Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
0012 - Challenges for Heads of Data in Scale-ups!
243 Plays1 year ago
๐ New episode on The Stacked Podcast featuring an incredible guest, Joao, the Head of Data at Plum. We dive deep into the Challenges of Building a Data Team in a Start-up. ๐๏ธโจ
๐ Introduction: In this episode, we take a journey with Joao, exploring key milestones in his career that led him to leadership in the dynamic world of data.
๐ Building from the Ground Up: When stepping into a scale-up as the Head of Data. Where do you begin? Joao shares his insights into the crucial first steps and priorities on the to-do list.
๐ป Tech Talk: Tech is the backbone of any data team. Joao discusses the overwhelming choices data leaders face when building their tech stack and how technology influences the hiring process.
๐ค Hiring Hacks:Hiring isn't just about filling roles; it's about building a successful team. Joao sheds light on how to decide which roles to hire, the crucial skill sets for an early data team, and how roles evolve as the function matures.
๐ Cultivating a Data Culture: Tech and team are essential, but how is it all perceived by the business? Joao shares his insights on growing a data culture in a start-up, what defines a good data culture, and common pitfalls to avoid.
๐ Measuring Impact:How do you know if your team is making a real impact? Joao discusses the metrics of success, showcasing value to secure more resources and bigger opportunities.
๐ก Facing Challenges:Joao opens up about the common challenge of being under-resourced as a Data Leader in a rapidly growing organisation and how to tackle it head-on.
๐ Beyond the Horizon:Explore with Joao the additional challenges Data Leaders in start-ups often face, gaining valuable insights into navigating the dynamic landscape.
Here is the link Joao spoke about - https://towardsdatascience.com/how-i-won-singapores-gpt-4-prompt-engineering-competition-34c195a93d41
Tune in now ๐ง
And give us a follow and a rating, your feedback means the world to me!
โจ Don't miss out โ it's a knowledge-packed episode you won't want to skip! ๐๐
#DataLeadership #Startups #PodcastLaunch #TechTalks
Recommended
Transcript
Introduction to Stacked Podcast
00:00:02
Speaker
Hello and welcome to the Stacked podcast brought to you by Cognify, the recruitment partner for modern data teams hosted by me, Harry Golub. Stacked with incredible content from the most influential and successful data teams, interviewing industry experts who share their invaluable journeys, groundbreaking projects, and most importantly, their key learnings. So get ready to join us as we uncover the dynamic world of modern data.
00:00:35
Speaker
Hello, everyone. Welcome to the podcast.
Guest Introduction: Jaul from Plum
00:00:37
Speaker
Today, I'm joined by Jaul. Jaul's the Head of Data at Plum, a fintech within London. Jaul, it's great to have you on the podcast. Thank you so much, Harry. It's really good to be here and honored by the invite. Looking forward to share some of my views and experiences.
Challenges in Building a Data Team in a Scale-Up
00:00:53
Speaker
Great. Well, today we're going to dive into the challenges of building a data team in a scale up. And let's start at the beginning. You're relatively new in your new role as the head of data at Plum. So where do you start building a team and where do you start when you first join a startup?
00:01:10
Speaker
Yeah, so you're right. So I joined Plum relatively recently, around six months ago now. Good time flies in August, 2023. Just emphasizing a bit more what Plum is all about. So we are a FinTech in London, as you said, Harry, focusing on helping people throughout their financial journey by giving them the tools for them to manage, save, and invest their money. What set this apart is probably like the aspect of focusing on automation and enabling effortless experiences around money.
00:01:37
Speaker
with things like automatic saving rules and recurring investments. We are on a mission of maximizing wealth for all. So you'll see us throughout the years and probably seen already expanding to different directions of wealth management, savings, and financial planning.
Jaul's Career Journey
00:01:52
Speaker
So it's a super interesting place to be in terms of data because we sit right into that intersection of how people manage their money and it's super heavy on behavioral challenges.
00:02:03
Speaker
How did you get to be where you are then? Maybe we should start there. It would be good to get a bit more of a background on your own career before you joined. Previous to Plum, I was leading the data function at Free Trade. It's another thing tech in London focuses more specifically on democratizing access to financial markets by building tools for people to invest for the long run.
00:02:25
Speaker
It was an interesting journey there, which I'm really proud of. I joined as the first data scientist there, joined a team of two at the time.
00:02:34
Speaker
And then we scaled the team to 10, went through a bumpy 2022, like many scaleups around the world, and then started transitioning more to heading the function, managing, and leadership. Prior to that, mixed bag, but emphasizing, obviously, scaleups and startups.
Value of Scale-Up Experience
00:02:51
Speaker
I was a senior data scientist at Bulb, leading all things related to data science on the marketing and growth space, and also worked for different companies like Forward 3D, an agency focused on digital marketing.
