E033: Sentiment Analysis

with James and Jane. now on to this episode.

Hi this James and Jane and here we are with another episode of a world of work podcast. We've got another summer special for you. We are part of our little collection of episodes here looking at engagement and experience and all things to do with that, and how you understand the employee perspective from the world of work. And today we've got a conversation, which we've had with Andy Roberts from weekly 10,

yeah. So just like all of these summer specials, they are little mini interviews that give you the opportunity to get behind the curtain on some of the people working and studying employee engagement. Andy Roberts, for me, was really interesting conversation, partially because I really love how he's using a data tool to empower managers to have better conversations with their employees. I think that's a it's the first time I've heard someone talk about using data in that way, and using it as a way to reassure senior managers and senior leadership that it's working and what's happening, but therefore giving them the ability to delegate it down to management. And I love, I absolutely love that. Yeah, there's

some really good stuff in this conversation coming up. I mean, we touch a little bit on some of the technical sides of things. So I certainly left with a better understanding of what sentiment analysis itself actually is, a little bit of an understanding of how the process works. Obviously, you know, total base level understanding. But I learned a little bit of that, as well as some reflections on how you can use for findings to help organizations. Yeah, and I now know what machine learning is, so that's a win.

In all seriousness, I think Andy explains really, really and simply how machine learning might be being is being used, certainly in his product, and might be in the industry and also because he comes from a business background, and because his background is not people, I think he comes trying to solve a problem that he is a manager had, and I always find that good purpose behind it, yeah, sort of motivation. So I really enjoyed

this conversation, as you'll probably guess, judging by the number of questions I asked, but I think it's probably just best to hand over to Andy in the conversation before that. Why don't we talk about getting in touch? Good, catch. So

if you are listening to this episode and you want to get in touch as always, www, dot the world, the WoW podcast, org, and we are always on Twitter at the WoW podcast. So whether you are lying on your beach, on your sun lounger, or you are still at work this summer, or on skis down south, get in touch. Let us know what you think of the episode. And for now, we'll leave you with Andy. All right, everyone. So

here we are in the core part of our conversation. Today, we are focusing on our conversation with Andy Roberts, and we're going to be speaking a little bit about things to do with sentiment analysis and a little bit of an exploration of exploring data within the world of people, and a little bit of people analytics as well. But before we jump into that, why don't we hand over to Andy, and Andy, could you introduce yourself to the audience and say a little bit about your background? Bit about your

background? Hi, yeah, thanks, James, yeah. I'm Andy Roberts, so I'm CEO and founder of weekly 10. And weekly 10 is a employee engagement platform which is focused on a very short term weekly check in between employees and their managers. It's designed to essentially replace traditional performance management platforms and engagement surveys with something which is more cultural and focused on continuous continuous communication. And as part of that, we have a machine learning platform which takes that data and and looks at things such as sentiment analysis and provides that data to to management to make, to make key decisions. My background itself is is actually one within the software and leadership space and and that's across finance e commerce and telecommunications industries. So I've worked in both startups and SMEs, as well as some large multinational corporations, so I've had good visibility on how engagement works in those organizations and how companies have and organizations have done that in the past, and in some cases, are still doing, still doing. So today,

that's a really good. An introduction, and you've got a quite a varied background, which is really helpful for this. So a combination of the sort of technical side and some of the IT side with some of the engagement side is kind of interesting for us. Um, just to start things off, when you speak about employee engagement, what exactly does that mean to you? What's your view of employee engagement? So

again, I guess the thing with these terms is there's no well defined definition, I guess, in the dictionary, but engagement really, really to myself and to us, is about the relationship between an organization and its employees to that two way relationship and how an employee feels towards that organization more widely. And obviously that organization is both the entity and a collection of people in itself, but it's very much about the way in which people feel about the organization they work in. And so

why is that important from an organizational perspective? I guess you kind of jumped out of some former parts of your career to really focus on it. What is it about engagement that you find so important organization? Well,

