Cole: Hi. Welcome to the People Analytics World Fireside Chats. I'm Cole Knapper from Agnostic. I'm here with Vincent Greco from Shopify. Vincent, it's good to see you.

Vincent: Yeah, good to see you too, Cole. Really happy to be here. Awesome.

Cole: Well, I wanted to start out just by getting to know you a little bit. And as a part of my research for this conversation, I saw you actually have a background in political science, and I don't think that's necessarily the typical I don't think there is a typical background for People Analytics, but I don't think that's necessarily the typical background. Anything that political science you feel brings something special or unique to our field.

Vincent: Yeah, no, that's a great question. And you're right. I think everyone sort of approaches People Analytics from, I think, a different sort of walk of life. Yeah, I would say from my perspective, I think where political science is helpful is like one, it's like super interdisciplinary. So it draws from economics, sociology, you know, stats, all, you know, a number of different disciplines. So I think you sort of have to borrow from a bunch of different areas, which I think is usually very helpful, particularly in like a PR context where you are sort of looking at what other folks are doing in different disciplines or domains. So I think there's sort of that piece and I think to build on that, I think more broadly, social science is nice because it is sort of a combination of, I think, thoughtful sort of data and statistical work, but also layering that with theory and sort of like understanding qualitatively the sort of mechanisms you're analysing or sort of the domain that you're analysing. So I think like those two things from my perspective have been really helpful and have helped guided me a little bit sort of in sort of coming into this space, not having a h.r. Or sort of talent background, but not being in People Analytics for about 8 or 9 years now.

Cole: Yeah. Well, one of the things I think is actually a strength of our field is the diversity of backgrounds that come into our field. Do you believe there's any kind of like unique superpower that your background brings into People Analytics that may you may not have seen others have?

Vincent: You know, I mean, I'm sure others do have this. I don't want to sort of say it's unique to to me, certainly. But, you know, I sort of come so my background is in political science, but a lot of my focus was in grad school on econometrics and particularly microeconometrics. And, you know, I think that training is super helpful because oftentimes you're focussed really carefully on, you know, using data to identify sort of causal mechanisms, causal effects. And so you're thinking really deeply about sort of the domain that you're studying, but also like like what could be confounding factors. How do you isolate sort of the the variable of interest? And I think, you know, and People Analytics, that's been super helpful because oftentimes we are sort of trying to isolate a particular thing or measure the impact of a particular initiative or program. So I think that training has been very helpful in terms of giving me the tools to sort of like analyse and answer those questions. But as I said, I'm sure other folks come from like an econ background and have that, but that's been very helpful for me, layered in with, like I said, some of the other social science work that I've done. I've done work in experimentation, behavioural economics, so I try to bring all of that into like the work that, that we do at Shopify.

Cole: I love that bringing in experimentation and behavioural economics. I think that's, that's a really kind of untapped area that we have to focus on. But so we kind of went way back. Why don't we go to the present here for a moment? So you're now leading People analytics at Shopify. Do you want to tell us a little bit about what that experience has been like?

Vincent: Yeah, it's been it's been great. So I've been at Shopify now for around 11 months and previously I was at another tech company, HubSpot. And I would say the unique thing about Shopify, or at least this particular role is so I'm a a director of data science and I don't sit within talent. I actually sit sort of in our data and engineering org. And so we have like a dotted line relationship into our talent organisation, but we actually don't report up through our chief people officer. So it's a little bit of a, of a different sort of working model and it's one that I've not done previously. So this is like the first time sort of not working directly with sort of talent folks, which has been interesting. I think there's been really interesting and sort of helpful aspects to that. Also, of course, like challenges. So yeah, I think it would be interesting to hear if others have sort of gone through that experience, sort of what their their viewpoint is. But it's been interesting so far sort of doing that.

Cole: Yeah, I won't say that you're quite maybe the the first monkey being shot into space, but you might be in this regard. And frankly, it was one of the reasons why I was excited about talking to you today. Um, but can we talk like. So let's dig into this for a minute because think, think we like, we kind of did a cursory pass. I want to dig into it. So you're sitting in a, in a functionally basically an engineering function and you're not reporting into like a chief people officer or someone like that. Are they a customer of yours? And if not, who are your customers? And and how is that different than perhaps sitting in a talent or a people function?

Vincent: Yeah, it's a great question. So so, yes. So talent is certainly one of our primary customers. So we obviously support all of the data and sort of systems work that sort of is required to run our talent organisation. But I think maybe somewhat uniquely, there's also just other Shopify wide initiatives that talent of course touches, but it also impacts the broader company that we also will sort of play a lead role in. You know, we can get into some of that I think, later in the conversation. So one of our bigger stakeholders, of course, is our CEO and founder, Tobi Lutke. So we work directly with him on a number of projects as well. We have like a direct line of communication with him alongside our chief talent officer. And oftentimes like, like the Venn diagram is such that like there needs of course like overlap. But Tobi of course is, you know, a very technical sort of developer at heart software sort of engineer. So I mean his ass oftentimes are sometimes a bit different or his focus areas are going to be different, oftentimes with the same like underlying idea or like the things that they want to do. But in terms of approach or what he actually wants built, it's often a bit different than what like a chief talent officer will ask for. So that does create an interesting dynamic where we're sometimes building things custom for him and then also things that are more broadly for like the talent organisation.