00:03:04
Speaker
page SPC and longer. So Lum is probably the third big scale up in a row. And the London kind of B2C tech space. Yeah, in terms of technical background, throughout my career, even though I'm now more focused on leadership and management, my background is all in data science and advanced analytics. So basically helping
00:03:25
Speaker
companies and teams figuring out insights, patterns, intelligence from their vast datasets focused on different areas.
Broad Knowledge in Data Leadership Roles
00:03:33
Speaker
Because I was privileged and I'll talk about why I think it's so good and important for someone to at least experience once a scale up or a startup, I think it shapes you in a unique way in terms of learnings and experiences.
00:03:48
Speaker
But I personally evolved to becoming quite a bit of a full-stack data scientist over my career operating across kind of different areas of data, data collection, ingestion, warehousing, machine learning, and A-B testing mainly, which I couldn't be more grateful for because it formed the foundations to be able to hopefully in an okay way lead data teams now.
00:04:11
Speaker
Yeah, I think that breadth of knowledge is essential to being a data leader. You have to have your specialism, but particularly obviously in the startup and scale-up space, you're always wearing many hats and you're looking after the end-to-end process. So having a good understanding of how it all fits in together, I think is really key. And the more the data, modern data stack and
00:04:36
Speaker
processes grow, there's a bigger convergence of needing to know how all of these systems communicate with each other. Absolutely. And just to add, look, startups and scallops are not necessarily always easy, but definitely in terms of value for personal growth.
00:04:53
Speaker
tends to be a super rich experience because it's environments that tend to be relatively unstructured, relatively undefined, the roles, different spectrums, obviously, but for the right individual that is trying to optimize for learning, personal growth, growing responsibilities is the perfect environment for learning.
00:05:11
Speaker
Perfect. I definitely agree. So you joined Plum six months ago as the Head of Data. What's first on your to-do list as a Head of Data in a new team? How do you approach that?
Aligning Data Teams with Business Goals
00:05:23
Speaker
Because it can be quite overwhelming.
00:05:25
Speaker
Absolutely. Very interesting, nevertheless. Look, I think it's all about trying to go as deep as possible into this triangle, these three dimensions, right? Business, team, and tech. To try to grasp the context as best as possible and as effectively as possible. On the business side, perhaps the most important thing in the very, very beginning is to understand what drove and influenced the business to invest in the senior data role.
00:05:51
Speaker
And again, there's multiple options here and multiple scenarios. The ones I experienced the most over the last years or heard other leaders talking about the most is either someone left the team and they want to continue in the same trajectory. So it's about maintaining a good trajectory that the team was before, but they need to fill a gap on leadership because the person left.
00:06:13
Speaker
That will definitely shape your role in terms of being more oriented towards extending what the person in your role was doing before. The other option is maybe the business wants to get more from data, more automation, more insights, more value, more intelligence.
00:06:30
Speaker
So they're looking to hire someone for a leadership rule, expecting some sort of stepwise change, which is the position I feel I was personally in as plum. And another examples, right? Particularly now in the era of AI, you also see a lot of hype type hiring. So business senior stakeholders, founders wanting to jump as fast as possible into the AI train.
00:06:52
Speaker
And thinking that the only thing preventing them from doing that is having the right data leader running the team. So I think it's super important to grasp that business context. And understanding if the ambitions from the leadership team is what do they expect from that data, fundamentally from the data team, right? And again, without going into specifics of what teams are here, I experienced really different range of kind of expectations towards the data org.
00:07:19
Speaker
ranging from being a more operational player in the business. And you tend to see this quite a lot with highly regulated businesses, where first and foremost, even though you can get amazing at the benefits from investing in data, first and foremost, data is not optional because you have regulatory reporting responsibilities, auditing responsibilities, etc.
00:07:40
Speaker
And then that's suddenly one end of the spectrum. The other end of the spectrum is obviously believing that data can be an independent value stream for the company, delivering revenue, cost reductions, et cetera.
00:07:51
Speaker
I'm very proud and happy to feel that I chose a company to join like Plun where definitely we are all aligned in terms of like what we want to do from data and expand the value streams we take from data. So, yeah, I think focusing on understanding that business picture is quite important. But then moving to the team and org, right?
Managing Inherited Data Teams
00:08:13
Speaker
Are you inheriting a team that was built before you? So you didn't have a choice in terms of like putting your own opinion into hiring or not? Or is your role about scaling up or building a team? I was personally very happy. I inherited an existing team as Plum. Very lucky, sorry, that everyone is excellent and really good to work with.
