My name is Hubert Wan. I’m a 24-year-old Electrical Engineer with a passion for Maths and Data Science. I graduated from uni last year and managed to get my first job at iProspect Sydney, working in the programmatic team.
It’s been an incredible experience so far. Everything in the Digital Marketing space is new for me, especially considering my engineering background. There are a lot of interesting challenges in digital marketing and the programmatic space, which has really engaged me and I’m excited for what’s coming up in the future in this space!
In this article I’ll be talking about what my experiences have been, how I’m adapting to this new environment, what I’ve learned from the marketing and programmatic worlds, and how I feel I have grown in the role.
Firstly, what is my role? As a programmatic media trader, I’m responsible for devising, executing and optimising digital programmatic campaigns. I report into the Head of Programmatic, and work alongside and within paid media teams to help drive overall performance for our clients who utilise display channels to achieve their business goals.
As I’ve been learning more about the programmatic space, I’ve also been building in-house tools to improve and automate processes, bringing greater efficiency, scalability and consistency to our product. This involves extracting data from Demand Side Platforms (DSPs) such as DBM and transforming data and metrics into an interpretable format.
After processing and interpreting this data, we make changes to our campaigns based on contextual and behavioural tactics that we are constantly developing and improving. On top of that, I’ve also been working on separate analysis using log-level data to gain further insights, such as frequency limits to users, and which conversions were viewable.
Another thing I’ve also been working on is optimising campaigns by budget pacing. In this process, I attempt to come up with algorithms to determine the optimal way to balance our tactics depending on performance, and test these algorithms on a day-to-day basis to improve CPA of the activities we’re running.
Coming from an Engineering and Mathematics background, I initially had no knowledge of what Digital Marketing was, how exactly it worked or what it was able to do. However, as I developed an understanding from working here at iProspect, I soon began to appreciate the many opportunities and initiatives that might be available thanks to our many interesting technology options. This has allowed us to build interesting strategies and tactics across our clients’ digital channels.
To me, these ideas are even more exciting in the programmatic and media trading space, with the ability to actively target segmented audiences in various ways, including through open bidding in the marketplace to secure preferred deals with other media sources, such as directly with publishers. It’s usually through open auctions that we can target segmented audiences through auctions, allowing us further flexibility and increasing the potential for greater, more consistent performance.
However, these options also increase the complexity of buying media. There are a lot of interesting questions when it comes to finding the best strategy or increasing the depth of strategies. The fact that even with DSPs internal algorithms a trading desk can make critical changes on top of this that can dramatically affect the performance of a campaign demonstrates the potential of a good strategy and a well-run trading desk.
The opportunity for growth, building new strategies on top of an ever-changing environment with both rapid and significant technological and social changes makes digital media to me an exciting place to be with a lot to learn.
Coming from an Engineering background and having some experience in computer programming, operational tasks and processes are things that are quite important to me. In the past three months in iProspect, I’ve found that you not only learn new things from a new environment but also understand that are always things not being done as well as they could be. These changes that I’ve been involved in making can also make life easier and tasks more achievable across the board, with or without programming experience.
It’s always important to scrutinise your processes, no matter how long you’ve been doing things a certain way. A common problem that seems to occur across the board is that a lot of tasks are extremely manual and repetitive. As the number of tasks increases, scalability and consistency become an increasingly larger problem. By being able to analyse processes and either improve them or remove redundancies, not only are we able to complete tasks more quickly and efficiently but it also creates extra time to build and expand on our capabilities, resulting in a more effective solution. The process of scrutinising processes also allows us to understand the steps we take to do things, which is the key for allowing the automation or optimisation of tedious and repetitive tasks.
One way of doing this is to have at least some basic knowledge of what the software you’re using really does when you do a certain task. This is important not only for my role but other roles too, as through understanding software, we can understand the capabilities of tools and use them to their strengths while avoiding their weaknesses.
As many of you know, although Excel is great as a calculator, it’s not so good at handling even moderate amounts of data, typically struggling with tens or hundreds of thousands of rows. On top of this, without macros or knowledge of VBA, Excel also provides poor solutions for scalability. To improve scalability for data processing and segmentation, I use languages like Python and SQL to build in-house solutions for my team, using methods and techniques that are better suited for handling large amounts of data.
Through working with data, I have also learned the importance of understanding hidden algorithms and ensuring reliable data sources. This is even more important with data-driven decision making. As the ability to handle large amounts of data increases, so does the potential for errors or inaccuracies due to mis-collection and misinterpreting data. The problems caused by these errors if not detected grows considerably worse and worse and is exacerbated by not being informed of changes made by other platforms that provide or store the data.
Therefore, it’s incredibly important to be vigilant and check what your data means and whether what’s reported is accurate or not (not only checking for false positives, negatives or name checks but also whether the way data is stored works as you think it does). This applies to both dedicated data teams and also any general users of the data.
Finally, I’ve found it helps to actively ask how people work on certain tasks. This helps me to understand what they do and reveals more efficient ways of doing similar tasks. It also allows us all to understand their processes and eventually come up with effective, scalable solutions for repetitive tasks that can solve everyone’s problems.
Through this journey I feel that I’ve grown a lot as well. Since this is my first job, I’ve encountered many new problems myself, especially in a workplace environment, with noticeable differences from the more academic background that I come from. It has taken me some time to adjust but it’s a challenge I’m somewhat excited about, having to carefully balance and manage different aspects and interests of working life.
I have also been improving as a developer in the past three months, having developed tools for use within my team. This is the first time I’ve been exposed to data and processes at a commercial level, which is exciting to someone that has previously explored data at a hobby level. The thing that excites me is that the insights that I find can also be immediately acted upon. I’ve also been developing further Python and SQL skills through various projects, learning more efficient and sound ways of handling data.
Through this process, I have become aware of scope creep. Scope creep for those who aren’t familiar is a development trap where without proper management, the goals of development continue to expand while being developed until it either takes considerably longer to complete original tasks or even failure of delivery can occur. My boss was able to manage the project well and it’s something I’m learning to try to identify when developing projects.
Where to from here?
It’s been a dizzying three months where I feel I have learned a lot and grown a lot from this experience, while helping the programmatic team with its ability to act more proactively rather than reactively. However, there is still a lot to learn and a lot to build. Some projects I’m still working on include analysing log-level data for extra insights not available by simply querying DSPs from their interface, building data warehouses to retain information and improving analysis of data to gain more insights such as using Time Series Analysis.
I’m also keen on teaching Python and SQL along with programming ideas and techniques to others within the company, either on demand or over time. Of course, I haven’t really taught programming languages, especially to people without programming languages, so this will be an interesting experience.
really can’t wait to work on building up our capabilities, not only within the programmatic team but also the company as a whole.