Cole: And would you say I mean, I think I know the answer to this, but it sounds like that's a pretty atypical experience of a People Analytics leader. And you've had prior experiences maybe seeing a different setup at other organisations. Can you contrast those two things for us and like what are the pros of having this? Because I would say you have an exceptional level of buy in to be working directly with the CEO, doing special projects for them and making that kind of Can you contrast the two and tell like pros and cons or something along those lines?

Vincent: Yes. Yeah, for sure. And and yeah. And I sort of look at it as like things that have been really fun and exciting to start. And I think the cons are more probably just challenges or things to sort of work through as we sort of continue to develop the model. But I think like the first and foremost, foremost in terms of like advantages or things that have been really exciting is as you said, I think one like the buy in, but really sort of the focus on sort of the product and engineering side of People Analytics versus like more of the consultative side. Not to say the consultative piece is not super useful and that's sort of where I've grown up and sort of People Analytics. So I think there's a ton of value in that. But in terms of like flexing a bit of a different muscle, the emphasis at Shopify is certainly a lot more on infrastructure systems, scalability. So the way we approach sort of data problems and the types of questions we are answering oftentimes is more with a product mindset and so we can talk more about that a little bit later. But that's been sort of an interesting sort of wrinkle and something that's pretty unique and also like a definitely a different set of challenges and sort of interesting sort of problem spaces, I would say that's been great.

Vincent: I think the, you know, alongside that, just like the ability to spin up resources super fast and sort of the cadence at which we work. So we, you know, will work oftentimes in like 1 or 2 week sprints and pretty like major projects. And so like the, the speed at which we're doing stuff and executing is really fast and like stuff moves and if you need something given the criticality of lot of the work that we do, we just get the resources that we need to execute. So that's been awesome. I haven't had to really ever make like a super hard, like, like, you know, ask for like resources. I think whenever we needed something, we were able to get it. So that's been, you know, really interesting. You know, I would say the challenges or the things that have been sort of a bit of adapting to is like one, as I mentioned, this sort of idea of supporting initiatives that are for all of Shopify versus like something that's a bit more talent specific and like how do you prioritise or like what's the right way to sort of, you know, sort of stack rank? Some of those tasks has been a bit interesting and challenging since we also don't report to either of those stakeholders, right? So in terms of setting our agenda and roadmap, it is a bit interesting because it's not as if my my manager is sketching out like what it is we should be doing because, you know, she sits in like data engineering and has no sort of background in talent.

Vincent: So a lot of it is trying to figure out as a leader, like, how do I sort of spend the team's energy and like, where do we invest time? So that's been both a challenge, but also something that's, I think, a good opportunity. And I think the last thing or the second thing is just, yeah, getting that visibility into talent needs and like building that deep empathy is a little bit trickier. Of course, when you're not sitting in the org directly. So you have to like be more intentional because that is something you take for granted. If you sit in talent like you know all the. Players, you know, like what's going on when you don't sit in that specifically. It is a bit trickier to find ways to like, really empathise with your, your stakeholders, you know.

Cole: Well, now I've got so many different directions. I want to go after what you just said, but. Well, so guess let's. Let's stay. Let's stay with the thread that we're on. So how do you build those relationships when you're not in a talent function? Like, how do you overcome those barriers?

Vincent: Yeah, I mean, it's sort of interesting, right? Because I think the other thing to note here is that Shopify is a digital by design company. So we are fully remote. We do not have like a dedicated office space, like we do have in-person meet ups and things like that. But we don't have an office where folks all go. So again, like you sort of layer that on top of the fact that you don't sit within talent. I do think it creates some interesting challenges. But that said, you know, look, one our chief talent officer is amazing. She's she's a fantastic partner. I have a great relationship with her. We meet, you know, as needed, like all, you know, talk through sort of projects and priorities. There's also a weekly, you know, talent leadership meeting that I attend. I have sort of regular check ins with each of our talent leads. So, I mean, there are mechanisms at least sort of more real time mechanisms to sort of understand that, you know, we also are very intentional on Shopify around just posting like objectives and things that we're like we're, we're sort of aiming to do over the course of the year. So I think there's usually not many questions around like what the main goals or priorities are. And I would say last year, particularly at Shopify, a lot of our biggest company initiatives were talent ones, so it was pretty obvious where to focus energy. So there wasn't a lot of ambiguity or question around like what we should be doing or where we should aim the team. So we'll see how that evolves over time. But that's been sort of the way, yeah.