00:08:34
Speaker
But it's fundamental to take as much time as needed to truly get to know the team. It sounds really, really obvious, but it's something that can easily be missed and rushed. And they're saying, especially if the team is already of a certain size, like seven, eight, onward, it's easy to join, start talking with senior stakeholders, start getting excited about like loads of different projects that can be done, but missing quite a fundamental part that you need, you need to be aligned with your team and understand them. Understand?
00:09:04
Speaker
what makes them love their job and staying around, but also what emotionally drains them, what problems have they tried to fix before and failed, what quests they started but didn't go to complete previous successes, et cetera. Nice. I love that. The business context always takes a long time to grasp. It's so important for you to then be able to
00:09:28
Speaker
Before you start implementing your roadmap and your plan, you need to know what you're looking to achieve. Your team, you won't be able to do anything without your team. Hiring can be a long process and you don't want to lose them if you can keep them. I really think that's great understanding how they think and how you're going to best empower them.
00:09:50
Speaker
And just to add quickly, which I think in the team domain is perhaps the most important thing is to really learn about
00:09:59
Speaker
your team's individual characters, right? Like drivers, motivations, streams, interests. Because scale-ups are environments where very often it's required for people to leave their comfort zones and go above and beyond and be self-starters. So it's critical to understand, like, really what drives them, how they think and how they work in order to make sure that they can succeed in environments like this.
00:10:25
Speaker
That's so important. That's a key role of a data leader, isn't it, to make sure that your team are shielded from the aspects which they don't need to concern themselves with so they can get on with their best work. I think empowerment is the other key point. I suppose you mentioned in your first statement, you've got the technology, the team, and the business as the core
Developing Technology Stack for Data Teams
00:10:48
Speaker
areas. Maybe let's start at the
00:10:50
Speaker
the technology. Technology is a huge part of how a data team operates. The modern data stack has been growing and growing. The new tool coming out every month with all of these vendors popping up. How do you make sense of it all and how do you approach building a technology stack which is fit for purpose for you and your team?
00:11:11
Speaker
Yeah, definitely. Look, I'm gonna answer mainly in two blocks. One more focus and kind of like what to focus in terms of tech once you join a company and then zooming a bit into the tech vendors if that's okay. Like, in terms of like what to do,
00:11:26
Speaker
With regards to tech, when you join a company as a data leader, I think understanding the broad technical landscape of the company is absolutely crucial. For example, at Plum, there's understanding how the systems are built. At Plum, for example, the back end is built using Python.
00:11:41
Speaker
which is interesting. It means certain dynamics and synergies between data folk and professionals that work already natively in Python, and backend engineers that are building products and features for users can sometimes be facilitated. So those aspects really change how we think about what to invest on in tech. Understanding how data flows in the business and what data is available. Again, in the example of Plum,
00:12:07
Speaker
It's a company that have hundreds of different data sources internally and external. So it's really interesting in the beginning to immediately understand that. So it gives you a sense of the opportunity and the sort of quests and projects you can land on that data. But one of the things I did when I joined, which I think was really useful, was to map the technical, the data technical domains and going deep into describing like the state of these domains and forming at least a two quarter roadmap for each.
00:12:37
Speaker
For us at Plum, but I believe for many companies out there, the technical domains of data were data collection and ingestion, data warehousing, business intelligence, and MLOps. So from a data perspective, these are the things I focused a lot when I joined a new company, going deep into the different domains, how the domains relate to each other and form an opinion and roadmap for the next six months.
00:13:01
Speaker
going to tech vendors and more directly answering your question.
Problem-Solving and Vendor Management
00:13:06
Speaker
Look, to be very honest and perhaps upsetting the
00:13:11
Speaker
like loads of different good people I've met over the years from vendors. I tend to be quite cold about tech vendors and focus first and foremost, almost exclusively on the problem statement I'm experiencing with my team in the moment. The constant approaches from vendors popping up.
00:13:32
Speaker
makes it very easy sometimes to do the other way around, right? Like trying to map a problem that is not really a problem to a new shiny vendor that he's speaking with. So this is something that I definitely try to do the best I can is to focus on our own problems and pay attention to the vendors that map the problems we are experiencing and not the vendors that we don't think are relevant or mapped to any problem we have.
00:13:58
Speaker
even if they could be useful, they're not something you should prioritize. The second thing is... Sorry, Harry. No, I just said I liked it. Obviously, you're focusing on yourself and your own problems. It's easy to get caught up. They have very big marketing budgets and they have persistent sales teams. Great, but it can easily get wrapped up and lost in all the excitement.
00:14:24
Speaker
Yeah, absolutely. And look, perhaps the main thing is like having a bias tickling scale ups and startups and tickly after 2022, 2023 macroeconomic situation. If there's one thing that anyone in startups learned is that cost and path to possibility matter a lot.