I think there's a lot of evidence out there that suggests that highly engaged employees have, you know, certainly high levels of productivity and lower levels of absenteeism. In fact, I think some of the latest, some latest statistics are other employees are up to 21% more productive when they're when they're highly engaged with an organization, and around 30 to 40% less likely to call in sick or leave the organization. So when we're actually looking at the the market out there these days, you could, you could often say that it's a candidate market in the fact that it's all about getting the right talent in the right roles, and it's expensive to hire and retain and make sure that people are in, you know, doing those roles that are, that are out there, and, you know, getting the right people in and retaining them is is critical, and you can traditionally do that by giving them pay rises and doing various different things, but essentially giving them a feeling within an organization that they contribute and that they, You know, have a desire to move this organization forward, is in many cases, much stronger way to to motivate and retain that talent. You

said a little bit at the beginning there about how highly engaged people are more productive. And so there's obviously some benefits around that. When when you look at people and when you think about it, how many or what sort of proportion of the population do you think are engaged in their roles at the minute, have you got any sense? Have you got any sense

of that? Yeah, well, it typically will vary per sector, okay, but I think if you're looking at some of the customers we've worked with, you may be looking at less than 15% in terms of highly engaged employees. I mean, typically you'll see that the employees would from a traditional performance pulse survey. You will see this, but also kind of in different ways that you look at people analytics is exactly the same. You'll typically split your employee population between engaged employees, so people who have a are really positive in their outlook towards an organization, have a sense of belonging and the fact that they want that organization to grow and do better. And then you'll have the kind of middle of the road, not engaged, but not necessarily highly disengaged employees, which would be your typical kind of coasting people. So the people who are you know that they're not necessarily damaging the organization, and they're not, you know, affecting others however, they're probably doing the minimum required in terms of what they're doing. And then finally, you have the disengaged side of things, which is where people can actually be actively negative to organization and detrimental in that they may affect the attitudes and opinions of other people in the organization and possibly even outside of the organization. Typically, we'd see that split across, I mean, typically, especially in the UK, where productivity is one the lowest in the, you know, in the developed world, you will see levels of engagement, certainly engaged employees, you know, less than 15% and then you'll find a large, the most proportion in, obviously, in that kind of coasting, that coasting

zone, sure. I mean, I've heard, you know, similar types of metrics before, but it's always amazing to me, the level of disengagement, or the lack of positive engagement that's out there. I mean, it's so fascinating and and it speaks to so much kind of unrealized potential, to some extent, unrealized opportunity, to both the. Productive, but also to have a higher quality of personal working experience, definitely,

definitely, I think, and it's, and it's a real concern that the majority of organizations haven't got to handle on on this problem, because it is clearly a competitive edge. If you can increase even by a small fraction, you're highly engaged. You're highly engaged staff, and more often than not, it's left down to individuals to engage with an organization. And you know, newsletters or, you know, the odd engagement survey from senior management, etc, is seen as a way of engaging employees. If you actually speak to employees, they, you know, they have have a different view. So, you know, I'd see it as being a way of empowering managers and empowering others to, you know, generate this network effect within an organization to to increase engagement, rather than just relying on very, very high level strategic initiatives. When

I think, I think what's really interesting about what you're saying is it brings together. So we talked in a previous episode about motivation theories, and we talked a little bit about self determination theory as one of one of the theories that was relevant. And I think what's really interesting about what you're saying is, is that it it relates really well to two of the three key elements of that theory, which is, you know, people want autonomy. They want the ability to engage fully and actually put back and have a two way conversation. And also, and I think this is the bit that I think organizations really struggle with, is this concept of relatedness, that employees want a relationship with their organization. And it is not a newsletter. I was, I was just reflecting on what you were talking about a little bit, and I know you're going to come on to some of the tools that that you guys use, but I remember I worked for an organization, and the CEO used to send out weekly newsletters about what he'd been up to, and he saw that as really powerful. I was like, Yeah, but do you know what the rest of everyone else is up to? Yeah. And it was a one way conversation, right? Yeah, and, and, don't get me wrong, it was a step further than the previous CEO, but it's still a long way from having a two way conversation, definitely,

definitely, and that that is it. It's do, do employees feel as though they have a role in the determination of where that organization goes and you know whether it is an you know whether it's a non profit, and there's different targets to a profit making company, but it's, do you have an influence, and what is your influence on the strategy of that organization and around that what does that mean in terms of whether it's recognition, in terms of whether that's, you know, a sense of achievement, etc. There are a number of different components in there which, you know, cannot be, as you say, cannot be fulfilled by, by a one way conversation.