Cole: That kind of got into the next question I was going to have because you mentioned, you know, with having a supervisor that's not necessarily a talent has like a talent background. You know, you've got the CEO over here, obviously not the head of town or I guess arguably they are, but then you're not in the talent function itself. So how do you set those priorities? But it sounds like with the company is so invested in the talent space with what they're already doing, that problem kind of solves itself, right?

Vincent: Absolutely. Yeah, I think that's right. And I would say like our our CEO is quite interested in sort of the talent aspects and certainly thinks of that as one of our competitive advantages. So he's very deeply involved in a lot of the projects and work that we do. And so as you said up front, like the problem or the the challenge has never been around getting folks to pay attention or to be invested in sort of the work. It's quite the opposite. It's more like how do you ruthlessly prioritise and like really like go to first principles around like where you need to focus because things are happening super fast and we're getting pulled in a lot of directions. So it's one of those things where it's like, how do you just get disciplined around where to spend time and make sure the team is focusing on the right things?

Cole: So how do you go to first principles in this space? Because I find this as an under-discussed topic for sure.

Vincent: Yeah, I mean, look, I'm sure everyone has sort of different like mechanisms and ways through which we do it. I mean, I think for me, you know, one of the big things is like, you know, one, like ensuring we understand like, like fundamentally where we're looking to go like, and how do we work backwards from that. So like, what is like the end outcome that we're driving towards? And if we sort of start from that and work backwards, how do we get there? So I think that's sort of a useful sort of mechanism to sort of understand that. I think with that will come like surfacing assumptions. So like what are we like thinking about? Like what are, what are we assuming is like, you know, either already in play resources that we have things that are already in motion. I think, you know, uncovering some of that and then really sort of going to like the what are the zero sort of priorities, the things that are absolute non-negotiables that have to get done in order for this to sort of move forward. So, you know, I think that's been the big thing. And of course, we work, I think like many like product and tech companies around, like the idea of like MVP's minimum viable products.

Vincent: But I think the difference is like the standard, the quality bar is like pretty high at Shopify. And so I would say we always say like an MVP is not an excuse to like ship a really crappy product, right? Like, like you can't just be like, Hey, well, it's an MVP. So, you know, so, so the idea really meaning that like you, you have the MVP still has to work really well and be useful. It can't just be something that you cobble together that just doesn't do what it needs to do. So I do think, like our thinking around MVP's is I want say it's different, but I think there's a bit more of a strict adherence to like maintaining a high quality bar. Even if you are trying to scope something that is sort of smaller in nature to get sort of traction and movement. So I do think that's another way we've sort of used that to like get to the the essence or like boil down like the thing that we need to do and make sure we get right.

Cole: Yeah. Well, talk, talk to me about because I want to go back to what you were saying earlier about being a product driven function and doing things through sprint cycles rather than, I guess first in, first out or something like that. Can you talk about that and has that been an adjustment in your career or have you always operated in that way?

Vincent: So I would say more recently it's been sort of a motion that I've gotten more comfortable with. I think it's, you know, I'm interested in your perspective, too, and sort of being in a few different places. But I do think a lot of, you know, there is a good amount of work sort of moving towards that idea of like not these long like waterfall type projects where you take six months to do something. And I think a lot more emphasis on getting things out there quickly and getting feedback. So I do think even if it's not like in the more like narrow parameters of like a sprint, I do think generally the work is moving towards something that's like shipping faster. So I would say that's been an okay adjustment. But I do think, as I said, given a lot of times we're working on really big things. So like, you know, obviously like Shopify talked about last year, how we overhauled our compensation system. And you can imagine that's like a pretty massive body of work. And I remember I think I was telling you this story, but like I was in sort of Toronto for like two weeks, sort of helping to to work on that project. And when you, you know, think of the idea of like, all right, cool, we're going to do all this, this massive undertaking and like try to do it within like a two week span.

Vincent: Like you, you have to be, again, like, just so ruthless with like, what you focus on, like what absolutely needs to work perfectly. Where can we iterate? So I do think like it's, it's really strengthened. I think some of those skills in sort of doing it with like high stakes, which has been I think a bit more of the case at Shopify. But again, I think just to like round out, we're just all about shipping all the time. So like even like managers and leads, I think there's a notion that like, hey, when I'm leading a team, I don't need to get my hands like sort of dirty there or I'm sort of like overseeing things. And I would say, like that is not like sort of the culture at Shopify. And I think all of our managers, directors, whoever you are, I think everyone is seen as builders on some level. So I think there's an idea that, you know, no matter what your role is, you are sort of actively involved in shipping things and actually getting stuff out into production and that's been fun. But again, like, yeah, a little bit different from what I've seen in other sort of organisations.

