Episode 23

Ron Chan on Balancing Technical and Soft Skills in Data Roles in the Age of AI

TIM: Ron Do you feel there's a balance between technical skills and soft skills for data roles? Where do you think that balance is? Maybe more interesting is how that's changing with AI. Do you view

RONALD: I guess in the space that I work in, which is health and, more specifically, real-world evidence and real-world data health the hard skills The technical skills are important, but I've always had this mantra, which is it doesn't matter if you have the best analysis in the world; if you can't explain it to the other person, then it's meaningless; it's useless. So that communication, that soft skill, being able to present, is becoming more and more critical, particularly, I think, as we move on to the move into the AI type age where you can do it quite quickly. With the help of AI, write SQL scripts. write Python scripts So I think my sense is that those soft skills are becoming more and more important.

TIM: Yeah, with Chachapiti picking up the slack of the basic bits, I wonder if then it sounds like the profile of who would typically be hired into a data analyst/data scientist role might also change if the blend of skills is going to be different.

RONALD: Yeah, I think so. I'm always high. I've always hired with a very strong intent of ensuring that the candidate has curiosity, and I think more than ever now, particularly that curiosity to go play with the new tool, to go play with prompts, to go test out features. I think that is the foundation of being a really good analyst or a really good data person.

TIM: Especially if we're in, let's assume, this kind of precipice of a very significant change in the way things are done, I guess if you don't carve out time to actually play around a little bit with these other tools, you might find yourself very quickly left behind doing things in a way that's now being 10x'd if you use AI to do the same task.

RONALD: Yeah, definitely. I remember when, like, before AI came out and when I was much younger, I always felt that way in terms of, Oh, I should really go spend time to go learn the latest Python, the latest SQL commands, or whatever is hot right now, the latest visualization tools, but I think now more than ever, given the rate of change is evolving so rapidly, it's much, much more important to carve out that bit of time to go be curious and go test and go play. It's just, yeah, harder to rely on remaining static now.

TIM: I've spoken to a few people at a few different companies recently—three of them, actually—who have basically said to their entire company, These are companies that sometimes, for thousands of people, stop working and don't do anything for the next month. Just use Claude or ChatGPT and rethink everything you're doing day to day. and it's a really interesting approach, which is quite bold, and I don't remember certainly in my career any point where there's been any new technology that has come along where companies have said something like that, and I guess the idea is to have like a step change in productivity and the way people are doing things. And almost like a forcing function, you could be so busy doing busy work, but if you don't have the time to actually explore these tools and how they apply in your day-to-day life, then it could be easy to miss out on them. Can you imagine that approach working? This kind of, Let's just implement it across the board as quickly as possible.

RONALD: Yes and no. I think it sounds like an interesting idea. I suppose in health it's still quite the data is still quite complex, and to be good with health data, you need to be good with analytics; you need to have some domain knowledge, and of course, there's the soft skill of communication. I don't feel like GPT or AI is at the stage yet where it can replace a lot of those skills. I think it can augment, so I think there are definitely benefits in using it to augment and make your job a bit faster and more efficient, but I don't know about replacing just yet, so the idea of just stopping what you're doing and going for it It might work, but I think a lot of guardrails need to be put in place on how you implement it and how you use it.

TIM: Particularly if you're using health data, I guess you can't just start pumping in people's sensitive health data to the cloud and asking an AI to come back with an answer.

RONALD: Definitely, but even take a simpler use case in the data that we collect in my program: we collect a lot of drug names, and the drug names are written in by the doctors, so doctors like to abbreviate, like to use acronyms, and they like to use shorthand, and we use AI to help us just at first pass to tell us what you think is the closest match. and sometimes the matches aren't even close, so without that domain knowledge, you would just assume Oh yeah, that seems correct, and then you would make a completely incorrect conclusion.

TIM: Where do you see AI most improving the hiring process? Have you seen it have any impact already? Like when you've been hiring roles yourself, what do you view? What are your thoughts there?