00:14:41
Speaker
So having a bias towards being, you know, towards solving the problem simple first and internally, it's really, really important. The way I think about this is like solving a problem statement you have in-house, it's not like a binary outcome of like it's solved or it's not solved because there's different spec, it's a spectrum of like how well you should solve the problem. And it's all about like,
00:15:05
Speaker
particularly in startups, is all about solving it well enough. And for that, loads of times a vendor might not be necessary. I have one example that I experienced recently around, for example, data catalogs. In order to solve the problem perfectly, you probably need to map it to a vendor that maps all your lineage and extracts all the documentation and descriptions of your DBT fields and Looker fields.
00:15:33
Speaker
There are tools out there that allow us to create a data catalog that your business stakeholders can use, or at least you can validate if your business stakeholders are interested in using a tool like that or not. An example is the Looker dictionary add-on, which we've been playing with and seems quite useful.
00:15:51
Speaker
Nice. So look, the tech is obviously one part and plays a big part of facilitating your role in data, but the team are the people that actually use that. So whenever you join a scale up, you mentioned obviously you might inherit this team, but
00:16:09
Speaker
kind of in the name of a scale up, growth is a big part of it. So hiring is really important and hiring the right team is hard and knowing what to hire can be hard. So how do you decide what roles to hire and when's the right time?
Economic Changes and Hiring Needs
00:16:25
Speaker
I suppose really relating that to the last year that we've had in the economic situation.
00:16:31
Speaker
You probably want that role, but knowing when the right time really to hire it is something I know a lot of leaders struggle with. So yeah, what's your advice on that?
00:16:40
Speaker
Yeah, it's a great question. One that certainly doesn't have a right or wrong answer. What certainly is the case is that there was a paradigm shift during 2022 where before that we were in a world macroeconomic environment with very low rates. Capital was relatively cheap. Money was flowing from VCs into startups really well and easily. And all of a sudden, there was a sense of that the music stopped. And
00:17:08
Speaker
It forced every team to think differently about scaling and moving a bit away from the growth at any cost kind of mantra towards like profitability and sustainability of that growth. And you see these across the spectrum, not only in hiring, we spoke about also new shiny vendors and paid partners to add to your data platform, but you also see these on the type of work that the data teams are asked to do and the projects that are prioritized.
00:17:35
Speaker
insights typically tend to look more into areas like profitability, return of the customers, LTV, et cetera, rather than just volumes of growth, et cetera. So it's quite interesting. So I guess after that change, and we'll see what 2024 brings us, but definitely we're still in a period of caution and kind of balance when it comes to hiring. From my experience, particularly over the last two years, it's really
00:18:04
Speaker
about two things. One is the feedback and observations internally regarding how under capacity the teams you're managing are. So that's the type of hiring that it's more like a need rather than to be able to do what we want to do in the present rather than a scaling or growing type of hiring. And that's most of the hiring that has been happening in the teams I worked on over the last two years. The other type of hiring is more like
00:18:32
Speaker
looking into the plans going forward and the new bets. An example of these areas, again, obviously AI, quite a new field requiring part new skills, but certainly new ways of thinking and approaching problems, which might justify directional hiring.
00:18:49
Speaker
We at Plum, for example, are ramping up our AI and ML initiatives since Q4 2023, which I'm personally very excited about. We are not yet accelerating massively the hiring in that area before we prove and land a couple of important projects, but we're certainly thinking about how to move talent internally towards that direction, how to educate ourselves.
00:19:12
Speaker
how to decide what roles to hire and when, I think is mostly a function of feedback and observations about how under-resourced your team is and what you want to do in the near future.
00:19:23
Speaker
I like what you mentioned about how you spin up projects and prove some value or trying to prove some value before getting the green light.
Evolving Skills in Startup Data Teams
00:19:32
Speaker
I think that makes a lot of data leaders, you have to go and fight for approval and fight for budget to get them new roles. And if you haven't got anything to show for it, other than what you have planned in your head, it's much harder to do that. So proving some value cases and then showing the executives a roadmap of
00:19:50
Speaker
why this higher will drive value I think is definitely a great strategy to unlock that budget.
00:19:57
Speaker
So I think this would be one that many of the listeners will be keen to understand, especially if they haven't worked in a startup. What skill sets do excel in a data team at its early stage? And I suppose on that, as a data function in a startup turns into a scale up, how does the skill set evolve as well? So two kind of compounded questions there. Great questions. Thanks, Harry. Look, I think
00:20:24
Speaker
Individuals wanting to or aspiring to join a startup or scale up needs to be technically and culturally ready to wear many hats. I see this as a good thing, as a positive thing that can drive immense personal growth and career development. But it's certainly something that is not for everyone. Not all the roles are exactly defined.
