In this podcast we discuss the difference between emotion and sentiment, how emotional measurement can drive major business decisions, and how brands use Canvs to understand their customers better.
TheNextCMO’s latest podcast is with Erinn Taylor the EVP of Product and Platform for Canvs. Erinn is a leader in gathering actionable insights through AI-driven analysis and held previous roles at Dynata and Critical Mix. In this podcast we discuss the difference between emotion and sentiment, how emotional measurement can drive major business decisions, and how brands use Canvs to understand their customers better.
And don't forget to check out Plannuh, the smartest way to build, manage and optimize your marketing check out our website
Kelsey Krapf: [00:00:00] Welcome to the official podcast of The Next CMO hosted by Plannuh makers of the first AI driven marketing leadership platform for quickly and easily creating winning marketing plans, maximizing budget impact and improving ROI. The Next CMO is a thought leadership podcast for those that are CMOs or wants to become one.
My name is Kelsey Krapf and I'm the senior marketing manager
Peter Mahoney: [00:00:24] and I'm Peter Mahoney, the founder and CEO of Plannuh and welcome to the next CMO podcast.
For this week, we have Erinn Taylor, the EVP of product and platform for Canvs as our guest, Erinn has changed the direction and vision of Canvs, and we're excited to have him on our show today to learn more about his actionable insights through AI driven emotion
So great to have you here today, Aaron, and maybe we can set the table by having you tell us a little bit about you and their Canvs story.
Erinn Taylor: [00:01:06] So sure. I'm a 20 year veteran of market research and SAS based product development. I first worked with companies such as decipher focus, vision, critical mix, and Dianetta so market research firms and software firms in the industry. And during that time I came to know Canvs, which is a sentiment analysis platform.
We think we do sentiment analysis better. That's really what it comes down to. We take text analysis and also from multiple sources that includes from. Social media such as Twitter or Facebook or YouTube, like open-ended texts such as from surveys or survey prompts, or even book reviews and so forth.
And we analyze those four topics, emotions and some traditional market research analysis to produce insights because we're an insights company.
Peter Mahoney: [00:01:52] How would a brand or a marketer in general use a platform like Canvs to try to understand their customers better, tell us a little bit about some specific applications of the technology.
Erinn Taylor: [00:02:06] . If I see you use a couple examples, so again, me and entertainment, and we do have customers like Disney.
One way that they utilize platform is that they have talent sometimes says things right. And it could be controversial. So they came to us and basically they knew something was in the post and they knew within an hour afterwards, they need to make sure that there's not going to be any sort of blow back.
Or any sort of dammit like damage control. So they need to know right away what people were saying about this comment that was made on a talk show. So they were able to use that and identify that actually 98% of people were commenting on that were saying, they agreed with it. They backed it up.
They liked it. So there wasn't a damage control and they were able to go into a meeting saying, this is how people took it. And. This is how we can run with that. So that's one example. Another example we did some research with company called LRW and they're working with the brand red bull. And they were able to, it was their PR programmatic brain science department.
They were able to utilize and look into your content and see that while most people were saying, they really enjoy the product. There's a group of people that were angry and they were like, so what are these people angry about? And they will drill it down to it. Canvs allows you to import also closed ends so you can filter and find segmentation and see how the differences between them and compare them.
And they're able to find that within this group of angry people, There was a higher level of angry people in mass smirch, which is like a Walmart or Costco versus a C store or a grocery store. So they're like, why are these people so angry? And these locations digging into it, they were able to find out that.
And this is a insight that they wouldn't have found otherwise. They wouldn't have noticed that angry wouldn't have noticed that it was related to the segmentation and then into it. What were they actually saying? And found out these consumers were stressed because they were going in these stores and they wanted to get in and get out quickly.
They couldn't find the product or when they got to the location that there's so many people in the aisle, they just didn't want to fight through all the people to get to the red bull. So what that created action for LRW and they're able to go back to. To rebel and say, here's some things you could do.