RONALD: not in the space that I hire in, and again I work for relatively small organizations in a relatively complex data environment not just yet, so still we are doing the traditional take-home assignments and having that interview process by the kind of presents back to us not to say that we're ignorant of the fact that with a take-home assignment we expect the candidates to run it through GPT, so I think the standard of what we're expecting is a lot higher nowadays but not necessarily having it replace what we do

TIM: And so you'd still be doing the, let's say, application, CV screen, initial interview, take-home test, technical interview, and that type of process.

RONALD: For now, and I think part of it is because in the space that I work, it's still relatively niche, so for any given role, there might only be, let's say, I'd be very happy to receive 10 resumes that were worthwhile.

TIM: So yeah, that's interesting. So if I think about other people I've spoken to recently, maybe in broader analytics areas, they are just getting inundated with CVs. Their main complaint is, I'm getting 500 applications. I could read them manually. I would read them and probably reject 450 of them manually. I literally cannot read them. I have to have some tool to do that kind of first pass for me, but it sounds like for you, maybe the market of potential candidates is quite narrow because they need the health domain experience as well as the analytic skills, so then, yeah, you don't have that kind of screening issue at the moment.

RONALD: Yeah, health analytics is generally a bit smaller in terms of the scale of organizations that exist. Definitely, as you say, the hard skills, the tech, the sort of domain knowledge, and being able to—and you don't have to be an expert in—I work in cancer research right now; you don't have to be an expert in oncology. A year 12 understanding of biology and the ability to quickly adapt and quickly learn if I tell you about a particular therapy area is really important. and then on top of that we're also looking for all-rounders where they can communicate; they can ideally do a tiny bit of project management on the side as well, so that makes it so that we're not so uniform in what we're looking at in terms of skill sets.

TIM: What about them thinking about what's possible? So if we think of where LLMs are at the moment, is there any step in the hiring process that you could imagine them being helpful in, or maybe not even thinking about current steps, but, oh, what is imperfect about the way we do hiring? What opportunities might there be for AI to help? for example, things like, I don't know, giving feedback to candidates or something like that

RONALD: Ah, it's a good question. So you mean using LLMs to speed up the hiring process in a way.

TIM: So, for example, some companies, okay, they're now using an LLM to do that CV screening, so they'll get the CV, they'll give it some kind of grading relative to the job description, and so we'll say, Okay, this is a—I don't know—a score of 80 versus a score of 60; here's its relative strengths and weaknesses. What I assume these companies are then going to do is share some of that information back with the candidates because instead of just giving them a rejection email, you could say, Hey, we rejected your application, but here are the areas why we rejected it. Here's how you could improve in the future. So there's probably some kind of ability to do feedback. Not that they're doing it yet, but I imagine that's where it's going to get to quite quickly. What about other things? Are there other things in the hiring process you feel like, either from your experience as a candidate or as a hiring manager, are like, "Wow, this is ripe for a bit of disruption? like this could really be improved potentially in the not too distant future with AI

RONALD: I personally use LLM to help me actually screen and draft up that interview process and to ensure that, for example, if I'm writing a job description or position description, I've got the key elements covered and what else I'm not thinking about—that's where I find it very useful to articulate the things I don't know and things I've missed. In terms of using it to screen candidates and particularly provide feedback, I'm not sure I feel like it's a bit of a brave new world situation going on there. I would like to think if I'd spent the time speaking to someone, they'd give me their own feedback rather than auto-generated feedback from a program.

TIM: Oh yeah, you're right. At the interview stage, I feel like the average candidate would expect human-customized feedback. You're right; maybe at the earlier stages, where it's just that you've just applied, you've completed an application and a CV, typically you'd get back just the what I call a sorry-not-sorry email. So maybe it's one step improvement on that; it might be AI. Yeah, it's not human, but maybe it's better than nothing at least.

RONALD: Hopefully the candidates are doing that themselves, though if they're not getting the job, they're feeding that job description and their own resume and going, Why do you think I'm not? I didn't get this job.