We used to in one of my old organizations, we used to talk a little bit about a phrase that I don't really like, but the underlines, you know, you I kind of like, which is colleague, share of voice, and that's, you know, as much as I said, I don't like the phrase, I kind of like what it's getting. Um, yeah, definitely. So, so that's a little bit around the topic of engagement, and some extent why it's important, and a little bit of a view of the current state. You know, if you're going to improve something, one of the first things that you need to do is to really measure it. So how do you go about, or how do you see organizations go about analyzing and understanding the current state in relation to engagement

Well, again, I guess the traditional way the is the engagement survey. And there are plenty of organizations out there still providing engagement surveys. I did, statistic very recently that organizations, I think less than I think, around about 60% of organizations are now still using engagement surveys, but the remainder of actually move away more towards people analytics, etc, which I think, in itself, is dangerous, but we might be able to talk about that when we when we talk about sentiment analysis in a bit more detail, but at the moment, it is about the engagement survey. And there are large organization, large organizations out there, which will have, you know, 12 questions. You'll answer the questions, and you'll get score back, and that will be run every six or 12 months, and it's, you know, it's an interesting concept, because it's a moment in time which, which, again, is which, I think is a challenge around those traditional means, because if you're asking somebody that question once a year or every six months, you don't know what happened the day before The day that you asked the question? You know, some questions dreadfully when I've been really annoyed the day after. I

know someone who used to send it out the day off the pay rise. Yeah, we always used to give

we basically, when I was involved in some of this stuff, what we would do is we do a series of problems, like at the start of the year you said you wanted more trade. In development. Look, here's all the great stuff we've done for you. You wanted this other stuff. We've done all this great stuff for you. Isn't everything nice now we do a survey exactly, and

that you know that really, and that's kind of where my career change came from, and why I started the company that I did was around the fact that these are seen as tick box exercises for one person, whether it's in management or HR, but it's essentially ticking their box for the year and seeing, seeing that progress and not actually using the measurement as something which is going to have an impact on the company's bottom line or the organization's outcomes. So they have to be using the right way. And the challenge with them as well is the fact that you often have the data, but then there's no empowerment for managers to actually use that data and take action and improve things. So you have to be careful that you're just not measuring, you know you're measuring something and praying that it goes up in many respects, but, and I guess the other measure often has been the performance reviews and looking at, you know, holes people have set, etc, but, you know, I think, as we've all come to see, and there's plenty of evidence out there that say that both employees and managers do not engage with with that process, But actually, looking at companies, often look at the number of goals set, for example, by by employees for the year. But again, they'll often be targets, and managers are kind of beating with a stick to kind of get their employees to set goals. So yeah, that's how, that's traditionally how we, you know, we see, when we go into organizations, we see how, how that's being done. But on the flip side, we are now starting to see, as I mentioned, with the with the drop in the number of engagement surveys being run, is use of people analytics. But that in itself, carries some danger, because often that means that the data is actually coming passively from employees. So whether it's looking at especially for large organizations, whether it's looking at attrition, changes in productivity, you know, number of meetings managers have, etc, this data doesn't directly come from employees. So in itself, you can actually be making, making the engagement side of things worse by by using, by using different techniques as

well. And by, by the way you describe that, you mean it's just kind of like observational, incidental data that people are taking to use for their analytics. Is that what you're getting at? Yeah, so, rather

than asking employees a question, actually just looking at the raw data and whether it's big data analysis or whether it's using things like machine learning to extrapolate and interpret that data and then make predictions off the back of that, that can be done just using employee data.