So they were able to do an end cap, put it closer to the front of the store, allow people to get in and get out and actually improve the the sales cycle for red bull and these mass Mar stores and reduced the anger for their consumer.
Peter Mahoney: [00:04:18] So do you have to be a big brand to take advantage of these kinds of capabilities or can smaller kinds of smaller companies take advantage of this kind of emotional analysis?
Erinn Taylor: [00:04:29] No, you don't have to be big company.
So we work with boutique firms. We work with large firms. We work with brands. It really doesn't matter. I think an example of that probably would be, I know I just went and got my little change to regional oil change from they probably five or 10 stores right afterwards, I got a text message that says here's an NPS study, zero to 10 scale.
But what's been missing is like a reply and say is a 10, this is a one, or this is a five. But I don't know why. So if they just did one more question as to why I wouldn't be able to answer specifically why I felt that way. And we have some examples of that too. We did another case study where we were looking actually in this case is related to historical data around people who went to see theatrical went to movie theaters, and we found that We asked a similar question to all these people.
There's a, this was a five point scale. They all said, yes. I definitely go to see this. This is a five, five, five. The followup question was, why did you want to go see this content? And then the answers in there were wildly different. Some were like, I love this movie idea let's concept.
I was like, I don't know. I just, maybe didn't want to go or didn't want to miss it or something. But so the nuance inside there was very different related to it and it creates some sort of action. Another way that we've seen it used to is like the voice of the consumer. So if we think about it, if like you check out Amazon or you check out your new little it could be like a small retail store, whatever you check out, there's a follow up question.
It's No. How was this experience? And to be a five point scale or a 10 point scale again, asking why, if I say I loved this experience, right? I love this experience. It might be important understanding why do I love it? I love the prices. I love the availability. I loved how easy it was to purchase work.
Vice versa. I hated it. Why? Maybe it was just because you didn't have to think. I wanted. Whereas the rest of the experience was great, but without that nuance, we don't know where to spend our time and energy. Like how do we know how to improve our process? If we just know that you didn't like it?
So th it's the nuance, there doesn't matter how big you are or how small you are in terms of an organization size, that sort of information. Can feed back in the product. And again, it's being more empathetic understanding what your consumer wants, can help you, change the ROI in terms of your overall or just the bottom line in general.
Peter Mahoney: [00:06:36] So talk about the amount of data that would be useful , for you to look at, , to make some meaningful kinds of insights come out of it.
Erinn Taylor: [00:06:46] Oh because we are trained on all this content that we have over the, that past seven years. We have a lot of we have a Corpus of information and we're able to understand the way people speak.
So that means we don't have to train our content. So traditional systems require you to take a whole bunch of data related to this, that, or the other, in order to start understanding what you're seeing in the content, and you do some sort of a NLP around that to cluster and find content. We've already done all that work for you.
It could be 50 really nuanced records of content. Like in a qualitative study, it could be 150, open ends. It could be 5,000 open ends. It could be a hundred thousand open ends or any sort of content, like book reviews, et cetera. So the content really doesn't matter as long as we can find the trends within it.
What people are saying as long as there are trends within it. So if I took content from various different sources, I had nothing to do with each other. You can have a whole bunch of topics as opposed to. Some sort of grouped together content, like a open-ended question in a survey around how you like your car.
We're able to find topics, emotions. If there are emotions in there, sometimes people say things that are not emotional, like I'm going to buy this. There's no emotion around that. That's an action. So there's just different nuances in there. And that's also one of the reasons we use. What we call a reaction rate which is like how engaged are people with this actual question or the content?
So if there are people that are really saying that really excited about it, there's gonna be more emotion in that. Versus if the, if it's not very exciting or there's nothing really going on with it, it's just, yes. I went there, we gauged like the different types of events and different types of things based on how much emotion that person actually puts into it.