TIM: Yeah, there you go. Okay, that would be the kind of humble and smart approach to the rejection: to take that upon yourself to self-improve. I personally feel like the current narrative in the market is one of increasing paranoia from both sides. A lot of companies are saying Oh, all these candidates are cheating with ChatGPT. Every CV I get seems to deviate a lot from reality; candidates are saying Oh, the ATS is filtering out my CV; the company's rejecting me unfairly, blah blah blah, so I feel like candidates aren't in a good mindset to have that humility to almost step back and go Oh, how could I improve my odds of getting a job next time? but I agree they probably should at some level

RONALD: Yeah, it's hard when I wouldn't say cheat, but when candidates leverage these tools in unfair ways and potentially misrepresent themselves, I can imagine the frustration. you read it You read a cover letter. You read a resume that sounds really good, and the reality doesn't reflect it. When you have gone through the interview process or even when you've hired the person, I guess it goes back to the importance of having those soft skills and being able to communicate and articulate what it is that you've done, the impact that you've had, because you can get away with pretending to have the hard skills now.

TIM: Yeah, and someone pointed out to me the other day that actually maybe ChatGPT could be just as helpful for the soft skills, at least soft written skills, to someone who's got amazing technical skills but is a bit too humble, doesn't know what companies are looking for, maybe hasn't quantified their achievements well enough. You could argue maybe an LLM could be just as good at helping them craft their own message as it would be for someone with soft skills to do some of the basic technical stuff. What do you reckon?

RONALD: not so sure the first thing that came to my mind was it's not so much the communication of their soft skills or being able to persuade you that my analysis is the one you should follow; it's does the person doing the analysis or doing the work understand the business context in which they're operating. So in theory, yes, if you understood the business context, we understood the project, and you fed that all into the LLM, you might get a response that makes sense. My guess is chances are that the candidate may not understand those things, and that's why they're struggling, not because they can't get the LLM to help them.

TIM: That's interesting because, yeah, we both in passing mentioned kind of cheating, or I think you used the word leveraging the technology. How do you view candidates using AI tools in the hiring process? Is it completely fair game? Is it completely off the table? Should you use them because you're going to use them in the job? Like, how do you philosophically think about that?

RONALD: I expect them to use it like any other tool that's available now, so if they're not using it, then I'd be asking why not.

TIM: and is there any limit to that? So I'll give you an example: we were recently hiring for some sales roles, and one of the questions in our evaluation process was, I think it was, I'll imagine it's day one for you at our company, Alooba. What are the three things you need from us to give you the best chance of being successful? it's a question that to me Was very personalized, like I wanted to know their exact specific opinion on it So I could tailor an onboarding program for that human, and it was striking how many candidates clearly had used ChatGPT for that question, which I found very odd because I'm like, I don't give a shit what an LLM thinks about this; I care what you think. I'm trying to form a relationship with you personally, but then someone said to me, Yeah, but candidates probably view that as part of the evaluation, like anything else. So if they answer that question poorly, they could easily not get to that next stage of the hiring process, so I hadn't thought of that. I'm like, okay, fair enough, so it's just like part of the game, part of the test. From your perspective, is there any limit on where candidates should be using these tools in the hiring process? If they were doing a remote interview and sitting there typing stuff into ChatGPT, would that be too much? Is there any limit?

RONALD: It's a good question, and it's hard to draw that line. I would almost say there isn't with the caveat of if you use it to get the job that you don't deserve in a way because you've just cheated too much, let's say. Oh, you leveraged it too much, let's say, then I think you're cheating yourself out of an opportunity because eventually you will be found out, and you probably won't make it past probation. However, in the interview setting, if someone were to answer your question about the top three things that you want them to do to help them onboard faster and help them onboard better if they used a language model to give me the answer, I think it's about probing; it's about that sincerity or that fidelity, if you will, of what you said, and does it actually reflect your reality, and asking why, and why is it that meeting the broader team is really important to you on your first day or whatever it is that they've asked for?

TIM: Yeah, the sincerity thing is interesting. I am reminded of another person who had used the tool to develop a strategy for that part of the business, let's say product strategy or sales strategy, but they presented it as if they'd created it themselves. Oh, I spent the week in creating this product strategy or whatever, but it was clearly just one minute prompt into GPT and an export. Not saying that the end result wasn't of some value, but I felt Oh, that's a bit disingenuous because you're presenting some work that you, it's like plagiarism in a way. It's almost as if you've gotten your friend to do it. What's the difference, in a sense? It felt uncomfortable for me seeing that. Do you feel like at least candidates should be forthcoming in how and why they've used the products?