Cole: Absolutely. I mean, I think the thing that I'm really loving about this conversation and why I was excited to have it is because when I talk with other People Analytics leaders, it is very uncommon for them to say things like we're ruthless, ruthlessly prioritising or we're constantly shipping things like that's just not in our lexicon. And so if you haven't kind of noticed from the type of questions I've been probing you on, I have this like overarching meta question or kind of this corpus I'm trying to build here, which is, is being a product function or is being a People Analytics team in an engineering function, like are you truly the trendsetter? And this is going to be the thing that the future of People Analytics looks like and everybody better get on board or they're going to be, you know, a dinosaur or something like that. And like, what's your perspective on it? Because like, I feel like what you're doing is very unique and I'm very excited about it.

Vincent: Yeah, well, thank you for that. That's super nice of you to say. I you know, I would say it really is a matter of I think you just I think as a leader. Right. You have to sort of read the room and sort of understand your company culture. So it's not to say that I think every company would operate the same way as Shopify and such like leaning heavy into like product is the way to go for everyone. I think you have to sort of understand your landscape and like where there are opportunities and where you need to focus. So, you know, I don't necessarily think it's something that would work for every company. But that said, I do think there are principles through this work that have been very helpful that I think can really scale across most People Analytics teams regardless of sort of your environment or landscape. And I think even just going through, you know, the product principles at Shopify and understanding those, it really shed light I think too, like on just some of the mistakes I've made sort of in the past and like building, you know, tools or products and other sort of, you know, People Analytics functions that maybe didn't sort of have the same impact or didn't work as well. You know, I think just a couple off the top of my head, right. I think one of the things that, you know, we emphasise quite a bit in sort of building a tool or a data product is really around like this idea of making like the most important things super easy to surface and to use, but also giving flexibility that you can do a ton more if you choose to do so.

Vincent: So like the idea being you want to build something such that like it's so easy to get to the most important insights, but you also want to acknowledge there are like super users out there that are going to want to do a ton more things. And what you don't want to do is have that built on a platform or a product that can't support that. So you want to be able to sort of build something such that it can scale with sort of the user and sort of their needs. And so we spend a lot of time. And I think what that really means oftentimes is you have to work in almost you have to almost counterintuitively focus on like the more complicated aspects of building something first and get that sorted, like all of the backend sort of systems engineering the infrastructure such that then it's very easy to build sort of that functionality versus I think the normal way of doing it where people start super simple and hacky and then try to introduce more stuff as they go. And I think what that ends up really creating is more complexity that's hard to unwind and then it actually ends up you rebuild the whole thing. So I think oftentimes we spend more time up front trying to, you know, sort through the really sophisticated bits such that it's much easier to work backwards from that and make it super simple and not have to redesign it later on.

Cole: That's really, really interesting because like, I know three of the kind of platforms that you've built. You know, you mentioned the compensation system. You also have built a surveying tool as well as a performance management system. And if you'll you'll bear with me for a second. At a prior organisation, we actually built our own surveying tool and performance management tool. And the challenge that I found is I thought I knew performance management, I thought I knew surveys until I decided to build a tool that did it. Yeah. And that's when you realise like, oh my goodness, this is way more complicated than you think. And you have to have such a ground level understanding of some of these things to do it. Can you, can you talk about that process at all, what it's like truly being a builder?

Vincent: Yeah, for sure. I mean, I think I think that's right. And I think, you know, often times, you know, I think the sort of questions that we ask up front and sort of encountering, you know, this idea of like, hey, how do we build this is really thinking about, well, what is the you know, what is sort of the the infrastructure and data models that we would need to even do something like this? Because oftentimes when you sort of start from that premise, you actually start to uncover all of the trickiness and like the challenges, as you sort of said, and actually getting something to scale. So I think oftentimes, you know, that's where we start. And and that's not to say that like there are projects or problems where you don't need to do all of that. Like it's, you know, I think we often try to think about the work we do as like is it a like, is it infrastructure, is it like a feature or is it like an experiment? And oftentimes, like I think in thinking about that, you you sort of uncover, well, where do you need to focus or what's going to be the sort of priority to make sure you get it right in order to sort of scale something? So, you know, again, I think part of it is really understanding, okay, if this is where we want to head, what is the sort of underlying like back end need to look like to support this? And, you know, I think in doing that, it's led to it's created some upfront work, certainly, but it's allowed us to move much faster once we have that. And I think the comp work was a really good example of like a lot of where the effort was, at least from our team was in building sort of the, the process.