Peter Mahoney: [00:08:21] And , how would you tell in a case like that? Cause I can imagine that, unless you put a bunch of exclamation points behind that statement, I'm going to buy this, that, that could be I'm going to buy this. I'm super excited or I'm going to buy this. I am a robot is it possible to discern some more information?
From that limited content, or do you need a little bit more to try to understand what's happening and , with terse texts like
Erinn Taylor: [00:08:46] that? In many cases we do find based on the extra things that they do. Again, we do curate that information. If we see like someone's using a bunch of explanation points, we can extend that right into with the term.
I really love this, or I. I'm going explanation of explanation for right. We can identify that and utilize that as an emotion. Whereas, just ongoing has nothing there. So to answer your question, yes. Explanation points, emojis. Different ways of saying things that we know to be like a common way that people are excited about stuff, again, a natural way to speak. We're just slang. Like we say things a certain way that we get used to, and those can, and we can identify those inside campus.
Peter Mahoney: [00:09:26] So from your experience, obviously you personally in your company has a great body of experience when it comes to understanding this kind of data. If you look across the marketers that you're working with on a regular basis are there questions that marketers should be asking of their customers through this kind of analysis that you see some common threads that people just aren't doing?
Just the obvious one or two things that everybody should be doing by leveraging these kinds of capabilities?
Erinn Taylor: [00:09:59] What we're seeing mostly is that due to a nurse shot people just aren't asking the questions. Because it's traditionally, it's either expensive. It takes too long, three to five days to get some sort of analysis back from it.
They don't know what to do with the data because there's too much of it and they can't dig into it and understand. So really just asking the questions is what we really want people to do. Why do you feel this way? What about this? Do you like, how is this something that will make your life better? By asking those questions, it, again, adds more to the overall story and you can actually hear what the consumer is saying.
And then we can create action from that. That context. But if you don't ask the question, there's nothing to actually utilize from that. And there's only so much you can get from a close ended question because we're now eating people and telling them what they think, what we think they're going to tell us.
Here's a whole list of things that we think you're going to say choose from them. But we're what we're missing is really what they're feeling, really what their intent is, because we just don't know. That's the one thing that we're trying to let everyone know that. The inertia doesn't have to happen anymore.
We can compress this data down to something that's near instant that you can utilize right after the data or the research has happened. And it's in a format that's easy to use, easy to digest. These are the filter, and then it can be, you can even tune it further. We have ways that if you have your own Content that is unique to the way that you operate or your own features or things like that.
Our system can actually learn about those things. And when you're, if you're doing like a regular feedback, a feedback survey or a tracking study, things like that, it can actually get smarter over time and actually start producing those things for you. Benchmarks, and if people are talking about certain things were less or a new feature that's been released, we have a game company that uses it to find out how their game features are.
Our landing with our customers, which really helps them out to understand are we spending money in the right places? Was this a great thing to do? And you can also listen for other things like marketing and your, in your, either in their game. Like they do positioning products in there, right?
Like you might see a Kip of cocaine or a Pepsi can inside of a video game. Who paid for that? Just I'm sure someone paid for that. So is that working? Are customers talking about it? Is it being placed in position properly? So knowing that allows for a marketer, it's did we spend our dollars the right way first off and or as the company who's selling the advertising here, look, people are talking about the product that you put in my video where my game or my, whatever.
So why don't you spend more with me? Yeah, that
Peter Mahoney: [00:12:20] makes fascinating. I tell you the one thing that I find vaccine around a trend in, in marketing around just deeply embedding a product placement with influencers, as an example, is trying to figure out how in the world do you measure that thing. And this is a really interesting way to try to get at some answers, at least.
So to that question, is it changing the discussion? Is there a discussion at all? There was, does anyone notice it right? It would be handy to know whether you're eliciting any kind of reaction at all. So it seems like that could be an interesting vein to go after, to try to understand if you can go after the large sums of money that are being spent on 16 year olds.