RONALD: It would be nice if they did, but even if they didn't, you can tell. You can tell because I had to do something similar. Not that I cheated, but I was writing the data strategy for my company, and I thought, Let's see what GPT has to say about this. It was a solid six out of ten, but it wasn't close. It wasn't specific enough; it didn't take into account a lot of the context in which I'm working, and it feels bad. It feels generic; it feels heartless in a way or soulless, so you can tell when people do that, and it just makes you look a bit silly; it makes you look like your answer is half-baked. and what's worse is you're going to probe this candidate on why did you do that, why did you say that, and if they can't give you a good response, then it's just a very strong sign.

TIM: particularly if the thing you're asking them about involves any level of their view, their opinion, like I'm asking you, this is your piece of analysis. What's your business recommendation? Are we going to buy this or not? Are we going to move into this market or not? It's got to be your view. You've got to own that on a certain level, and so yes, that must be incredibly apparent if they haven't created themselves in those probing questions as you say.

RONALD: Exactly, and I guess on the other hand, if it's to create, if you give them some data tables and you're asking them, Hey, write me a SQL script to extract this output, I'm not particularly fussed as to how it's done, provided they understand it and provided that if I asked them to modify it, they're able to do it, but how to get there initially, I'm not fussed.

TIM: And then maybe that's another key, actually, so you, as a data expert, use GPT to create a first pass at a data strategy, which saved you 50 hours thinking of it from scratch, and then you could go and take it from a six to a 10, let's say, but someone with no experience in data at all would have bunged it into GPT, gotten the six out of 10 or five out of 10, and got Oh, this looks pretty good because they're not in a position to know. Otherwise, maybe that's critical to them; it's a tool that people who are already experts can leverage to make their lives easier, basically.

RONALD: Definitely, and we always use GPT where I work at Omico as a first pass. It's a great tool to create a draft, and we must always remember it's a draft. It's always a five or six out of ten draft.

TIM: Yeah, that is the scary part, and I think part of it is that its grammar is flawless, so if you're a kind of grammar cop like I am and you can spot your own typos and spelling from a mile away and you get something that just is on point with formatting, grammar, spelling, and everything absolutely perfect. Then it's even more deceptive in a way because it's got the presentation of something that feels authentic.

RONALD: Yeah, it writes well, but you give it five minutes of critical thinking, and then you realize Oh, actually this doesn't make any sense whatsoever.

TIM: Thinking a little bit more broadly now about hiring and the kinds of people you look for, what you evaluate, and what your view is on cultural fit. I feel like It's been a kind of buzzword, particularly in the HR circles, for the last 10 years. I have very mixed feelings about a cultural fit. How do you view it? Is it important to you? If so, how do you measure it?

RONALD: It's a good question, so first off, yes, it is important to me, but I think the definition of cultural fit here is to do with the values that the person has on how they approach their work, so as I said before, curiosity is really important, I find, or I think so, is the ownership and the level of responsibility the person is willing to take on their work, so I don't know if you'd call them cultural items or whether or not they call you by just your work ethic or your work values. But for me, what's most important when I want to talk to candidates is: Are they curious people? Do they take ownership and responsibility for what they do? And then are they willing to take feedback?

TIM: And how do you evaluate these? Are these things you're probing for in the interview? For example, for the feedback one, do you just dump some negative feedback on them and just see what happens?

RONALD: Sometimes, but not usually, you can ask around it, so curiosity you might ask, Give me an example of an approach you took on a problem or a project, and how am I? What are some other ways you could have done it in retrospect? and see if they can think of any responsibility and, particularly, communication. they're harder Some of them are quite hard to tell during an interview, and the reality is measuring success on these things is tough because you don't know until probably three to six months in whether or not you've made the right hiring decision. I always say that when I hire candidates, sometimes two or three rounds of interviews are needed, and it's a 50/50 Sometimes you make great decisions. Sometimes you're like, Oh my God, how did I not pick that up? and you miss it.

