GOOGLE SEARCH CONSOLE FOR MARKETERS 4

{PODCAST} In-Ear Insights: Nontraditional Data Sources

In this episode of In-Ear Insights, Katie and Chris discuss what to do when you data sources off the beaten path. What if the data you have doesn’t answer the questions you have? Where do you go to get such data? Listen in as they explore Mastodon, cryptocurrencies, and so much more.

[podcastsponsor]

Listen to the audio here:

Download the MP3 audio here.

Machine-Generated Transcript

What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for listening to the episode.

Christopher Penn
In this week’s in In-Ear Insights we are talking about alternate data sources for either highly regulated or completely unregulated industries. Recently we were doing some work just trying to figure out is cryptocurrency being used in a certain industry sector and found out through a whole bunch of digging that no because it’s extremely hard to get set up using crypto currencies like Bitcoin and Ethereum and Stella lumens and all these different currencies but brought up the discussion topic of how do you go about getting information in order to do predictive analytics to do planning strategy when the sources that your normal go tues don’t have the information? So Katie, you want to expand and expound upon this really interesting challenge?

Katie Robbert
Yeah, I think one of the things that that we as marketers run into, and we all do it myself and you included is we sort of get, we put on those blinders. And we kind of get stumped when our normal sources such as a Google Analytics or an SEO tool, or our social media data just aren’t providing the answers. And it can, if you’re not familiar with working in a data set, if you’re used to just like looking at a dashboard or pulling down, you know, your rent your regular, you know, engagement and likes metrics, working with an outside data set can be a little intimidating, but what you and I are finding is that is the only way to really get the full story of what’s happening. And so, we were recently, you know, just randomly chatting about a couple of things in our slack channels. And you came upon this idea to investigate, which you know, often happens, you know, you’ll take a look at data on A Saturday night of data that we wouldn’t normally look at for the company. But that might answer an interesting question. So you’ve used third party data sources such as census data, Bureau of Labor Statistics, you know, you’ve looked at data sets that one wouldn’t necessarily think to pull into their marketing set. So I guess I would pose the question back to you of where does one even start to think about what other data is out there?

Christopher Penn
It depends on what you’re looking for. So a couple of years ago, we were doing some work for was it was a Pfizer, the the pharmaceutical company and one of the things that was a discussion point with them was they cannot use certain marketing tactics they you know, it is extremely difficult for them to use Google ads or Facebook or any of these things to market their products. They cannot for example, they cannot mention their product by name. What a what a difficult challenge for our a marketer to be able to say like, yeah, you cannot mention your product by name, you can only talk about what they call disease, state awareness. Here’s the problem, you’ll learn more about the problem. And by the way, we have the solution for it in the form of this pill or that pill. And so when we talk about understanding a topic, one of the first places to look is what people are talking about and what people are searching for. So search tools are relatively agnostic, although they tend to work with the main search engines, Google thing, etc. But there are engines like duck, duck go, for example, that are sort of more fringe search engines that don’t have any the features of Google but index places that Google frankly will not return results on. Things like illegal drugs, things like sex worker services, all this stuff that the mainstream technology companies have said, we don’t want to be a part of this world. The nature of the internet and humanity is that there’s a part of humanity that will always be want to that and They will build services around it. So you to answer your question, the first place you have to think is what is it that people search for? And is there a search engine built to handle that? And then what are the things people are having discussions about? And can you find a social network that addresses that. So, for example, Twitter bands, a large number of things, but this is app and service that came out two years ago, called mastodon, that’s a decentralized Twitter essentially, where you run your own server that hooks into a federation of other servers. And you essentially provide your own little mini communities while having the greater community involved. It’s a bit like slack in some ways. But imagine if there was also a general all Slack channel, right? We’re very noisy channel, it would be there. Knowing this exists, means you can probably find a niche for pharmaceuticals or whatever, and be able to extract data from it. So that’s where I would start

Katie Robbert
That’s interesting. I didn’t even know that mastodon was a thing. I always learned something new from you on these Monday morning recordings. You know, this is going to come as no surprise to you. But the first thing I would say is, what is the question you’re trying to answer? And so what’s the plan? You know, so if you’re, if you’re struggling with not getting the answers from the data that you have, normally that you’re normally accessing, then I would say, really start to pick it apart. And so you know, let’s say you want to know, you know, incomes, well, you’re going to get some of that information inferred. But are there other places where people are reporting their annual income? The answer is yes. And so it’s looking at the Bureau of Labor Statistics. It’s looking at census data. Maybe it’s looking at salary. com and pulling from their API. Hi, maybe it’s looking at Glassdoor. It depends on the context in which you’re trying to get that salary information. But there are a lot of other ways to answer that question. And I’d say, the best way to start thinking about where you get that information is really understanding what data points you need in order to answer the question is a lot like a KPI map, where here’s your goal. And then here are the metrics that you need in order to answer the question of, are we meeting our goal, it’s essentially the same thing.

