head on over to Trust insights.ai slash gi podcast we can catch on the channel of your choice.
Thanks for tuning in.
We’ll talk to you next time.
Need help making your marketing platforms processes and people work smarter.
Visit Trust insights.ai today and learn how we can help you deliver more impact.
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!
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.