Erinn Taylor: [00:13:07] No, that's very interesting. So Published a blog post or they're in the year talking about this very thing. So we did some research around the MTV awards, five comm, and we noticed an insight in there that we would wasn't weren't expecting it's that. Toyota was coming up as a advertiser for this.
And people were talking about it. They really liked the placement or the cars or the ads that were happening within the show. So that's an insight that we didn't expect to find. And then we were able to highlight to the organization so they can take it back to their advertisers and say, look.
People are talking about it. We should keep doing this. We would love to have you as a continued advertiser for our programming. And then we're seeing that a lot with also this, the general agencies, right? So that's one of the reasons they use our campaign feature to track, ad testing.
They'll do again, they'll do things like track to see how their recipe. Did are they have their video around recipes or who are around making cakes and that cause on YouTube, like who's listening to it, who's watching to it. And also is it comparable and benchmarking it versus our competitors doing the same thing.
And now how do I do what they did if they're out performing mine or if I'm okay, we're performing, let's do that again. So it gives you a lot of opportunity to create one benchmarks, to understand how people are reacting and then just be able to then replicate, right? Because without any sort of KPI without any sort of.
Understanding of what the numbers mean. You can't actually go out and replicate and produce the same thing or better. Again,
Peter Mahoney: [00:14:31] that makes a lot of sense. And one thing that's come up a few times. I'd love your opinion on what do you think of NPS?
Erinn Taylor: [00:14:39] That is a loaded question.
Peter Mahoney: [00:14:41] I, on purpose, I tried to load it.
Erinn Taylor: [00:14:44] So NPS I think I mentioned earlier, I think NPS, isn't going to die. It's been asked for many years and we have the same conversation every single year in market research. Are we going to stop using NPS, right? But yet it keeps coming back. Sometimes people load it into their bonuses.
Sometimes they load it into their organization, overall KPI, et cetera. It is a scale, it is a way of measuring something that will they come up with something similar or better maybe. But again, every year we had the same conversation about NPS and is it going to die? It's not gonna die in time soon.
But what I do think Brock is wrong with it is the follow up question that I mentioned earlier is the fact that after you find out. That pre people are willing to share this, recommend this, whatever, ask them why. And just don't prescribe to them. Why you're asking, just ask them this unaided awareness question.
How, why do you feel this way? And then just let them emote and then take that information. And what Canvs can do with it is help you understand what they're actually saying in a very quick, concise way. Again, we have tracking studies that are NPS studies. So one of our bigger use cases they track, like I mentioned, gaming company earlier.
They track it every single week, how people are responding to the game, there's an NPS type question in there, and then they're looking at it to see how do we make it better? How do we get more people? And specifically, not only just for getting more people, but it's also around the ad dollars that go into the gaming.
Cause it's like a mobile based game, which sometimes those are free. So how are you getting the ad dollars from that? How do we improve it to make sure that people are not being coming disengaged with that product and then moving on to something else?
Peter Mahoney: [00:16:18] Yeah, it's interesting. And I find that. So many people and probably because as you mentioned a minute ago, people are building NPS into their incentive systems and have, but it's gaining even more popularity.
And as a result, you just see it everywhere and you tire of it sometimes because every single company keeps asking me my scale of one to 10, how likely am I going to recommend the product? And you know exactly what it is. And and the frustrating thing is many people will ask. The follow-on question, but you just have this deep seated belief that no one's ever going to see that or do anything with the data.
And it feels like an incredible waste to have that feedback. So you feel like, eh, I don't want to provide the feedback now because no, one's going to look at it. I've literally never been followed up with, by responding to someone's MPS study. Even though sometimes I had some very specific concerns.
So it seems like it's a. Less than perfectly effective solution for people.
Erinn Taylor: [00:17:21] That's a hundred percent correct. And that's the inertia that I was mentioned earlier. That's where they asked the question sometimes. And when they do ask the question, what do you do with it? And either it's too expensive or it takes too much time to analyze or, I don't know what to do when I look at the data, I don't even know what I'm supposed to do with it.
