Peter Day is director of engineering at Quantcast, a company that provides AI-driven audience insights, targeting and measurement, to improve ad campaigns.
Day is also our latest (and aptly named) ‘Day in the Life’ interviewee.
Econsultancy: Please describe your job: What do you do? And who do you report to?
Peter Day: I’m the Director of Engineering for Quantcast, an AI technology company focused on the marketing and publishing industries. We use apply machine learning to our unique data drawn from 100 million online destinations to uncover insights to help brands get closer to their audiences and create ad campaigns that help them grow. I’m responsible for our team of engineers in Europe and report to our SVP of Engineering in San Francisco.
My team is made up of data modelling scientists, research scientists and other experts who come up with new ways of understanding consumer behaviour online using machine learning and AI. We are constantly iterating on our algorithms, tweaking them to react to changing consumer behavioural patterns, and inventing new ways for brands to connect with their audiences. This produces up-to-the-minute live audience insights that get used by marketers in their creative process.
E: What kind of skills do you need to be effective in your role?
PD: While a high level of expertise in machine learning is important, creativity is essential. There are around 50 PhDs among the team at Quantcast. Developing software is about starting with nothing and building something, so you need to be resourceful as well as smart. As a leader, I focus on cultivating an environment that combines the two so people can really thrive in their roles.
E: Tell us about a typical working day…
PD: While there really isn’t a typical day, one thing I make sure we do as a team is a daily “stand-up” meeting at 10am. As a rule we aim to minimise meetings but have found having this daily touchpoint is really important. After that, I go where I’m needed.
I try and balance my day between people and data. I’m often spending time with media agencies and senior marketers educating them about AI and machine learning. I also try and carve out time each day to focus on the data we’re working with, whether that’s building data models or writing my own software to keep myself in “coding mode”. Spend too long away from the data as an engineering leader and you can’t effectively advise your team on how to tackle challenges. When I’m not doing that, I’ll be trying to throw regular surprises and challenges at my team to keep them thinking.
E: What do you love about your job? What sucks?
PD: The best part of my job is solving seemingly impossible problems with clever and interesting people. I love getting stuck into the detail of what we’re doing which reminds me Quantcast really is the pulse of the internet. Deep learning is already taking our understanding of human behaviour to new heights, so being able to spot changes in data patterns and react to online behaviour is fascinating.
What sucks? Not having enough time. Our customers are all over the world which I means I travel a lot. I enjoy spending time with them, however I’d like to spend more time with the team and family.
E: What kind of goals do you have? What are the most useful metrics and KPIs for measuring success?
PD: We set our own goal and metrics, which we review regularly. Specific measures differ by project, but there’s always a focus on giving value to our customers. An example might be improving advertising performance by 10% in the next three months.
Marketers are under increasing pressure to show value from their budgets. Where they succeed, we succeed. Our measurement tools allow brands to more accurately tie their activity to brand growth so, wherever possible, we’ll benchmark our success by the success of our customers.
E: What are your favourite tools to help you to get the job done?
PD: In my role, effective communication is vital given how crucial the engineering function is to the business. As a result, anything that allows me to communicate more effectively has value. Quantcast currently employs more than 800 people across 10 countries with one in four of those in a product or engineering role – so there are a lot of people to communicate with.
Some of the people I work with prefer different styles of communication. Some of the best conversations I have with my boss are via online messaging, whereas when I speak to the commercial lead in Europe, video conference or the phone can be more effective.
E: How did you get into ad tech, and where might you go from here?
PD: I studied machine learning at university, and after that the most interesting industry to leverage my skills was in finance because of the huge amounts of data the sector was dealing with. However, I soon found myself running a massive department spending more and more time on the day-to-day complexities of running a large organisation rather than focusing on what I’m passionate about.
At the same time, the digital advertising space was exploding and producing vast quantities of data. I was familiar with Quantcast’s CEO, Konrad Feldman, since he’s had a similar path to me. I found out that the company took machine learning seriously, so that’s what sealed the deal for me.
E: Which brands have you been impressed by recently when it comes to machine learning led marketing/advertising?
PD: We’ve been working with travel brand Tui closely for a couple of years now and they’ve done a fantastic job at leveraging AI in their campaigns. They are completely open to what machine learning is telling them and adapting their strategy to emerging patterns in consumer data. As a result, they not only understand their own audience but also their competitors’, which helps them to tailor their messaging and product set to market. They are going beyond using machine learning just for targeting, also understanding existing and potential customers in much more nuanced ways.
E: Do you have any advice for people who want to work with data in adtech?
PD: To work with data in adtech, you need to get your hands dirty. Don’t rely on other people to do stuff for you. Most of the tools available these days are straightforward if you put the effort in to learn how to use them.
Most magic that emerges from data tends to be through exploration, so my advice would be to skill-up, learn SQL, learn data analysis techniques, and listen to your creativity. You’ll find that doing it directly will help you ask the right questions.
Subscribers can download Econsultancy’s Trend Briefing: Artificial Intelligence (AI).