00:20:49
Speaker
And even the roles that are defined, there's changes guaranteed in an environment like a scale-up. So your role will likely evolve. I think...
00:20:59
Speaker
This is particularly important in the very early days. If you end up being one of the first or the first hire in the data team, certainly you should be able or are expected to operate across the data stacks. So think about engaging and growing full stack kind of data skills, covering all parts of data, but also being ready mentally to take different responsibilities. From a personal skills perspective,
00:21:24
Speaker
I tend to quite explicitly have these almost kind of hidden scorecard. I keep one-night interview candidates over hiring manager interviews and try to look for signals that flash for personal traits. Self-starter, proactiveness, adaptability, and pragmatism. Starting with the end, pragmatism, it's really important to deliver the simple first before the complicated and having a really good eye for quick wins.
00:21:54
Speaker
This is perhaps, and other people in your podcast spoke about this, is perhaps one of the most impactful ways of influencing or impressing stakeholders in the startup is landing those things that have relatively low effort and relatively high impact. So your quick wins.
00:22:13
Speaker
Obviously adaptability, you need to be able to adapt to that change in a fast-paced environment. And lastly, as I mentioned, that productivity aspect of being able to spot opportunities to be better and passionately own those. I really like to see
00:22:31
Speaker
the enthusiasm of colleagues inside and outside of my team, teams like spotting an opportunity or an area where something can be better and like owning that and driving the change in that area without asking you for permission. It's something I don't miss from the corporate environments where you have to go through rest tape and layers of approval in terms of deviating a bit from your project and justifying that certainly in a startup.
00:22:57
Speaker
That's not only not a problem, but it's something valued and something incentivized. So, in terms of skill sets, I would go for those. Nice. Yeah, you have to, it comes back to, I suppose, your experience, that full stack, that breadth of knowledge and understanding how the data flows and then it's utilized is so important when you're going to be wearing many hats. I love the pragmatism.
00:23:22
Speaker
finding the simple wins, proving the value and identifying them can be hard. You know, you can have a long list of projects. How do you know which ones to tackle first? So being able to identify that is definitely a skill. So now you've got your tech, you've got your team.
00:23:41
Speaker
But how's that all then going to be perceived by the business? So this is more, I suppose, the cultural aspects.
Building a Strong Data Culture
00:23:48
Speaker
And most startup scale-ups tend to be more tech-focused anyway. So maybe have a better understanding of data than the more traditional business. But what is a good data culture if there is such a thing?
00:24:04
Speaker
is making sure, like, to me, a good data culture is making sure that decisions are as data-driven as possible. And I want to emphasize that as data-driven as possible, the statement itself accepts that there's decisions that
00:24:19
Speaker
cannot or should not be necessarily fully data driven. So being a passionate data professional, it doesn't invalidate this. And it's important to understand that there's always an aspect of certain groups of decisions that are driven by other things, but not necessarily by data for reasons like we can't get data on a certain intuition or certain gut feeling opportunity we have or for other reasons, right?
00:24:45
Speaker
So it's about making sure that the decisions that should be data driven are data driven.
00:24:51
Speaker
And in that camp, availability and accessibility of data in order to make decisions, it's everything. Again, the topic of self-service analytics, super interesting. There's different levels of skepticism and opinions across what self-service should be pursued. And more and more, I think, full and perfect self-service analytics are a bit utopian.
00:25:17
Speaker
but guarantee a level of accessibility and availability through a good business intelligence strategy using Looker or Tableau or the tools. It's absolutely crucial for a strong data culture.
00:25:31
Speaker
Brilliant. How do you improve this? Because that's a key part of how I suppose a data team interacts with the business and that's sort of how it's perceived. So as a leader, if there's areas where this is maybe more department specific, have you got any strategies or advice of how leaders can improve that? There's a few different ways. One I personally use is like,
00:25:58
Speaker
You know, as head of data, I'm leading a function that is kind of independent from other product squads or product teams. So I don't necessarily have a vested interest in kind of, basically, let me rephrase that part. Basically, I normally benefit from being like an independent participant in different discussions. So one way we try to evangelize data culture
00:26:23
Speaker
this particularly works on companies that are still relatively small, perhaps around below 200 people higher than that, it's not possible to scale this, is to make sure you have data professionals and data leadership sat in different product and leadership discussions and working as a challenger
00:26:42
Speaker
and basically interrogating the quality of the metrics, the quality of the statements, and how data-driven some of the decisions are, and trying to raise the bar in that area. That's something that definitely I try to do a lot here at Plum, and in different roles day in, day out. It's not so much a concrete action. It's more like a way of approaching your position
00:27:05
Speaker
in the company. When it comes to tech, if you assume, we typically, we like to think about Looker, but this applies to any business intelligence tool as like the window of the shop, right? Like it's what people, is the way people interact with data is what they see. But behind the scenes, there's a massive operation of data collection ingestion, hundreds of pipelines, hundreds of tables, DBT, all sorts of things.