TIM: I'm laughing because we've all been there, but I'm also a big fan of football, and I heard an interview on the weekend with David Moyes, who used to manage Man United and Everton and West Ham—all these big clubs. He's managed a thousand Premier League games or something ridiculous, and they're asking him about some of the players he'd signed and were there any that he regretted and how quickly did he know. and he said for some of them it was literally their first training session where he goes, Oh my God, what have I done? and they've spent 10 million pounds, so if that level of investment, you can still, it's still a flip of a coin at some level, then yeah, I think we can all be excused for not making the perfect hiring decision every time for our hundred K year data rolls.

RONALD: Yeah, and the reality is you have, what, two, three hours at most to make that decision, and everyone's in a process where both parties are new to each other; everyone's probably a bit guarded, and everyone's trying to put their best foot forward, so it makes sense.

TIM: And in that process, are you trying to explicitly measure cultural fits? So you mentioned three of those values. Are they things candidates would then be scored against on a rubric? How do you actually approach deciding who has or has not met those criteria?

RONALD: Traditionally, when I have interviews with candidates, I have two rounds of interviews. The first round is going to be technical; it's around can you do the job, and the second round interview is solely dedicated to will you do the job? Do you have fit? I'm not just looking at yes, obviously the three components that I met and I talked about curiosity, ownership, and willingness to take feedback, but also then it's how do I think this person will get along with the rest of the team as well? Again, having that, because I don't work in a large organization, everyone in my team needs to be able to work effectively together. I've made mistakes in the past where I've hired that super competent, super technically amazing, brilliant genius type of person, and the first week into it, I've just regretted it massively. I was just, Oh my God, this person can't work with others; they just can't communicate well; they think their way is The best way and the only way to do things and it's a horrible situation to be in

TIM: Yeah, I'm laughing again because I've had the identical experience. I'm sure we all have. I used to work at a business that was, I'd say, dominated by curious geniuses who were very challenging to work with. They added incredible value in a certain scenario, but I feel like almost it was once the company was at a certain stage in terms of its scale that it just didn't work anymore. Because you couldn't over-rely on one person just running as a lone wolf

RONALD: Yeah, and it doesn't require a big company to hit that stage. The moment you hit three, four, or five people, you're already there in a team. oh

TIM: Thinking back to some of the candidates you've interviewed who maybe haven't made it, the ones you didn't hire, are there any recurring themes among that set of candidates? Any kind of reasons why they would typically not be hired? And just that connects to any misconceptions those types of candidates might have about the process. For example, the ones who've just focused exclusively on technical skills, they've just done their hundred hours of Udemy, and they've neglected the other bits of the business. Yeah, can you think of any typical types of characters you've come across?

RONALD: There are a few, so yes, people are over-relying on technical skills. The other group that I tend to see a bit is people who are very hand-wavy about what they do in their role, and they can't be specific about their contribution. I worked on a big project together, and it was a hundred million dollar project. and yeah, it was great outcomes, and what did you do exactly? Oh, I did a bit of analysis, project management, and I dealt with the client. Okay, so what did you achieve as a result of that? Oh, the project was great; the client was super happy, and okay, thanks. Then you haven't really told me much; that's one that immediately raises hairs on my back, and that's going to be a no.

TIM: That's just to pick up on that kind of type. Let's call him a type for a second that sounds almost like you're talking points politician high-level positioning. I wonder if, in candidates, then I just thought of this right now: should they almost be thinking of different audiences in their interviews? because they're going to have a very different experience interviewing with you versus interviewing with a typical recruiter or a typical HR person. In all likelihood, you're probably going to get into more details on the more technical side of things; you're going to really interrogate what they've done. whereas maybe on average someone in talent acquisition is looking for slightly higher-level things; they don't want to get into the details of a model, and so maybe they've almost optimized their process for that kind of 15-minute quick chat, whereas they need to also have an in-depth ability to back that up. What do you reckon?

RONALD: That's true; I hadn't considered it that way before, but that does make a lot of sense. I would say also to candidates who perhaps are more junior or have worked on big projects and maybe they haven't contributed all that much is to also just be upfront to say, Yeah, I worked on a big project; my role was just to, I don't know, do something very small on the side. And that's okay because I did that role well, and I took responsibility for that piece. That's the thing that, at the end of the day, I think a lot of people are looking for when they're hiring: that ability to learn and to engage because the likelihood is candidates have never done what they've been asked to do before in a new role.