Vincent: And for those of extract, transform load, like the ETL process on the data side to be able to basically spin up like the state of the world like, like what is compensation currently look at Shopify and how do we adjust? Like here are some knobs, things that we can adjust if we were to like try something different, you know, as a company, here's like, we can do that and here's like the UI to do that, and then here's what the impact would look like. Like here's what would change and here's what that would actually do for us. And so have that sort of feed into like a reporting layer that sort of shows decision makers, hey, like here's the the impact or the implications of making X, Y, Z changes to our comp system. And I think having that super tight feedback loop where you can just do that and do it infinite number of times because you have an awesome sort of data model, you know, in sort of an infrastructure sort of underlying it, it allowed us to move immensely quickly, much, much, much faster than if we did something like that was like focussed on a very specific problem. And then someone came back and said, Well, but what if we did this? And then we had to rebuild the whole thing to support answering that question. Building something way more generalisable allowed us to basically cover anything anyone ever wanted to see. And so we were able to like move a lot faster once we had that built.

Cole: Absolutely. Well, I imagine if if a existing People Analytics team tried to pivot to this type of model, they would probably struggle a little bit because they might not have the right capabilities. So what capabilities did you have to, you know, buy or build to to do this, to build a team in house? Excuse me.

Vincent: Yeah, sorry. Yeah, it's a great it's a great question. And so I will say like I am a bit spoilt in that way because I think, you know, when I joined Shopify, the PR team already existed was a great team that was already starting to do amazing work. So I can't take certainly the credit for all of the awesome stuff in terms of how that team was constructed. But and obviously building teams myself and then also now inheriting a really high powered team, I would say, you know, again, depending on what you're like aiming to do, if you are sort of focusing more on the product side, you know, I think the investment we've made on, you know, data engineering and data developers sort of both back end and front end again, like I think that's been immensely helpful. And I will say data science at Shopify is more engineering focussed. So all of our data scientists are expected to like build extractors and like work on data pipelines and do data modelling in addition to the more I would say like research or statistical based work. So I think because of that strong engineering focus, everyone sort of comes with the basis of like understanding how do we set up the right infrastructure and systems to be able to answer the questions that we want to answer. Now that's not to say, though, that like on top of that, having just deep domain expertise and really understanding how to frame questions and use data to answer questions, which is not a pure certainly a pure engineering skill is also very important. So I think, you know, our team superpowers certainly is more on the engineering and infrastructure side. I think what we've made amazing strides over the last 12 months is really deeply understanding the business and our CEO and like what it is he wants. And I think with that, we're able to sort of really focus on the right things at the right time, which is, I think, unlocked a lot of impact.

Cole: Well, let's dig into that a little bit because. One of my immediate thoughts is, wow, having a team with that, that level of skill must be extremely expensive. And so I also think that that means that the team has to be producing a ton of value on the other side, especially, you know, I think about the current context of what's going on out in the world with, you know, cost pressures, inflation, layoffs and those type of things. Teams have to be kind of justifying their existence and to do that effectively and you've got that right level of buy in from your leaders. How does your team navigate that ability to make such an outsized impact to ensure that it's going to navigate these kind of uncertain waters effectively?

Vincent: Yeah. And look, and I think I think that goes back to what we were talking about earlier, which is really around prioritisation and focusing on the right problems. Because when you think about some of the work that, you know, maybe some People Analytics teams have been doing over the last year, year and a half, right? You have things around, you know, retention efforts during the height of like, you know, sort of the labour market sort of movement and everyone sort of jumping jobs and all that sort of good stuff. You know, there was a lot of emphasis on retention and obviously there's a huge cost impact to to that and sort of the opportunity cost of losing someone. So there's sort of that work. There was, of course, more recently things around like, you know, layoffs and sort of downsizing. And how do you think about the economics of that also huge impact. You know, obviously Shopify, we were working on compensation, which is an enormous cost. Obviously, when you think about it from the context of the entire company. So I really think a lot of it is like just aiming at the right problems and making sure that in what you're doing, you're sort of focusing on things that are of high value and high impact to the business. And I always tell like newer leaders starting out. Particularly if you're not maybe in the position that I'm in a Shopify where you're already, like you said, have a bit of that buy in and it's more just like prioritisation, you know, I think if you're starting out and trying to build a brand and build sort of a focus area, I tell folks to really think about your work as like a portfolio and in any sort of portfolio you're trying to find some sort of mix of work to sort of mitigate against risk, right? And so what that means, I think practically is like finding some of those like quick wins that everyone talks about that like, you know, are hard to find, frankly, nowadays.

Vincent: I think we've all maybe hit a lot of those quick wins, but like there's usually some things you can do fairly quickly that won't have the highest impact but will have some measurable impact. And then there may be longer term, bigger things, whether it's on the infrastructure or system side or data side that are going to take longer to do. But you still need to sort of start working on those things. So I try to like tell folks having this bundle of like projects such that, you know, you are sort of safeguarding against a few different scenarios allows you some of the time to build up sort of that, that, you know, sort of repertoire of like projects work and like that resume of like here's the value we've added. And that way you're not like sort of just going after things that seem pie in the sky. Like really? Yeah, huge, huge, like savings, but like, you're not ready to do it. You can't work on comp yet. You don't even have the data in a, in a data warehouse to sort of do the analysis. Right? So I mean, again, I think you have to sort of meet the company where it is, but also think about your work as that portfolio and figure out where you can sort of optimise it to find that balance such that you can buy some time to get to those higher impact projects. Yeah.