00:27:31
Speaker
So treating Looker as a customer-facing product or your BI tool in general as a product, something that requires a roadmap, requires understanding your users, in this case, stakeholders, internal users, and has different iterations, training, et cetera, is absolutely crucial in order to keep a healthy data culture.
00:27:54
Speaker
It's not always very obvious in the short term how certain actions you take in Looker or putting together data catalogs or rolling out data training company-wise affect the data culture. In the short term, it's hard to measure those effects, but hopefully these things become more visible after you consistently do them for a curious time.
00:28:18
Speaker
Yeah, I think that's, that product mindset's obviously been quite, there's been quite a lot of talk about that in recent years and seeing data as a product. I think that's important for a data team to think about the user experience of what they're building, whether you're
00:28:34
Speaker
BI and you're building a dashboard or you're in the data platform team, who are the consumers of your product and what is going to be their experience. I think that's definitely a mindset shift that is slowly happening within the industry and is very important to enabling the person who's downstream from you to do their work. Because sometimes the best way for you to do something isn't necessarily what the end user wants. And I think that's often missed. So I love that.
00:29:03
Speaker
Absolutely. And just to add to that, acknowledging that there's no such thing as one or two user personas, even within, you know, end user personas, even within a small company with 200 people or so. There's multiple data stakeholders and people with different data appetites and data needs. So it's really easy also, just being the devil's advocate here now, to over-optimize to certain specific niche use cases. And in doing that,
00:29:30
Speaker
you end up detrimenting the data experience for other user groups. So look, we really try to avoid chasing perfection in this area and focus on directional improvement, making sure month in, month out, or quarter in, quarter out, we continue to be better data culture. We continue to raise the bar in terms of decisions, how those decisions are sustained by data.
00:29:57
Speaker
evangelizing the importance of non-data people to invest in their own data literacy as well because they are consumers of data and manipulate data and have to put metrics and decisions together. So all of those things consistently through time hopefully land an impact.
00:30:15
Speaker
Brilliant. Well, look, that's brilliant. And we've touched upon some of the core areas there and the building blocks of a successful data team in a startup scale-out. So this all leads to impact. So it'd be great to hear how your data team at Plum measures impact and share some of the work and some of the projects that you are doing and what impact that's having on Plum as a business.
Measuring Data Team Impact
00:30:43
Speaker
Yeah, selling the ROI of data teams or communicating that, again, as previous people in this show and this podcast described much more elegantly than me, it's a bit of a million dollar question and something that doesn't have a sixth answer for. In the context of, and that applies to any data team operating in any company of any size or industry, but particularly in scale ups or startups.
00:31:12
Speaker
The fact that data is completely multidisciplinary by nature, so you don't have multiple data divisions, you tend to be still in one data team that have different specialisms. The fact that what the stakeholder sees is the absolute tip of the iceberg normally from a technical perspective.
00:31:31
Speaker
The fact that data teams sit in an intersection of departments, some teams operate closer to business, some closer to engineering, but certainly all of them traverse product, means it's super hard to demonstrate direct ROI in pound against every single action or every single project you do.
00:31:50
Speaker
So for me, it's a lot about trying. Perhaps this is a theme across our conversation here. It's about making sure you're progressing towards the right direction, making sure you're responding right to the challenges around prioritization, planning, and value, and iterating to make those relationships with your senior stakeholders better. I tried to use three things to do these better on my day-to-day.
00:32:16
Speaker
One is to try to have fixed schedule updates to leadership regarding the challenges your team is facing, progress towards key projects, costs, because ROI is a function of costs. So it's quite important to communicate the costs of running data in your company.
00:32:34
Speaker
And the KPI is a success metrics that define how well you're building your data platform. So, making sure that you have a monthly, quarterly, weekly, whatever works for you, but fixed cadence and relatively fixed format update prepared for some key stakeholders regarding all things data.
00:32:54
Speaker
normally tends to attract the right questions and satisfy a lot of the questions regarding what are they getting from investing in data. The other thing is making sure that intelligence through data, so insights, it's constant. It's a constant stream and not the one-off or a special thing that leads to celebration.