You know, there are even, there are even stories out there at the moment about none of our customers, thankfully, of using this at the moment around actually recording all the the typing that you do on your on your laptop, for example, so actually capturing everything you've typed in. So looking at productivity from that perspective, looking at, you know, all kinds of stuff, whether it's keyword analysis in that, whether it's how active you are within, you know, within within that, within that period. So that in itself, although you might be getting some useful data, it is the wrong message to it to employees from an engagement perspective as well. So you might be able to measure things, but you're not, you know, at the same time, you may well be crossing the barrier of trust with with your employees. Well,

that, I mean, it's just, it's an extraordinary way to think about it, isn't it? Because effectively, what you're saying is we don't believe you're able to articulate what the problems are, so we're going to try and identify them directly from what your behavior or what you're what you're actually doing, but in doing so, we may damage the relationship so much you don't want to be great, yeah, experiment or influence of the experiment, that kind of thing, right? Yeah, yeah, yeah. So, so

these are, these are two of the ways that companies are currently trying to address these issues, yeah. And what are the alternatives?

So, you know, the alternative, certainly, from my perspective, is about changing the culture of the way in which feedback is provided. So typically, as we've just mentioned, it's very much a there's a combination of ways that feedbacks elicited from what we what we've just mentioned. One is from a survey, which is done regularly, but maybe every six months, or what? How? View, and some organizations have now started doing those monthly or even weekly. But the problem with those reviews themselves and those questions is they're often leading questions. So how are you feeling between one and 10? How do you feel about the organization between one and 10, etc? What you're doing is you're you're setting the context and then asking somebody a relatively closed question behind that, I believe that it's really about changing the cultural organization from from the ground up, and encouraging the conversations and empowering managers to do so. Often, middle management is seen as a bit of a dirty word, and there's a big focus on pure leadership, but the skill of management is is as important as ever, and managers need to be empowered to get feedback from their employees and actually act on that feedback throughout the organization, rather than it going purely to senior management and bypassing managers because many issues and many you know, much of what you're trying to achieve through these strategic initiatives can be achieved by managers if you give them the right tools. So you know the approach of having something which is whether it's weekly, bi weekly or something, but something which is there, whether it's in the calendar or whether it's using a software platform to provide continual feedback and have that conversation makes managers job much easier and actually more effective, and also opens up the organization to employees, and there are many other different ways of doing that. So for example, you can have peer recognition built into those, those type of things as well, where you can encourage people to think about the job that other people have been doing, and you know, good work that's being done, and share that work. And then also through employee exposure, which is often which is often under underrated. So actually exposing good work and allowing managers to do that in a way which is consistent across the organization. So I believe it's very much about having that conversation and having consistent conversations across the organization, not just relying on, you know, the 10 or 15% of really proactive managers, but actually have sort of a framework there which managers can build upon and provide that support. Yeah, there's

some really interesting stuff in there. I think, in terms of where I'd like to go next to this, I'd like to jump onto actual sentiment analysis itself. So, I mean, we talked about collecting different types of data. You get metrics in from your engagement surveys. You get so you can get, like, quantities of data in from those looking at, I guess, response rates and scoring to different questions. You can get, you know, text based responses to feedback, where you're eliciting opinions and all that kind of stuff. If you've got a mixed pool of data like that, where does sentiment analysis come in? I guess, what is sentiment analysis and how do you use the data to to analyze sentiment. So

sentiment analysis essentially the understanding of the emotion and potentially subjectivity within natural language. So it's you know, if you're if we're having this conversation now, and you say something, then I can pretty much immediately based on the language that you've used, but without seeing your face or your reaction in your face, I can pretty much tell whether you're being positive, whether you're being negative, whether you're being constructive, etc. And you know, I can decipher that because of my experience of those words and how they've been used in the past. And that's purely what sentiment analysis is, it's about taking natural language and working out what it means from, from a, you know, a level of subjectivity, I guess, from, from a level of emotion. And that's that sentiment analysis. Now, where it comes into its own when we're talking about people analytics is the fact that, as I mentioned, the importance of having these conversations and having these open conversations rather than questions, is that, because we have this text data, we have we've collected data of these conversations, We can actually then tell how the organization, from a sentiment perspective, is so, whether it's a department level, organization level, you know, what is the prevailing trend of sentiment across across An organization? Because you have this text. Gotcha. Sorry, but at the same time of which I've got many

