Now the company is looking to use its gameplay insights to deliver real benefits to the business and to improve customer experience.
For example, by feeding player activity into a dashboard and looping that back into the CRM system, designers can make informed decisions in terms of what to build next based upon where players are actually spending their time in the game, "instead of what they are reading about on Reddit", said Jon Gleicher, business and gameplay insights manager at Blizzard, speaking at Gartner's Business Intelligence and Analytics Summit in London this week.
Gleicher explained that the high-volume, unstructured internal gameplay data, called Battlement, is passed through a Kafka network and stored in a Hadoop cluster with Cloudera. It is then sorted using the extract, transform and load (ETL) capability of Sync Sort DMX and stored in a Teradata warehouse.
Blizzard's business intelligence (BI) and data analysts generally "attack this data in SQL", or Python depending on the data. For the "last mile of analysis" the team uses Tableau for visualising the data in a simple format to pass up the business chain.
This transactional, in-game data is valuable in two ways. The 'strategic value' is where Blizzard can become smarter about its customers, 'operational value' is based around making Blizzard aware of what is going on with its business. This involves data scientists making sure the unstructured data can be reported up to executive level and across to business leaders, with insights such a revenue per-user and subscription data.
Gleicher explains how it took him some time to recognise the value in this data: "When I first got to Blizzard I was wholly uninterested in the operational value add. I wasn't that interested, frankly, in reporting up to the CFO our daily revenue-per-user.
"I wasn't that interested in reporting to the business leads the peak concurrency of their game day-to-day. Now I have been brought back to the centre because there is a part of this journey that involves evangelising data usage and winning people over with data."
As well as this sort of top-line information, Blizzard works with a lot of gameplay data to help developers keep the balance of the games right.
For example, if a certain hero becomes too powerful because players have found an unfair edge, game developers can watch for these sorts of in-game developments in real time and fix the environment, much like a security specialist would monitor a network for unusual activity and threats.
The next step for the Blizzard BI team is around delivering not just operational business value from its data but also strategic value, or in other words "making a better experience for our players", says Gleicher.
He uses the example of bad matchmaking leading to players having a bad experience, by either getting beaten by a superior player or winning too easily, and the risk of subsequently losing customers because of this.
To explain this strategic value add, Gleicher talks about how the BI team changed fundamentally six months ago with the introduction of a new role: engagement managers.
Before this the game developers and BI team were being overseen by a project manager that brought customer requirements in and mediated between the two teams. This tactical approach was "meeting expectations but we didn't feel like we were pushing the boundaries, not reaping dividends from this data asset not because it was anyone's fault but because people didn't know what to ask for," says Gleicher.
So Blizzard looked into a more strategic approach by replacing project managers with engagement managers. The new job spec called for: "MBA people, quantitative and technical. So they understand the business of our games and also can speak IT and big data," says Gleicher. They "own this strategic partnership, so they are only successful if the games are yielding value from data and analytics."
So how does Blizzard bring the strategic and operational streams together to add genuine value First, Gleicher thinks the new proactive approach to business intelligence, alongside the reactive side of day-to-day report building, has helped derive value for Blizzard.
Second, the team is structured around what he calls "engagement pods" where a developer from each game will sit nearby the engagement manager, a data scientist and a report developer to discuss how they can use BI to improve each game.
As Gleicher said, when World of Warcraft was being developed ten years ago "no one was thinking we should capture data around this," they just thought it would be cool to be able to kill dragons in a virtual world. Now Blizzard is doing both.