Recently (19-21 November) I went to the Strata + Hadoop World conference in Barcelona. In particular a talk by Jagex, creators of Runescape really stood out for me.
Runescape is a free massively multiplayer online fantasy role-playing game (MMORPG). It has over ten years’ worth of game content, and 220 million player accounts since its launch in 2001.
Jagex had a problem. It needed a way to recommend relevant content to new users from 10 years’ worth of available quests. The risk was if new users were recommended a battle with for example Queen Black Dragon (at the time the hardest dragon in the game) when they weren’t ready to, it could disengage new players, and damage revenue lines.
The world Jagex created is data rich, and its solution was to create a bi-directional recommendation system for Runescape. In English? Jagex’s data scientists created a data feedback loop enabling Runescape to use its current data to recommend highly relevant content to its players based on player interests, delivered within the game mechanisms, and continue to feed new data back in to keep this up to date.
The user focused bi-directional model drove increased revenue lines, improved player engagement, and quest completion rates. It allows Jagex to see which content is popular to users, and plan new content accordingly. I found myself asking, how could this apply to the advertising industry and user experience?
I thought - it’s simple right? Adding user data to advertising campaigns provides a more targeted way for advertisers to reach users and receive a higher return on investment (ROI). However in reality it’s not as simple as the Jagex example.
The advertising world and its data is complicated, particularly user data. To achieve accurate user behavior attributes requires scale. To get this, advertisers often have to combine first and third party data, which can come from a variety of sources (e.g. various DMPs). When you look at who your users are across devices it gets even more complicated. Essentially it is not a closed loop, and the system lacks transparency.
Big players like Amazon, Facebook and Google have the scale to build up profiles of users based on behaviors, and make it available to advertisers or marketers who might want to, for example, identify new prospective customers and drive traffic to their sites.
Collecting data on its own is not a quick fix (to feed it back into the business you need the skills in-house to do this), and big data is not always accessible for smaller organizations. This is where Demand Side Platforms (DMPs) come in as a way to access third party data (and collect and store your own if you’re a media company). This undoubtedly is already going to give a higher ROI to advertisers as there is less wastage of ad spend.
There are examples of data boosting campaign effectiveness, particularly around increasing brand awareness through targeted advertising. Nugg.ad references increased brand awareness and affinity for Toyota of 39% and 33% respectively, with many more examples like this.
Then what about the end user experience? This is definitely a challenge for the ad industry (unlike for Jagex’s situation) even with targeting. It raises issues of privacy and what is and what isn’t considered “creepy”. If the recipient of that advert finds this invasive, could this do more damage than good to the brand? But without this targeting are we just serving up Queen Black Dragons that will repel users from your products and wasting ad spend?
There is no easy answer. However I do think there is a subtle difference between logging into something and getting recommendations based on your behaviors as opposed to being targeted adverts for something you didn’t know you’d expressed an interest in, or you may not want your data collected at all.
That said I think there is a lot to be learned from the Jagex story. One of the reasons it resonated with me so strongly is that it is a really neat example of what the advertising industry is not able to do at scale with its data. But the principles are the same. Knowing and learning from your users, and utilising this knowledge to tailor targeted recommendations can result in a higher return on investment, and (hopefully) a better user experience. And there will not be a one size fits all.
If the ad industry could get its data in order, this investment in data could be a win-win-win; for advertisers, media companies, and users. But the journey shouldn’t stop at one successful campaign. Look at what worked/what didn’t – now feed that data back into your decision making!
Find out more about: