What’s the role of data scientists on online advertising?
12 Nov, 2013
By Tony Evans, Corporate Development Director at Crimtan.
Nick just joined our advertising company as a data scientist. In fact, he’s one of the 50% of staff who are data scientists. But what’s happened to the chain-smoking, Martini-swilling folk we see in Mad Men, who used to be the driving force behind advertising? Data – and it has completely changed the way that the online advertising industry works.
In a digital environment like online advertising, data has always been used to measure simple things like clicks and conversions and, more recently, to deliver more relevant ads to web users. But technology such as real time bidding and advertising exchanges that allow new ways to buy and sell advertising has catapulted data science to the forefront of ‘must-have’ skills in the industry.
Agencies and traditional ad networks suddenly found themselves having to train planners, buyers and sales staff about complex new technologies. But more than that, everyone needed to recruit mathematicians and data analysts and, almost overnight, agency recruitment companies flocked to universities in search of numerate graduates.
The industry has been revolutionised. Besides an increase in whiteboards and shorter lunch breaks, almost everyone spends more time working with Excel spreadsheets and online data dashboards – and those at the data deep end have had to get to get to grips with new analytics software like Hadoop.
While access to vastly more user data has given many agencies a boost, many still lack the skills, resources and technology to use campaign data to its potential, especially in their battle to compete with the likes of Google and Facebook. Instead they have turned to technology companies to deliver the specialist, data-driven campaigns and insights that many advertisers now insist upon.
But what does this mean for advertisers? The most significant change is that it has democratised online advertising; businesses of all sizes can use new technology and get hold of valuable campaign data – and this can all take place with their marketing budget going much further. Display advertising is now a realistic prospect for many businesses that were wholly reliant on search.
Online cookies allow advertisers to improve advertising efficiency. Sadly, they don’t look like this. Photograph: Getty Images
Retargeting techniques (displaying adverts to visitors after they have left your site) have also become cost effective as site visitors can be bought one at a time when they visit other sites – online cookies also lead to the content of the advert being relevant to users’ interests.
Advertisers can also use a technique called ‘lookalike targeting’ to target customers with attributes similar to existing online visitors – and this, in turn, improves advertising efficiency. Data scientists identify visitor attributes – such as interests, location and demographics – before then delivering adverts to other web users with similar attributes. Branding campaigns also benefit, avoiding inappropriate site content and ensuring that adverts appear in the area of the screen visible before you scroll down.
Data now allows advertisers to see exactly what data variables – from individual web sites, content type, user interests and, amongst others, location – are most effective. By establishing what works best, this data can be applied intelligently – and in real-time – to deliver adverts only to those web users most likely to respond positively to specific messages.
But advertising agencies still need humans to interpret the data and place it in the context of the bigger picture. For example, a campaign might be performing exceptionally well in a particular region, but could this be because you just got some great local PR? What data science does best is to give more control back to advertisers – it gives them the information they need to make intelligent decision about where and how to spend their marketing budget most effectively.
This article appeared in The Guardian.