00:33:15
Speaker
Again, I've been in companies where because of different reasons, cultural investment in data, the roles, the ratio of analysts to data engineers or analytics engineers was lower, then the activities related to learning more about your users or learning about the behaviors that are driving
00:33:34
Speaker
the most impact for the company was relatively rare because the team was more focused in optimizing the data platform to make sure that other operational tasks were performed. So here at Plum, we want to make sure and we are making sure that intelligence through data is constant. It's not something we celebrate. It's part of what we ship constantly ranging from nuggets, small insights that can affect or stimulate someone to think
00:34:00
Speaker
differently about their users or about the features they're building, all the way to deeper dives, insights where we'll go deep into an area that we haven't covered much. We start with a range of hypotheses and we prepare a pack that describes general knowledge about an area we didn't know much before. So yeah, really making sure intelligence, that the team is shipping intelligence and knowledge about the user base is important. And third,
00:34:27
Speaker
I try to be, and we need to be better at this, and it's something I constantly try to iterate, but be clear about your prioritization rationale and transparent about what your team members have live in terms of projects and why. So not only what they are doing, but try to have a strong level of knowledge about
00:34:48
Speaker
why the things were prioritized in a certain way and the importance to do those. And having a documented prioritization rationale, it's normally great and it protects you from different questions around prioritization.
00:35:03
Speaker
Nice. I like that end. That why is so important as to helping others explain why you've got that prioritization in place. And I think that can help with building your own rationale and understanding of what your prioritization is, understanding that why.
00:35:21
Speaker
is. So you've been at Plum for six months. We're recording this at the beginning of the year in January.
Future Goals for Plum's Data Team
00:35:31
Speaker
What does success look like in data at Plum over the next 12 months for both you and your team?
00:35:39
Speaker
There's a few different things, but broadly speaking, and this is also a reflection of how we organize ourselves. So the team is slowly and sustainably, but nevertheless growing and trending up in size. We have two hires joining. We are 12 in size with me. So that's quite exciting. And we organize ourselves in like different areas. So we work together under one mission statement of maximizing the value of data. But we obviously recognize the importance of the different responsibilities and roles and areas of focus.
00:36:09
Speaker
you
00:36:09
Speaker
So for 2024, it's a lot about three things, really. One is to really go deeper into understanding our user base and feeling the last batch of gaps, fundamental gaps we have in terms of understanding why users use plums, stick around, love, some churn. Understand fundamentally our user base by ramping up what we do in terms of deep dives, advanced analytics, and insights is going to definitely continue to be an area we focus a lot.
00:36:39
Speaker
The other one is continuing the modernization of our data platform and making sure we really build the best in class in this area. I'm super proud of what we achieved already prior to me. So obviously, I'm very far from being responsible for all the successes in that area. But the foundations are right, and we really identified
00:37:05
Speaker
the problem statements regarding our data platform and the scalability of some of the operations we have in data. And we know exactly what we need to do, and we have the right people in place. So it will be a lot around elevating the data platform to be in the place we want to be. And hopefully, you'll hear us ramping up our participation in public forums, meetups, blog posts, and sharing some of the exciting stuff we're doing in this area.
00:37:32
Speaker
Lastly is delivering on the promise of AI. So, as a data team, one of the sub-functions of data is related to data science and machine learning. And as I said in the beginning, 64, we're doing big bets in that area. And we're going to continue to do so.
00:37:52
Speaker
And hopefully deliver both internal data science, machine learning and AI products. So products that allow us to do things we already do, but much better, much faster, much cheaper, as well as customer facing AI products. So yeah, those are the quests we're going after.
00:38:09
Speaker
Very exciting,
Connecting with Jaul and Closing Thoughts
00:38:11
Speaker
very exciting. Well look, it's been great to cover off these really key points and I think really you've shared some really useful information as to some of the challenges that it heads of data, it's got a scale up space, but also for people looking to move into the startup world. So if anyone wants to learn more about Plum, then feel free to connect with Joel. I don't know if he'll be hiring at the time, but I'm sure he'll always appreciate a connection.
00:38:39
Speaker
So yeah, thank you for your insight, Cheryl. We've just got the final round of quick fire questions, which we asked every guest on the show. First off, how do you assess a job opportunity in your career and how do you know it's the right move for you?
00:38:55
Speaker
In the spirit of being a quick fire round, a quick straight the point answer on this one, I tend to make sure I evaluate the role in terms of understanding if it constitutes an opportunity to engage on something new. So to me, it's all about optimizing for personal growth at this phase of my career and trying to experience roles that give me something I didn't experience before and force me to leave my comfort zone.
00:39:23
Speaker
So this could be new responsibilities, a new tech stack, a bigger team, different industry, there's different ways to do that. But yeah, I tend to optimize for roles that are promising when it comes to personal growth.