TIM: And even if they had a very small piece of a very big project, presumably then still being able to quantify something, some demonstration that at least for their unit they delivered some kind of value, whatever that metric was, rather than talking in vagaries, would help their cause.

RONALD: Definitely 100%.

TIM: So apart from that particular type, someone is maybe just operating at a superficial level; it doesn't give enough details. I wouldn't say inauthentic; maybe it's a bit harsh, but I can't quite get into the nuts and bolts of what they've contributed.

RONALD: It just makes it sound like they're lying a bit because they can't tell me what they've done; they just touch this very broad project, the other type. I would say that I have come across people who fail a second-round interview, so those are people who are just far too sure of themselves, far too almost morally superior, if you will, that they don't make mistakes, or they took the best method, and they know they did the best thing ever, but they made the best decision. And there's always a question How do you know you've made the best decision? How is that even possible? So they're the people that I Yeah, they typically run into the two heads are smarter than one type issue that I spoke about before.

TIM: And do you sense that it's a lack of humility on their part, or do they feel like this is the best strategy to get the job? Or, yeah, where do you think the gap is?

RONALD: So the gap for me is it would be a tell for they don't—they lack retrospection; they don't retrospectively think about what they've done or how they've done it. They just maybe they're very good at forward-looking, but they also aren't very good at backward-looking. Maybe they're doing it to impress, and maybe it would work on some people, but unfortunately for them, I don't get impressed by it.

TIM: It's probably also a market-based thing as well, isn't it? Like, I can imagine it's probably drawing stereotypes now, but a sort of American style of culture, business culture, is a bit more in your face, a little bit more aggressive, a little bit more salesy. Maybe that would dominate more than a slightly humbler tech startup in Australia. Maybe there's just a different expectation of how they would behave as well, perhaps.

RONALD: Yeah, I think so. It's hard doing interviews. We've all been there. Not only are you trying to answer the question while you're also trying to establish that connection with your interviewer, so being able to be able to clue in and go Hey, maybe Ron doesn't actually like people who are too boastful, and people who like the people who are a bit more humble and can retrospect a bit would be important. but I take your point; it is difficult.

TIM: You'd think if they'd done agile methodology that would be baked in. Isn't it like Sprint's Anomaly, ending with a retrospective meeting every two weeks, thinking about what you could have done better? Could you do better? A lot of tech companies do meditation. It's all about reflection. You would hope that, yeah, if they could connect that to their interview performance, then they could learn some lessons, perhaps.

RONALD: Yeah, I agree with you, Tim; we're not asking for that much, are we?

TIM: Speaking of not asking for too much, what about hiring people with no domain knowledge at all of your industry?

RONALD: The most important thing is curiosity and willingness to learn, so I have a great story around this, which is one of my colleagues that I used to work with, and she's still, she has since gone to another company, but I hold her in very high regard, and she works in health rural evidence looking at she does all sorts of complex working with health data. She started her journey looking at metals and alloys and figuring out how to best make them. I don't even know what she did; actually, she did a PhD in mythology, and if she can switch careers and jump into health data and be highly regarded, I think a lot of people can.

TIM: Yeah, for sure, and at the end of the day, data's data. It's really important. Every business has a conversion funnel; every business has a warehouse, and you need the same technical skills, and the curiosity applies equally no matter what the domain. I feel like domain knowledge is like there's a pro and a con to it. If you hire someone from your industry, they've been in the industry for their entire careers; all they know is that industry. Presumably, the learning curve is zero, so that's a massive benefit, but maybe myopic; you could argue especially if it's like a more traditional industry. whereas from outside you could add a fresh perspective perhaps

RONALD: It can certainly be dogmatic to the domain knowledge of your last role, which would be a problem. I think where the benefit is awareness of how that ecosystem that you're working in operates, so to give a really solid example I'm working in health data; it's really nice when candidates come and they already know how hospitals and general practice and specialists and how that health system all integrates together. because they can look at the data we have and draw insights from it that I probably didn't know before. I'm like Oh, that's a really good point. I didn't think of that, but yes, being very dogmatic around this is how we should analyze the data and this is what the data is and how we should interpret it. That is problematic.