Cole: I feel like there's an article to be written there about if you're approaching your team like a portfolio, how to lead people analytics, like a hedge fund manager or something.

Speaker3: Yes.

Vincent: I'm sure there is. There is, you know, some theory, right? A portfolio optimisation that you can apply to like a people setting. But I don't think it's a bad instinct, frankly. I do think like because again, I think maybe there's a tendency to overindex on any one of those things. And I think that's a mistake. Because I will say, too, is if you're going after only those, you know, quote unquote, high impact, like, you know, like big cost saving measures, you lose the opportunity maybe to innovate or do something that's, you know, sort of not on folks radars. And I don't want to sort of give the impression that, like, the work that we do should have just practical impact all the time in like the very narrow short term sense. Of course, some of the biggest I think, wins in People Analytics have been projects that are longer tailed and more exploratory in nature. Obviously you think about all the stuff that Google did way back when, but a lot of that like wasn't tied to like immediate impact, right? It was something that was an investment. And so you do want to make sure you're building some space to find some of those opportunities in addition to like the more immediate stuff that's in front of you.

Cole: Yeah. Where are you long and where are you short? You got to show those positions for sure. Well, let me let's dig into this a little bit. So you're working with a CEO, right? And I wanted to get into one piece of nuance that comes with working with a stakeholder about the difference between being data informed and data led, especially when within the frame of reference of being kind of evidence based in people analytics. So can you talk about that at all in terms of how you're helping influence decisions for leaders in your organisation?

Vincent: Yeah, I really love that framing and, you know, and I think one of our tenants at Shopify is to be sort of data informed, not data led. And I think, you know, the reality is, you know, we think about like research and data as like. Important tools that are like one set of inputs to help us understand sort of the problem space or domain. But as everyone who works in data knows, like their data has a ton of issues, there's a lot of context missing from it. There's a lot of domain and subject matter expertise that's not sort of embedded within or encoded within data in sort of the immediate term. And so like, you know, this goes back to the social science sort of piece that I was mentioning earlier. But, you know, I don't think it's controversial to say like they're these are additive things, right? So like, you know, the more you can sort of bring, you know, not just data, but some of the subject matter expertise or domain knowledge to sort of elevate or uplift sort of the data, That's sort of the I think that's sort of the secret sauce, right? I don't think anyone sort of looks at data or should not look at data As Yeah, it just tells me what to do and I go and do it. I think we all expect everyone to engage with their domain in a way that they don't just blindly sort of follow what the data says and they have enough sort of understanding to sort of push back or ask questions or be sort of interpretive of the data to make sure that they understand what it is they're doing.

Vincent: So I would say like that's been that's a huge tenant for us. And I would say not only is that I think just better like science and research, right? I think also it's of course just a much better way to get buy in from stakeholders, particularly those that maybe are not from a data background. So if you come in, you're just sort of like, look, the data says this and you're not doing it, ergo you should change. You know, I think if anyone's maybe tried that doesn't go typically doesn't go very well or you're not going to get, I think, a ton of traction with that sort of mindset. So I do think you have to approach it a little bit of a humble inquiry and sort of use data as an input. But certainly that is not the only piece of information that one should use when sort of making a decision. So I think that's for us the way we try to approach it. And certainly we've seen a lot of value in that, not just from like the quality of decision making, but I think also in terms of building more connective tissue with some other teams that are not data teams but need to sort of like implement some of the stuff that we're actually working on.

Cole: Yeah, absolutely. Well, tell me about this. So being data informed and making data informed decisions, usually in the context of People Analytics, that means that we're providing data informed for others to make decisions. One of the things that I've been noodling on lately is this concept of should People Analytics itself be the decision maker rather than only informing others decision making? I don't know. Can you talk about that and have you thought about that at all in the context of being data informed?

Vincent: That is interesting. I have not thought about that super deeply to date, but I think that does make I think that makes a ton of sense. Conditional on I think the idea that you do know your domain well. And and I think oftentimes and I sort of alluded to this earlier, but I think sometimes maybe a challenge within our space is that if we don't sit super closely with some of the practitioners or we don't have that like on the ground sort of understanding, there is sometimes a disconnect between like what it is we're sort of either recommending or sort of like extrapolating from the data versus like what the reality is or like what are the pitfalls or challenges we would maybe face in implementing something that we don't sort of have the context on? And I will say earlier, my early in my career, I certainly ran into that where, you know, and I think everyone maybe has this example of like going off sitting in a room, building a model sort of devoid of context, finding something interesting, and then, you know, going and being like, Hey, look at this cool thing. And then someone very quickly, you know, sort of, you know, is able to show the flaw in your reasoning or like the reason why you sort of see this trend. And so I will say, like, if you can build that sort of domain expertise and you can sort of come to the table with that knowledge, yeah, like why why wouldn't we also be a part of that decision making, I think body Right? Like it makes a ton of sense that we would be involved there. But I think it is sort of dependent on you investing the time to really understand the sort of space that you're working in.