00:39:37
Speaker
Nice. And you also touched a bit upon that in your first answer, right? It's about understanding what you're stepping into is a really big, big part of that. So second question, what is your best advice for people in an interview? Something that obviously came up before, but I think it's so important, especially over the last three years I've conducted
00:40:00
Speaker
hundreds of interviews in different companies. And I see these on and on. Be yourself, right? Avoid memorizing answers to cliche questions that you think are going to come up. Ultimately, it's quite hard to keep kind of a role play or something you're not necessarily for long enough. And it will create the wrong impression if that happens and you end up engaging again with the team and not being able to play that role.
00:40:28
Speaker
Really be yourself, particularly in scale ups. That's super important. People want to connect with you from a personal and human perspective, first and foremost.
00:40:37
Speaker
And second, this is a bit of a... Yeah, I think it's interesting when I see this one and on again. Don't assume interview questions have to be very hard with traps or nasty angles to it. Normally, questions are meant to be simple, and I see this all the time. It prevents good candidates from answering correctly things they actually know because they're assuming that the question can't be that straightforward.
00:41:05
Speaker
It's a reflection of something a lot of data people and technical people have, which is a slight tendency to overcomplicate. So try to avoid that and hopefully it will give you better results.
00:41:17
Speaker
Yeah, I agree on the over complication. Sometimes it really is as simple. And what I think to add that my advice, which I always give candidates is you can always ask the interviewer for clarification if they're expecting more. I think that's something that not a lot of people do, you know, you finish your answer and say, you know,
00:41:35
Speaker
Was that the answer you were expecting? Do you want me to elaborate further? Something like that, you throw the ball into the interviewer's court for them to then say, no, that's great, or give the interviewer the opportunity to be able to allow you to elaborate. So yeah, great, great advice there. Final piece, final question. If you could recommend one resource to the audience to help them up skill, what would it be?
00:41:59
Speaker
I'm going to be really annoying and recommend a few different things that I've been finding useful and touch on different topics. One is a book by Cheap UN regarding MLOps called Designing Machine Learning Systems. It's something I've been referencing a lot over the last six months and prior companies where we had to develop and mature kind of machine learning systems from scratch in environments that are kind of fast-paced. So definitely one of the best in clubs in that area, in my opinion.
00:42:29
Speaker
On another topic equally important for startups and something that, to be completely honest, we need to continue to improve at Plum and in different places. It's regarding A-B testing, particularly this book called, Trustworthy Online Controlled Experiments, A Practical Guide to A-B Testing is by Ron Kohavi, Diane Tang, and Yashu. Definitely packed with gems and knowledge, important knowledge in this area.
00:42:55
Speaker
On prompt engineering, even though AI has been the buzz, the topic for 2023 and certainly will stay the same for 2024, I think it's super important to understand prompt engineering. So it's becoming almost like a non-technical skill, right? It's how to develop and craft prompts that allow you to maximize the value you get from an LLM response.
00:43:20
Speaker
And there's a particular article I read recently, hopefully we can share on the link, which I found super interesting and touches on the different techniques that can be used. And then on a non-technical camp completely, something I finished reading recently and it's regarding a topic of enormous importance, climate change.
00:43:40
Speaker
in the city of London. It's called Breast Tackling the Climate Change Emergency by Sadiq and the Mayor of London. And I highlight this book to just emphasize that it is important to kind of learn about the world we live in. And also because this book touches on the importance of data reporting and evidence in showing and demonstrating
00:44:02
Speaker
why certain existential crisis and subjects have to be addressed. And so there's different, there's interesting reflections on how the team use data to show that they needed to invest more in this area.
00:44:15
Speaker
Amazing. Well, look, that is great. We'll, uh, we'll link the posts and jab tag all the, all the books in as well in the comments. Um, Joel, I really appreciate your time. It's been great to, to unpick the challenges of the growing and managing a data team within a scale up. Yeah. I'm sure the audience have learned a lot. Thank you for your time. Harry, thank you so much. It was a pleasure. Thank you so much. We'll see you next week, everyone.
00:44:45
Speaker
Well, that's it for this week. Thank you so, so much for tuning in. I really hope you've learned something. I know I have. The Stack podcast aims to share real journeys and lessons that empower you and the entire community. Together, we aim to unlock new perspectives and overcome challenges in the ever evolving landscape of modern data.
00:45:06
Speaker
Today's episode was brought to you by Cognify, the recruitment partner for modern data teams. If you've enjoyed today's episode, hit that follow button to stay updated with our latest releases. More importantly, if you believe this episode could benefit someone you know, please share it with them. We're always on the lookout for new guests who have inspiring stories and valuable lessons to share with our community.
00:45:29
Speaker
If you or someone you know fits that bill, please don't hesitate to reach out. I've been Harry Gollop from Cognify, your host and guide on this data-driven journey. Until next time, over and out.