TIM: I wonder if thinking about it now, we're going to get more people working in analytics who are domain experts and can now leverage a tool like ChatGPT to do the technical stuff they never knew how to do or it was too difficult based on their skillset, and so now we're going to get people who know the ins and outs of their business or their product or their function so well, and now they're going to get this cool tool that helps them do the stuff they couldn't do before.

RONALD: I think we'll move there eventually. I don't know when, but that concept of no-code programming has been on the horizon for quite some time, but slowly we're shifting there, so yeah, maybe one day analytics will be just someone who's really good at asking questions around the data rather than actually manipulating tables and spreadsheets or whatnot.

TIM: I spoke to someone this morning, actually, who's doing a little interesting project in their company. Their goal over the next month or so is to see if they can make their data analytics function obsolete by using GPT, so they're going to try all these different tools that claim to be GPT on top of your warehouse or dashboarding or whatever. he'd used some already thought Oh my God, these are 10 years away, so this is maybe not going to work, but they're going to press on, and the idea is to have you almost like that self-service thing for ad hoc queries. You can just do the large level, the natural language, to get your data back. I feel like this is the sort of thing that has disaster written all over it at a certain level, but

RONALD: It sounds good, doesn't it?

It sounds wonderful; my question is who's around to tell you that your interpretation is wrong or the output's rubbish?

TIM: I can think of times I've worked for years as an analytics professional, and the difference between a left join and an inner join in a hundred lines of SQL code can be the difference between it being completely wrong or perfectly right and off by orders of magnitude and something as subtle as that you could never possibly know unless you knew exactly how the tables connected to each other. So the number of different possible areas you could make is almost infinite, but the devil's advocate would be maybe a business where they have one analyst serving a hundred people, then at the moment a lot of people will be getting no answers at all or not even trying to do a data-driven way. That's just gut feeling the whole way, so maybe it's better to have 10 answers, eight of them are bullshit, and two of them are right, and that's better than it was before. I don't know.

RONALD: Look, if that's your situation, then I think the organization has broader challenges than not enough analysts; it has a data culture or a data strategy issue.

TIM: Can you think in your past of any particularly tough hiring decision you had to make, and if so, what factors did you consider when you were making that tough hiring decision?

RONALD: hiring decisions, so we're talking about being unsure of the candidate or

TIM: This could be someone who you've hired and regretted; this could be someone who you haven't hired and regretted; this could be someone who maybe you had two amazing candidates, and you could only hire one of them any time that you were caught at a bit of a crossroads, maybe, and had to make a tough call.

RONALD: Yeah, look, the biggest regret I've made is I think I said it before: hiring those super smart, technically super smart, and capable people who I think, Oh great, if they could come in, they could implement all this amazing stuff that they've done before, and some of them I'll look up, and they've done image pattern recognition and images, and they've been involved in all sorts of sophisticated research programs. And it's true they're technically very gifted, as the main regret is they fail to grasp the core concepts of the business, and that's a challenge. Actually, the inverse is also true for me, where I've hired people who are tech culturally, if you will, a very good fit but technically very poor, and we've been able to actually train those people, and it takes time to train them up technically, but because they had a good understanding of the business and the willingness to learn and they had the right aptitude to fit into the business, that actually ended up pretty well. It actually worked out fantastically.

TIM: And presumably for those people, they must've had the mindset and realized they needed to learn some technical skills and then wanted to do it as well.

RONALD: and the humility to accept that Oh, I don't know much, but I'm willing to listen, and I'm not going to assume I know the answer.

TIM: Yeah, so that almost growth mindset I guess you call it these days, and was that something you knew during the hiring process, or were you pleasantly surprised by that once they joined and saw Oh, actually they're really going for it.

RONALD: It was a gamble, so for that particular candidate I'm thinking of, it was a bit of a gamble that had that sort of come from an adjacent domain that was they would be able to learn that, but the business or the technical concepts fast enough, but they'd never done it; they'd never been an analyst before. I was actually hired from a consultant role. I'd never been a consultant before, but I could sense that they were, I wouldn't say desperate, but they were really willing to give it a go. I don't know anything about this field, but I'm willing to learn and willing to give it my best shot. I'm willing to apply myself to do that. and it really worked; it really came through. It was one of the best hires we had.