Cole: Well, tell me. And this. This could be just a lark. So tell me to switch gears if it is. But in doing research for the podcast, I found that you were a contributor or a researcher as a part of this group called Empirical Implications of Theoretical Methods. Or Yeah, is does that have anything to do with what we're talking about right now? And what is that group?

Speaker3: Yeah, yeah, yeah. No, that's awesome.

Vincent: Yeah. So it was, it's a, it's a program that's like NSF funded. It was something I did in grad school. I spent a summer at, at Princeton sort of working there. And really the, the focus is like, how do you bring to bear like formal, like mathematical and economic models to like practical applied sort of problems. And and I think, again, this sort of ties back to some of the econometrics pieces that I talked about earlier. But oftentimes, like like models are really useful, right? Mathematical models, economic models are really useful in part because, one, it sort of surfaces assumptions that you're making in a very explicit way. So you have to put that on paper. So I think that's always important for understanding, you know, how you're approaching the analysis or the work that you're doing. But two, it's like logically consistent, right? So again, like in sort of doing work in that way you're able to derive like hypotheses, like what you would expect to see given sort of a formal set of rules. And then you can go and use data and collect data to test against those.

Vincent: And I think that's certainly how science I think should be done or generally should be done, and it's a great toolkit for that reason. So a lot of the work that I did, there was a lot on just like, how do you apply that? And so before I was in People Analytics, I was actually in program evaluation, and a lot of the work I was doing was sort of measuring the the impact of energy efficiency programs and like how do you sort of quantify if someone participates in like an energy savings program, how do you measure how much energy was actually saved in doing that? And like, how much can you attribute to the program? So it's like a really nice mix of like you need to have theory and like really understand like, well, what's the mechanism? How would this actually work? And then like, how would I collect data in a very rigorous and robust way to sort of answer that question causally and not be sort of subject to confounding relationships that, of course, exist in the in the real world.

Cole: Well, let's let's talk about that, because I feel like the elephant in the room of doing science and experiments is this concept of causality. And I know from my research background that people only speak about causality with trepidation and that kind of thing. And so what role does causality and causal inference play in the work that you do with People Analytics?

Vincent: So I personally think it's a really helpful set of tools, not because I think you can answer everything in a causal way, of course, because I think you're absolutely right to say there's a ton of challenges in doing that, particularly if you don't have a experimental sort of like a randomised controlled trial. So I do think there's challenges there. But, but again, like the way you then would approach it, right? So most of econometric work is sort of trying to approximate an experiment. So how can you set up an observational data study or like a collecting data in a non-experimental way? How can you set that up in a way such that it gives you some of the same leverage you would get from an experiment? So a good example, right? And something we did at HubSpot, which I was really proud of. I think we had an analyst work on this, which was amazing, a really good work, but it was sort of at the at the very, you know, sort of onset of the pandemic. Right? And, you know, we had at HubSpot, we already had sort of a remote work population prior to prior to the sort of pandemic. And then, of course, we had sort of office closures on a very specific date. And we had this really interesting natural experiment where we had a bunch of folks that were in office all of a sudden had a transition to remote. But we also had this sort of control group or comparison group that were always remote. And we had sort of, you know, pre post the office closures, those two groups that we can compare.

Vincent: And so I think it's stuff like that where if you can find in your work like either quasi experiments or sort of natural sort of phenomena that sort of happen that set up an experimental framework. So like the remote work, one is a very obviously popular one, I think one that maybe gets less attention, but I think also could be very useful. We haven't done this work yet, but I think about like a strike prices for restricted stock units, right? So strike price is like when you got the equity from the company that you're in. You know what the the price of the stock was at that time. That's sort of exogenous what we say, right? It's random in a sense. It doesn't sort of tie into your decision to join the company per se. And everyone has a different value. And you can imagine like understanding that variation. And since it's sort of you can sort of approximate that as sort of randomly assigned. What the impact of that is over time. So what are the delta in strike price? What does that mean for like retention, like over the first couple of years? Like does it have any impact? And that can maybe tell you something about the relative impact of, you know, equity grants as a retention mechanism, like stuff like that, that you can I think, you know, just think about the natural world and say, hey, like, where are these opportunities to sort of collect data in that way and answer a question a bit more of a causal sort of approach?