so and when you're thinking about the use of sentiment analysis, is that predominantly in a conversation, a conversation inverted commerce, because I realized that there's data and textable, but a conversation between the individual and the organization. Between the individual and manager, or between the individual and other people in terms of, who does the person think they're giving feedback to?

It's, it's, it's potentially any, any of any of the above that you've just kind of covered, now that the area we focus in on is the open questions. So typically, you know the standing questions between, say, a manager and employee, because those are the conversations which really judge what the prevailing sentiment is. Now the input that sentiment analysis is best used as a trend and at a higher level than, say, just an individual level, but it gives you, it kind of gives you a view of the kind of water cooler talk that you don't get from an engagement survey. So you're actually getting underneath the skin of the organization and understanding, you know, what, what does it, you know, mean? What these what do these conversations mean? But there are, there are other analysis, rather than just the, you know, the pure ago view of whether an organization is positive or negative, because you can also group together so you can, you can work out whether the organization is positive or negative about a certain client or customer or about a certain project, for example. So there are also different techniques you can use, using similar, you know, kind of AI and machine learning technology that that will give you, give you more detail.

Yeah. And then just one another question for me, how do you cope with or Does, does sentiment analysis cope with people where maybe the language that you're analyzing in isn't their first language? Or do you find that there are differences when people have, I don't know. For example, grown up in different English speaking countries?

Yes. So there are different impacts to what we would call baseline sentiment between both whether your first language is English, for example, or whether you're in a certain role type. So we actually, the system that we develop actually looks at what we call role archetype, which is the fact that maybe a software engineer has a completely different baseline sentiment to a sales director, for example, and also different we

both look at each other because you're just, I know that James is thinking without even looking how cool that is. Yeah. Mm, how cool that is, and also how much I desperately want to see what those baselines are, yeah, and that. And

so it does vary, and also the first language, or whether your second language, English has an impact. But the point of sentiment, and the way in which we use it is that it's not so much about the sentiment ratio that we would generate, it's more about the change in sentiment. So you have a baseline sentiment, a prevailing sentiment, and it's when that changes, it's when there's potentially concerns, or when you know something's potentially happening. So it's more about the change in sentiment rather than its baseline.

And so do some organizations use this for major change programs and things like that? So for example, if they were going through a restructure, would they use it to see what detrimental effects it's having on the workforce, for example, exactly.

So some of our customers will use this both looking backwards, so from a retrospective perspective, and also looking forward. So for example, you can bring in data about maybe it was financial results, and this also gives you an idea of how engage your staff are about financial results. So if you release financial results, you know, did that impact the sentiment? And you can also look at with, you know, a slightly lower level where there's a change in sentiment when a key leader within the organization left. So you can look both backwards and then also look and use that data to do some level of predictive analysis based on kind of what's happened before, and the way in which the model, you know, expects things to change based on based on that sentiment. Yeah,

I've seen, um, I went to a Behavioral Finance Conference a little while ago, and this is a bit of, a bit of an aside, I guess. But when I was there, there were some people talking about using natural language processing and analyzing sort of Twitter data to look at automated investment decisions is that loosely, the same type of approach that you'd use or similar?