TIM: What about thinking now in terms of diversity? Have you seen any interesting or innovative ways any companies have gone about improving the diversity of how they source talent?

RONALD: I think since COVID, my general feeling is that it's improved. It's improved because people are more mindful when they look for talent to be that, that almost talent-centered, if you will, because we know that we can work from home now, and it is a method that works, but then also they're more willing to accept that everyone has their own life, their own circumstance. So I think in more broad terms it's improved. I don't know about specifics; nothing comes to my mind immediately when you say that.

TIM: The one you've touched on I feel is potentially the most profound one. If you can hire people remotely, then suddenly the potential candidate pool is astronomically larger, so in an example, sorry, in our business we basically are anyone from anywhere in the world, so then if you think of people's diversity across different geographies, then that's a pretty easy way to go. Maybe not easy, but a kind of big bang for your buck there, so the fact that you're you would have remote hiring, I think, does make a big improvement, especially for people that might not live in the middle of the city or live near the office. I remember seeing some research around also people with physical disabilities, and the improvement in their odds of getting roles during COVID during a remote interview process was astronomical. So I feel like it opened up some opportunities for people who were previously quite disadvantaged.

RONALD: Yeah, and I don't know about you, Tim, but for me, we have remote workers, and it's fantastic, but it does present additional management opportunities and challenges because it becomes more of a question of, okay, if people aren't in the office together sharing that same physical space, how do we promote that team spirit or team culture? How do we effectively update each other? Still, video conference calls aren't as authentic, if you will, as in-person meetings, so it's still a bit harder, so I think that their management challenges that, and maybe we accept that there's going to be a five or maybe 10 percent drop-off in efficiency because everyone's remote, but we gain that through diversity, and we gain that through other ways. So I don't know; that's a tough one in many ways.

TIM: Yeah, I agree there's nothing like a meeting in real life in three dimensions. Having recently met some of our team who I hadn't seen since they started working for us because they live on other continents and then seeing them in IRL and giving them a hug was a pretty nice moment, so being able to do that on a regular basis would be better, but then, yeah, it's definitely a trade-off because if you can hire remotely, then the size of your talent pool is a thousand times bigger than if you can only hire within 20 kilometers of your office. So it can be such a game changer.

RONALD: Yeah, and I actually think so. I have a rule where it's fine if you work remotely or if you live remotely; even at least once a quarter we meet in person as a team, and we sit next to each other and have those water cooler conversations, and we have a ritual where we go out for lunch, so I work at UNSW. So along Antarctica, there are tons of nice Asian restaurants, so we usually go to one of them. I'm always pleasantly surprised at just the spontaneous, curious questions that come up, and they're fantastic ideas, like why did we not talk about this sooner? And they just come up over lunch. I think that speaks to the power of having people in the same space, which is harder with remote staff, but there are ways around it.

TIM: Yeah, and I feel like with meeting people in real life, there's almost like a diminishing return, and if you meet people once a quarter in an in-depth way, you have a night out, you have a long lunch, something like that can be almost as valuable as seeing them. I don't know three times a week for three months. It's because it's more about the quality rather than the quantity I found.

RONALD: Yeah, no, I'd second that, and it's deliberately putting time away, much like doing retrospection, deliberately putting time away to have those what are called the conversations.

TIM: Yeah, absolutely. My final question for you, Ron, would be—and this is a big question, dumping a bit of a strange one on you—is there anyone you consider to be a hiring hero to you, or anyone that you've seen do hiring very well in the past? This could be a person or it could be a company, and if there's no one in particular that springs to mind, what would this hiring hero mythical character look like, do you think?

RONALD: I'm trying to think if I have a hiring hero. I don't have any single organization or person in mind, but if I were to amalgamate all of them into one person, I would say it's something like hiring them so that in reality, when you hire someone, you're going to have to work with them really closely, so I'd want to be comfortable if I'm hiring someone that I would be able to go to the pub with them afterwards, and maybe one day I can report to them. If I'm comfortable with these two things, then I think it's going to be okay.

TIM: awesome