Gone are the days where advertising and sales were the sole remit of creative-thinking Mad Men types, who wildly conjectured who a product’s target audience might be, and constructed campaigns which may appeal to them. Today, advertising is a much more precise, data-driven field. Of course, creativity has a role to play, but so do audience targeting and automation. We sat down with Anna Cremers, Chief Data Scientist at nugg.ad, to discuss her work and the evolution of advertising.
Tell us a little about yourself and your work.
My name is Anna Cremers, and I work as Director Data Science at nugg.ad. nugg.ad is involved in predictive behavioural advertising- so we’re compiling and offering predictive audience segments to the players in the online advertising ecosystem.
As you can imagine, as a predictive product, data science is more or less the core of what our company offers to our clients. We have huge amounts of data going through our systems- we have, for example, a market reach of about 85% in Germany. We also have market coverage in other countries like France and Denmark- so if you take all of that into account, you can imagine that we have to deal with tons of data.
We have a real-time approach, so we deliver our audience segments in real-time, meaning that with every click a statistical model is updated. So we need super-fast algorithms to serve our needs. Â We also need a hands-on data science team to get things going.
Are there particular technologies you use to process this data, or is your technology custom-made?
It’s a mixture of both. So we work with a Hadoop system for data storage, but for our predictive models we use self-built algorithms. Â From an analyzing perspective we’re working with typical software programs like R- so a combination of technologies.
You’ve worked at nugg.ad for seven years- how has your role changed in this time?
When I started, nugg.ad was a start-up. I was around the fifteenth person to join the company. nugg.ad had existed for about a year at that point, so the data problems and challenges were different. We could work with a slower algorithm, for example, because the dataset was much smaller. As any business grows, you have to adjust the tools and storage system, because it’s going to explode if the data exceeds a certain amount.
When you mature as a company, you can start to look into serving special requirements of your clients. You can hone in on the data, build data-driven products. As we grew, we saw this more and more- the need for client-specific reports, client-specific insights, which requires adjustments within the system.
What changes within the advertising industry have you seen during your time at nugg.ad?
At the moment, there’s a clear shift towards automation. You have, both in terms of supply and demand, more automatic tools. Now, you not only have an ad server but an SSP solution, and not only an agency but a DSP solution, so the whole market is becoming more automated.  I think the whole market changed to a more programmatic, more automated way of doing things.
Onto women in data science. Do you think it’s still a man’s game?  Do you think that’s changing?
That’s a really tough question. Every time I attend a conference, I see that it’s a male-driven market and a male-driven field because 90-95% of the audience are male. This is particularly true for speaker line-ups as well. So I think there’s still some sort of mismatch.
But when I look at my company and especially at my team, we have 80% females. But that might be linked to the fact there’s a female heading the team. I think it’s still special to work as a female in a certain role and in a technical environment.
I do see more women- especially young women- becoming interested in this field. I think we will see more women in technical roles, because so many work environments are becoming more data-driven and technical, which makes it essential to have more employees who understand that side of the business. So, in theory, the gender gap should get smaller, but for now, the conferences- and maybe the industry- still seem to be male-dominated.
Why do you think this is?
Well I don’t think you can take a big-picture view and say schools discourage girls from taking Maths, for instance- I honestly don’t believe that’s true. I think the gap starts around the time you start considering careers- there’s so many other interesting avenues you might want to pursue, that you might overlook science and statistics. The gender gap isn’t something you really think about at this point.
For example, looking back at my own professional career, I first started studying sociology- which is a classically female field, I would say. I came to the conclusion that I’m really not that interested in sociological theories, but I am interested in the statistical and methodological part. So I switched to Statistics as a study and finished my Masters in Statistics. I only found out during my studies that this was the direction I wanted to take.
But when I started at nugg.ad, I didn’t think “You’re going to be one of a few females working in this field”. I’ve seen more and more women move into this field- so maybe 10, 20 years down the line, there will no longer be this gender imbalance.
Do you have any advice for young women- or young people in general- looking to move into this field?
I really think it’s the field of the future.  But you shouldn’t underestimate a role in data science- it’s not just crunching numbers and being able to handle data. You really need a full scope of skills. You have to understand how programming works.  You need to have a good understanding of methods, and how to apply those methods to certain data problems. You have to have the business knowledge to understand which kind of questions you want to answer, and also maybe build products out of data.
I think you also have to know in which wider field you’d like to work- we have data science in advertising, medicine, pharmaceuticals and many other areas. But if you know your business and you know your data, I think it’s a really interesting field to work in.
Looking to the future, what are you working on at the moment?
One of the interesting things we’re working on is starting to integrate other kinds of data sources as a predictive baseline.  That’s interesting because you have to adjust your algorithms and your whole data flow. I think we’ll be using these kind of techniques more and more.
Looking at the market, what I think will also be important in the future is focusing more and more on bespoke  audiences and customized segments- so that we really interpret the specific client’s needs with the data. And then, bringing the best out of that data with predictive algorithms.
Any thoughts on the future of data science in the advertising industry, or of data science in general?
I think the future of online advertising is a more automated future- and maybe also more consolidated data ecosystems, where you could add in more and more automation.
In terms of the future of data science; it’s already growing more and more important within every industry, but I what I would hope for is more great initiatives like DataKind, for example. These kind of initiatives are looking into using data science for social good, and tackling real-world problems. I think there are great data science initiatives out there, where you can interact and explore with data and maybe try to make the world better, even if that doesn’t help you to make money.
(Image credit: Wikipedia Commons)