So, yeah, the use case you mentioned is a key use case that probably the biggest use case outside of what we're talking about in terms of people analytics would be around marketing analysis. So, and that's very much similar to what you mentioned. So monitoring social media, bringing that data in and looking at how people are talking about products, about companies more widely, and indeed, some of that data is being used for employee sentiment analysis. However, within within, within our approach, we don't use that data. We typically don't think it's that accurate for employee data. Rather, we use the data which is internal within an organization. But, yeah, it's exactly the same principle. It's about understanding what that text means from a within the context that it's being used, what it means from a, you know, a sentiment and emotion perspective

and just on that. So the actual sentiment analysis, if I understand it, you know at all, is effectively, you get text. You get lots of texts. You analyze that text, you look at keywords, and then you look at, presumably, relationships between the words that are in there, and some sort of sense of positivity for each word. How does it work? And

yes and no. So the way in which the way in which our platform works is, it is a machine learning based platform. So in the way machine learning works is, essentially you have an artificial intelligence algorithm. So you have algorithm which is essentially a load of code, right? So it's a list of instructions, and what the algorithm does is it learns what positive, what negative, and what neutral looks like. So it's not so much about looking for a positive word. It's the fact that the algorithm has learned what positive looks like and what negative looks like. So it actually takes away the subjectivity of us deciding what a positive and negative word is. And actually it's similar to as I was saying before, in that if you say something to me, I naturally know, because of my neural network in my brain very quickly, whether you're being positive or negative, and that's exactly what our platform does in terms of you give it some text. It knows from past experience what positive and negative looks like, and it provides a response with essentially the ratio of how positive or negative that that text is so. So that's the way it works, rather than looking for specific keywords. Positive

is, you give it a lot of data and a lot of instances to work through, and then it defines it,

yeah. So we actually get a lot of our data from different sources, outside of, outside of the people space, really. So we will pull in lots of different data, which typically we will use something called supervised Well, we use a combination of what we call supervised and unsupervised machine learning. So supervised machine learning is where you provide a piece of text and you also provide it with a score. So you say, this is text, this is positive, or this is text and this is negative. Now one of the data sources for that, a great data source is review data, because when you write a review, you say something which is positive or negative, and then you give it a star rating, great. And that, if you give that to an AI, it can train up very quickly, and it can tell whether the next response is, you know, the next input is possible negative. But we can also use unsupervised learning, which uses a clustering technique, so essentially takes lots of data in and works out relationships between different groups of text, and then from that, it can decipher actually it falls into this category, and then it gives you a bit more subtlety in terms of how that works. So we actually use a combination of that supervised and unsupervised learning for our algorithms. Sorry, for example,

slightly silent. It's my mind slightly blown, just because I'm trying to think through all of the implications of what you're talking about. And I think particularly, I just want to rewind a little bit, which I know is not directly related, but I'm really interested you talk about this conversation between potentially, managers and employees, given that that's probably the strongest relationship, and you talk about it in two ways. One is you talk about it as a learning opportunity for organizations to track trends, and the other, you talk about it as facilitating a management tool to help people. Is that, is that right? And if so, from flashing back to your old life, what matters more? What's the bigger value?

The biggest value, fundamentally, is in the person to person conversation between a manager and employee, because that is the culture that you build in an organization. What the Analytics does is it gives you a bow whether of your organization as to whether you're making progress. Or you're not making progress in an accurate way. So and it's not about and it's about being very transparent with employees as well. They should know that anonymously, they are tracking sentiment with an organization. But certainly the power that the most powerful element of those two has to be the conversations that happen in organization, because that that's what, that's what an organization runs on. It doesn't run on predictive analytics.

Yeah, that makes perfect sense. When you're getting data. Do you get it purely from your sort of structured and open feedback tools and conversations? Or do you, or others do things like troll messaging, troll email traffic, troll phone conversations, where does, where does the line end on that kind of stuff. So we take a very,

very strict line that we just use the data that comes in via the via the updates, so via the regular conversations that come in through through our platform. We don't look at email, we don't look at instant messaging conversations. It's very much from those conversations. Now, the other reason for that, as well as that being something which means that employees know what's going on, is that we also know the context behind that. So we know, for example, that a question has been asked around, what have your successes this week? So we expect that response to be more positive than the average, for example. So we can then actually build a more accurate picture of that, of that sentiment when it comes to the management view. What this is really about is reuse, is the fact that as almost as a side effect of having lots of data in this space, we were able to provide very accurate analytics. From a sentiment

perspective, you're also on a very practical level. You know, one of the things consistently the and you mentioned it at the opening, I think around trusting middle management and junior managers, yeah. And I think what, by having this byproduct of data, you give senior managers and leaders a way of saying, Look, you can still monitor what's going on. You can still get information. You don't have to be involved in every conversation. And actually, if you let your managers get on and manage, and give them the tools to do so you will get more and better and richer data that will give you a better understanding of how to improve the the wider organization anyway,

definitely. So you're building that culture by having those conversations, and then, yeah, exactly, you're then also getting that, you know, the data at the top level. And you know, because that means that managers are getting value out of that whole process. So are employees. And it's not something you're forcing people to go down, down the route of filling out a survey, for example, and this is much more accurate data than a survey, because what you're doing is you're actually getting into those conversations rather than asking the questions that you specifically want answers to. So So, yeah, it is about building that building that empowerment. So,

again, I guess that's the really positive side of it. James has just asked the previous question James to ask, I think we've, we're pretty much asking to everyone who's involved in this kind of space, because we understand the challenges you've alluded to. It, ethics, around machine learning, artificial intelligence, data, particularly data, is just huge at the moment in terms of concerns, I guess my question is, you sound like you've got very clear boundaries. We know there are other organizations who maybe don't see it that way. What do you think the risks are for employees for organizations, as this kind of understanding of how we can use data to change employee relationships goes forward and grows.

So I think if you looked at what we, you know, we talked earlier about there is danger that employees are ignored because they, you know, data is just taken passively rather than actually actively, from from from people, from the conversations, but they clearly are issues in terms of privacy. What I would say is that GDPR actually does a pretty good job of protecting employees from certain aspects. So for example, if we look at, if we look at profiling, GDPR specifically calls out profiling as being something you require consent for. So what we don't do is, we don't provide sentiment analysis for an individual. So we couldn't say, James, this is your sentiment for the year. What we do is at an aggregate level, a department or an organization level, or even role type level, we're able to pull out that data so just

just so clear. That is primarily because of GDPR legislation that kind of handcuffs organizations them to where they can get. It.

And just to be clear, GDPR is a European General Data Protection Regulation piece. So, yeah, it's a European focus. But that is because of that.

Yeah, so we, we made the decision that we don't believe providing sentiment per user, yeah, something which is, which is ethically correct. Now the reason being, although we want there to be an open, transparent conversation between employees, managers, senior management, more widely in an organization, sentiment analysis is profiling. So what you're doing is you're pulling data out and making an estimate based on what somebody said. It's not something they've specifically said, if you know what I mean. So there is, there is a danger. There is a danger there ethically, that you're, you know, making decisions based on what somebody said when maybe they didn't mean that, for example. So

it's someone else's interpretation or translation, albeit a very educated guesswork, educated, informed one, it's still a translation exactly you shouldn't be able to make,

you know, a decision to fire somebody, for example, based on that data, because it's profiling data. So we see sentiment analysis as being very much a bellwether, really, of an organization, and when you see something changing. So if you're comparing sentiment across different office locations or different role types, you see a change in that. Then it's about looking at what's actually happening underneath, and trying to investigate that, and talking to people about that. So what GDPR is very good at protecting individuals, even in the workplace, which is sometimes not, you know, not seen as personal data, but, but a lot of it is at doing that. Now, obviously, that doesn't apply to the US, and it doesn't apply to many other parts of the world. However, EU, EU companies, must comply to that regardless of where they provide that service. So it does provide a good level of protection in that respect. But yeah, it is about working also with companies who are ethical and transparent about the way in which this data is used.