Christopher Penn
It is. And the other thing is that we have to remember, at the end of the day, we’re dealing with humans and humans for good or ill are very, very predictable. They will tend to do the same things or try to find outlets for whatever it is they’re interested in. And if one isn’t there, people will create it. So understanding that there are parallel lateral data sources that can give you the same answers. There was a fascinating study about a year and a half ago, someone used the imagery from Google Street maps, you know, the Street View where they, they drive that silly looking car down your street, you know, takes photos of everything so that you can drag it up. They use that and they identified the vehicle makes and models in every driveway. And it found that was a better predictor of household income than certain other data sources. Because someone who has, you know, an old jalopy versus someone who has like a Porsche in their, in their driveway. Those people, by default have different levels of household income. There’s people that simply cannot afford, you know, a Tesla Model X vehicle. And so there’s a lateral data source in a good part of our work. And the work that marketers need to be thinking about is what runs lateral to the, to the data that you’re trying to get hold on if you can’t get ahold of that data. So say you’re a pharmaceutical company, for example, and you’re trying to gauge how probably Are opioids and other substances in the marketplace? Right? Yeah, you’re trying to figure out how much of an illegal market is it you can’t go just ask Google that, right? You can’t just go and serve Twitter. But you absolutely could if you were motivated, and you knew that it existed, and you get that from talking to customers and, you know, continuous learning, go to someplace like one of the black market exchange is on the dark web and just scan it like you don’t need to be doing anything illicit you can copying pasting data is pretty straightforward, and it’s totally legal. But you could see like, what are the illegal drugs being listed? In what quantity by day and now you have a data source that you can use as a benchmark to say like, yep, this is this particular thing is growing, this thing is shrinking, etc, etc. So how what runs parallel to your thing, what runs parallel to your line of inquiry and that, like you said, goes back to do you have a Goal, do you know what it is you’re looking for? And what else is related to that thing?

Katie Robbert
What happens when, so let’s say I am, you know, your everyday marketer, I don’t have coding skills, I don’t have access to AI tools or scrapers that can pull these conversations. And I myself am not very, you know, comfortable with unstructured data. What What advice would you have for a marketer who is seeking more information, but doesn’t necessarily have the tools to manipulate the data to add to their will?

Christopher Penn
You have really two choices. One choice is you can buy those skills can hire in somebody you can hire and Trust Insights, wink wink, nudge nudge. Or you can build those skills. You can you can learn those capabilities. There’s absolutely Nothing that, for example I do that another human being can’t learn. In fact I in terms of sophistication, and overall skill, I would put myself on sort of the lower end of what is possible, right? I don’t have multiple PhDs and advanced mathematics. I’m more of a hacker than I am a statistician. You don’t need advanced skills, like crazy PhD level skills to do some of the stuff What you need is never ending curiosity and absolute stubbornness to find the answer to explore every possible Avenue and to say, Okay, I need to learn this to do this. Okay, fine. I’ll learn that so that I can do that. It’s like, if you absolutely positively wanted to learn how to make sushi, you would learn how to cook rice, you would learn how to cut catch fish, you learn how to cut fish, etc, etc. It might take you some time to get there. But you can get there. It is a doable and noble thing. That’s if you have the time if you don’t have the time, you gotta buy it. I mean, there’s there’s no substitute for that if you don’t have the time or the money, figure out how to make one or the other.

Katie Robbert
What advice would you have for someone who is who does have the time. But they’re looking at, you know, third party data sources that they’re not familiar with? What are some of the red flags that people should look for? Before proceeding with some of these data sets?

Christopher Penn
Like so many of my analogies, it comes down to imagine it being a cooking analogy, right? If you have an ingredient Do you care where you bought the ingredient from? So out where we live there’s sort of this discount surplus free chain called Big Lots that sells stuff that is almost always expired. is you know from random misplace events and things like that it’s kind of an odds and ends store. You don’t buy your groceries there, right? That’s not a trustworthy source for fresh food. It’s fine if you like buying spam, because spam never expires. But you wouldn’t buy your main ingredients that you would buy your main ingredients from a more reputable vendor. The same is true for data. If the vendor isn’t reputable, probably don’t get your data from there. So like you were alluding to earlier, things like the Bureau of Labor Statistics, government websites, academic institutions, these are all places that generally have high quality ingredients, high quality data, major search engines, so SEO tools, things like that, that as you get away from that and start going to the more fringe edges of the web, the information naturally becomes less than or less trustworthy. And I’m not sure that I could teach somebody other than by example, like this is trustworthy. This is not trustworthy, although I will say if you have not already done it, you actually positively should go and download Google’s search quality rating guidelines. Google hires hundreds of people every year to look at samples of web pages and rate them on their on things like expertise, authority and trustworthiness. And that’s the training data that they used to train their algorithms. And in that hundred and 67 page book, they say, here’s what we look for for trustworthiness are the things that credentials displayed. His authorship clearly defined his contact information available, and there’s all these markers that they use as as proxies for trustworthiness. Now, that doesn’t mean that it’s foolproof, but certainly people who are cranking out low quality crap or not going to go to the extent extent, like one things they say, for high quality content is there should be if it’s a, if it’s a survey, there should be an extensive disclosed methodology, where the data came from time frame, all the stuff that you’d expect to see. Crap data from crap marketers typically has none of those things. So I’d say that’s a really good starting place to learn. How do you? How does Google when the most sophisticated organizations in the world make the determination about what’s crap and what isn’t?