Cole: Well, and you can take the within subjects in between subjects design, because within subjects, people who came in with the same strike price and between subjects, those who came in in different cohorts or groups at different strike prices too. I love I love that mindset. And you're you're a man for my heart, Vincent, in the sense I love natural experiments. And I think we even talked about it on a prior episode of this with Alec Levinson with his economics background. Yes. That was that was definitely something that I'm a big fan of. Maybe we switch gears here for a moment because in acts that I've been grinding lately and I'm curious just because of your perspective now leading a People Analytics team that's outside of a people function is career paths for folks in People Analytics. Like what? Like if you're a new entrant into People Analytics, you're coming in into an individual contributor role. What is your career path and are you going to have to do organisational? Plinko Like I see most People Analytics leaders do nowadays is just go between organisations to have a career in People Analytics or maybe should we all be just jumping ship into our engineering and product functions right now? Like what? What's your perspective on careers and People Analytics?

Vincent: Yeah, that's a great question. And you know, I can certainly speak from my own experience, but also just from, I think, managing others over the last, you know, ten years or so. But my advice always is to like, I try to build skills that are agnostic to like a domain, right? So like, you try to find things that I think would translate well outside of a specific sort of area. And obviously that is dependent somewhat on your own interests and like what you want to do. So if you love talent and want to be in sort of that space for your career, that's totally fine and makes a ton of sense. But that said, you know, I've always tried to make sure, you know, the skills that I were I develop are sort of translatable to other sort of disciplines and domains. I think the good news with People Analytics in general is that like it really is at this interesting intersection of like, you know, data and statistics, sort of business consulting strategies, sort of that like talent strategy, acumen, the business acumen and sort of now more recently, as you said, sort of more on like the product side. And so I think if you are setting this up such that you are getting a little bit of exposure in each of those domains, I do think like it gives you a lot of flexibility. And I can say even in my own career, while I've stayed in People Analytics for for a good portion of it, I've had plenty of opportunities potentially to go like outside of PR, I've chosen always to stay in it because I like it. But it's never been a matter of not having the right sort of skills or experience. So I do think for others that are maybe joining this space or starting to develop those skills, I think they're very transferable. So I think just continuing to lean in there.

Cole: Yeah, absolutely. I like that answer. I like that a lot, actually, of just, you know, if you've kind of got an antifragile set of skills, you can virtually go and do anything that you want, right? Totally. So I love that kind of methodology. I think we're kind of getting near the end of this on time, but I want to kick it to you. Vincent, is there anything that we haven't covered that you want to make sure that the audience is aware of?

Vincent: Well, thank you once, first and foremost. I mean, thanks. This is a fun conversation. Really appreciate you having me and just being able to talk through these things. It's been it's been great. And yeah, all I can say is, you know, sort of going back to my point earlier, you know, I do think like there's a lot of attention often paid to like the, I would say the the sort of shiny things around like analytics and data and sort of getting into the more sophisticated methods. But I think, you know, building sort of that really strong like detective skills and intuition around like how to approach data is just so valuable. So I would just say the one thing I would love to leave folks with if you are starting out in this space is just hone in on that. Don't get distracted by the, the super fancy things like really have a good strong understanding of like the underlying stats, the data piece, the sort of how to answer and ask good questions. That stuff is so much more important and transferable compared to like niche skills. So that would be my advice there.

Cole: Well, actually, let me ask you this real quick before we wrap up. I think this is important. Does someone need to have a PhD to be good at People Analytics? Like is that a necessary upon entry to have the skills that are necessary to have a career?

Vincent: No, no, for sure not. I mean, I don't think that at all. I think it's a matter of you can get that through PhD training. Right. And doing sort of that. And I know you know that as well. But you know, you can absolutely do that in an applied way and just sort of do that right at a college or not even in college, frankly, you can start working on some of these things, you know, doesn't require even a college degree. I think a lot of this stuff is, you know, doing the work to, like, understand things and of course, like doing some self teaching. But a lot of it is like applied. You just have to do it. You have to get the reps. And that's maybe the one challenge of PhD programs, right? Is like you don't always get that more direct exposure of actually working with like hard data or things that are super messy. You get the more curated like easier things to sort of work through. So I will say like that sometimes is more of a detriment than it is a help.

Cole: Well, that's why actually I love People Analytics because the PhD prepares you with the theoretical, the scientific, the, you know, statistical, the technical backgrounds, the experimental to really be able to understand some of these things. And then People Analytics gives you the repetitions, right. And the ability to apply it in the real world. And so I think it's such an exciting discipline because of that. And I think you can, you know, especially as time progresses with the amount of resources that are out there online, you can find a lot of these things and do a lot of that self teaching, like you mentioned. But this has been an excellent discussion, Vince, and I've really been looking forward to it. And and so I thank you so much for joining today. Um, but this has been the People Analytics World Fireside chat. I'm Cole Knapper from Agnostic and Vincent Greco from Shopify. Thank you so much for joining us today.

Speaker3: Thank you.