Okay, that's really, really great. Thank you. I think that's all my questions. I think it'd be really useful if maybe you could share with listeners how they might learn a little bit more about what you do and about a little bit about your organization, because I feel like we're literally scratching the skirt. Yeah,

clearly, we're at the start of it, so it'll be great to help them find out more. Yeah, sure. So

so we actually, as part of as part of our platform, we also have a research part of the organization work very closely with a number of UK based universities on on both the engagement side, so both the kind of low level conversations and and you know when it's best to have those what the best questions to ask are, for example. And then also on the sentiment analysis side, and we build psychological models on top of sentiment analysis to actually make more sense of just more sentiment. So we actually publish quite a lot of our information on our website as well, which is weekly ten.com and there's also details in there. We also look at certain, you know, blog articles we have one on kind of four day week, and how that could impact productivity in the UK, if it was, if it was potentially adopted, and various different, different things there. So you can find more information about the product and also, also some of our indie work on our website. We're also obviously on social media these days, so you'll see us on Twitter, Facebook and LinkedIn,

yeah, and we will, as we always do when we put out this episode, we will point people in the right direction on our social media and website as well. Because I know that if you're interested in this kind of area, and I know a lot of the people, particularly some of my fellow masters students and some of the HR people that I work with, really interested in this stuff, and they want to get to a place where a place where the data collection process is has more elements to it than a simple collection process, right? Which is about so I think definitely, okay,

really annoyingly, we've run out of time. Yeah, I've got a list of questions in front of me, but I will go away and think about them, and then maybe we'll come back to you. But for now, it's probably just enough for us to say, thank you very much for your time. It's

been really, really interesting, learning about what you do and your background and also understanding some more about sentiments, analysis and how it works. Anything from you. James, no, just reiterate

the facts. It was really interesting, great to learn a little bit more fantastic. Little bit more fantastic topic. Really, really useful. So

thank you. So thanks a lot, Andy, and we look forward to speaking to you again. Great. Thanks a lot. See you soon, right? So,

welcome back. That was our conversation with Andy from weekly 10. I thought it was a really great conversation, and you did ask a lot of questions. Yeah,

and I do feel like I tended to a fan girl, slightly awkward, sorry, Andy, but I just think, certainly from my point of view, I found it really refreshing how he talked about ethics, but also he talked about collecting of data only for the purposes of this and only looking at data that is specifically for analysis. And to me what he was saying, not only the ethical argument, but the practical one, which is by understanding the questions that the data is given in relation to. So for example, I think the example he gives is where they've asked about successes for the week. Yeah, they know it's likely to be more positive. They can get much better understanding of what that trend might be like, yeah? So for me, that was, that was really helpful. Takeaway, yeah, I

thought it was good. I mean, I thought the whole conversation was good. I think the starting of a conversation with some framing around engagement and how it works and what it means for organizations is always helpful. Like Andy said, Everybody's got a different view of what engagement

really means. My letter refers to it as a slippery fish. Yeah, OK, slippery fish for definition, hard to grab onto. I

totally get that. So I thought that was useful. And I thought some of a conversation about the different types of data that were out there were good as well. And like you, I really liked that ethics point. I think, I think it's absolutely right that people are aware that data can be used for all kinds of purposes, and that sentiment analysis works across all kinds of different types of data sources. But I think there is a bit of an ethical obligation to approach things like sentiment analysis in the right there's

also a hidden, a hidden nugget of Steve Jobs esque culture, which is, hire good people and then get out the way, right? So there is an implication of, if you give managers the tools to better have open dialog and then share that information, you can much more effectively shape an organization, but if you try and control what you mean? Interesting?

Yeah, and we touched on that. I mean, we didn't really drill into it as much as I thought we might about the, you know, the place for empowerment and autonomy and all that kind of stuff. But that's very much implicit in what's, I think, throughout, and trusting your managers to be managers, and investing in them and helping them. And, yeah, I think that's really powerful, alright, so for me, that was a great conversation as ever. You can get in touch with us at the usual places. You can get us on Facebook, LinkedIn, other places, Instagram, Twitter, at the WoW podcast, that's probably our most, most frequented of a social media site. So you can

and if you're interested in Andy and his work and his organization the week attend, then check out our Twitter feed when we launch this episode, and you'll be able to find the links there. Yeah, there'll