What's bubbling to the surface. So that's where we're trying to change. And we have been successfully able to change the mind of the companies that they can now look at that data and make decisions around it. And that's also why we use emotion as a better nuance than a value score, because.
Again, just knowing that something's positive and negative. I can't do anything with that. It's not programmatically capable of advancing in any detail. If I know they're angry, right? And people are moaning, they're angry. And I am able to identify that I can put them in their VOC loop where immediately someone can contact the person and prevent them from causing more damage, solve their problem.
I could see how these people are angry. Don't worry about the people that are super happy right now. We'll get to them in a couple of weeks because they're not going to do any damage. They're going to tell people they like our product. So that's where we're finding are getting our cadence and getting up, running around people, understanding what's in that, that open-ended content after the NPS.
And that's really where we started into research. Again, we are meeting their entertainment and they're just looking at, the syndicated products on TV. And next thing you know, our customers are saying, you know what? We have a whole bunch of other data. I bet you. If I threw this data into the system, I would be able to understand it.
Cause I can't do anything with it right now. We're just wasting it and we need to, we need to have a better NPS. We need to have a better, whatever our KPIs are, low here's one. Maybe you can do it. And we find out that they don't like this. They don't like that. They love this was per our, our dollars in the right spot.
Peter Mahoney: [00:18:59] That's great. I really appreciate you sharing your insights here. I think we're coming to the end of our scheduled time. Before we ask our last question, I just wanted to make sure that you help us understand and our listeners, how can they learn more about Canvs and some more of your insights?
Erinn Taylor: [00:19:17] Absolutely. So Canvs.ai. So C a N V S. Dot AI, no AI in there. You can also find us on LinkedIn. You can find us on Twitter and you can find us on Facebook. Great.
Peter Mahoney: [00:19:28] And by the way, I love the start. I have to take a moment to talk about the name because we had a great discussion about this too. And 'cause it's very clever.
So canvas, as in canvassing your audience and doing a study and the missing a at the end refers to the fact that you can understand text with misspellings and texts in the real world that you see in, or how did I do,
Erinn Taylor: [00:19:51] was that accurate? That's perfect. That's exactly right. So you're able to, yeah, we're able to find misspellings.
We're able to find synonyms things that are like we're clustering. So there's a lot of things that are happening. We're not a word cloud. We are actually topic analysis that we're looking at how people are saying things and grouping appropriately. And that's absolutely right.
Peter Mahoney: [00:20:09] That's great. Kelsey, I think you have one more question,
Kelsey Krapf: [00:20:12] given all this information and all this insight with, emotional and semantics, how, what advice would you give to marketers or, CMOs or those aspiring to be one.
Erinn Taylor: [00:20:23] In terms of just straight advice to someone taking over or working directly with markings, whatnot as the first thought to stare down the balls, meaning get closer to your sales, get closer to your product so that you're aligned on your messaging. Have a plan. And then make sure that plan is accountable.
So you have some sort of KPI that you can track against that to make sure that your plan is working. In terms of utilizing AI, utilizing sentiment analysis, obviously I would recommend Canvs to understand better what's happening related to the texts, but more and more than that is getting closer to the consumer, understanding what the customer's doing insane and what they ultimately wants you to do because our job is.
As marketers is to reach the people that want to use our products and making sure that we have in our product portfolio or services, things that they actually want.
Peter Mahoney: [00:21:12] That's great advice. Really appreciate that, Aaron. So I think with that, we're going to wrap up right. Kelsey.
Kelsey Krapf: [00:21:18] Yes. Thank you so much for your time today, Aaron, what a fascinating conversation, excited to, relearn this again, listen to it too again, but make sure to follow the next GMO and Plannuh on Twitter and LinkedIn.
And if you have any ideas for topics or guests, you can visit our website or email them at the next CMO app. Plannuh dot com. Stay tuned for the next community launching soon and have a great day, everyone.