Katie Robbert
I would add to that, and something that you don’t necessarily need any training for or supplemental software for is, any good data set will have some sort of a methodology disclosure on the website or within the data set when you download it, which will basically tell you, here’s how we collect the data, here’s the date range, here’s what’s included. And it’s that checks and balances. Because what you don’t want to have happen is you come across, you know, Joe’s data warehouse that promises you know, every kind of data that you’ve ever wanted. And then once you download the data, you find that it has, you know, gaps of date. So it’s like it’s missing April and it’s missing July. But it says that it’s otherwise collected completely consistently, and that’s probably your problem, not theirs, you know, so there’s definitely those types of things that you can look for within a decent quality day. Set. Now if it’s disclosing up front that there are missing data points, that’s something that you can work with, you can say that this is an incomplete data set, but it’s a sample of whatever it is you’re looking for. So there’s ways to use that information. But one of the first things you should look for is a methodology statement to say, this is how we collect our data. This is what we do with it. This is where it comes from.

Christopher Penn
When you were doing opioid research, how did you differentiate truth and falsehood when you were interviewing people who were using drugs illegally? How did you tell like, Okay, this guy is telling the truth, or this guy is like full of it.

Katie Robbert
So that is a easy question to answer, but it’s not going to provide the answer that you’re looking for, for this podcast. The system that I worked with was a piece of software that had gone through clinical trials. And we had actually done the trial of I’m going to interview you in person, versus I’m going to ask you the same question in front of people. computer screen. And what we found through the clinical trial was that people were more likely to be honest, sitting in front of a computer screen than they were answering a person. And so what we did was we then commercialized the clinical trial product and put that into substance abuse facilities as part of the intake tool. So the idea was that when people were answering questions about their opioid use and abuse, they were being truthful with the computer. So we didn’t necessarily say is this true or false? Because there were checks and balances built into the survey itself, where we go ask you the same question two different ways. If we got two different responses, we knew that you weren’t being truthful verse. So we would toss that out, versus the people who answered the question the same way consistently. So those kinds of checks and balances were built in. The actual issue that we ran into was the sample size of the network changed quarter over quarter because The substance abuse facilities would go on and off the network, depending on whether or not they were providing data. That was the bigger challenge was an inconsistent network and representative sample across the United States, because there were pockets of the United States that just weren’t using the product, they just weren’t using the assessment tool, or suddenly, California would come online, and then they would drop out the next quarter. So that was actually the bigger issue was the representative sample versus the actual truthfulness of the quality of the data.

Christopher Penn
But that sounds not dissimilar to what marketers face when they are dealing with a lot of their data sources, particularly anything that involves social media data. A, a topic may come for discussion that fades away very quickly, and then it comes, you know, comes back. I know certainly on the political side of what you just said is very, very relevant because people will give a different answer about like who they’re going to vote for in person because they may be a social stigma for supporting one or another but they will answer differently if they think they’re talking to a machine. We know this from the book, everybody lies, people will Google for things they would never say out loud to another human being. So that I think, even though on at first glance is not something that seems relevant, it is 100% relevant to the way market is collecting data back to your original question.

Katie Robbert
Yeah, it is interesting to think about it in that context. I hadn’t sort of put two and two together in that sense, but it does it make sense. People are tend to be more truthful to a computer than they do face to face. I mean, you see it in very ridiculous examples of people starting fights on the internet, that they wouldn’t start in person, like, would you say that to my face? No, but would you say it from the comfort of, you know, your desk chair and your keyboard? Absolutely. And it’s just those ridiculous things, but I feel like that’s getting a little bit off track of where we started, where, you know, the challenges that marketers face is that they may not be getting all of the answers. Is to the questions from their usual data sources. So don’t be afraid to step outside of what you think is, you know, a usable data source, you may find that you’re getting answers from places to your point about the, you know, the Google Street View, and the inferred incomes, that you’re getting answers from places that you wouldn’t necessarily think you could. So don’t be afraid to step outside of that comfort zone and at least explore what’s possible at least do a little bit of research to see have I turned every rock over or do I have what I have?

Christopher Penn
And I would say as a final summary point, make it your homework once a month to try and find a new data source that you haven’t worked with in the past and see if there’s a there there if it is a useful source of data for your marketing or not. As always, you can find out more information about all this stuff at our website, Trust insights.ai. Please make sure you subscribe to the newsletter, our YouTube channel and And of course, if you’re listening to this podcast because someone just forward to your link, you can subscribe to the podcast there. Thanks for listening and we’ll talk to you next time.


Need help with your marketing AI 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, INBOX INSIGHTS. 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 episodes every Wednesday.

Leave a Reply

Your email address will not be published. Required fields are marked *

Pin It on Pinterest

Share This