Justin Johnson

Exactly - and game designers would probably be horrified to learn about some of those player experiences and where players are getting stuck.

They'd be horrified! So even in standard boxed retail there's a strong case for capturing that data, particularly at the stages of focus-testing throughout the title.

But in Sodium's case we found out early on that the way the scene flow was constructed - it's a bit like the design of a theme park in a way, and theme park designers spend a lot of time working out the flow of people.

It's exactly the same thing - and it turns out that a lot of those big retailers are very heavily into metrics and user focus. But it's exactly the same principles with gaming.

We realised there were certain bottlenecks, certain dead areas that people weren't getting to that we needed them to get to - they weren't going to hit our conversion points, our engagement points.

One key example is that people weren't getting through to that important fifth level where they'd be able to make a decision on whether they paid for the rest of the game or not, which is crucial - you need everyone to get to that point. It's understandable if not everyone decides to buy, but at least you've got to get them to the cash till.

We saw various bottlenecks there, and people getting stuck at level three on a boss battle. There were even certain issues with the user interface, which were purely game design issues, being exposed by the metrics.

Cutting a long story short, once we had visibility on that we were able to remedy it. We measured it again once we launched and saw that, sure enough, those log jams had been smoothed out and people were getting where they needed to get to.

In Sodium we saw a 67 per cent jump in conversion, because people were actually getting to the right point - so that was probably the most startling statistic.

It certainly made a big difference to the revenue we were getting, for sure. This is the thing - metrics shouldn't just be used to get retrospective analysis on a game. For example, some payment providers give you metrics on how much people have paid and where they were in the world, etc.

But we call that retrospective - it's after the conversion analysis. It won't tell you what was happening with the players at that point, and it won't let you modify that story, to be able to get the players to convert more - either to experience the game in a fuller way, or to get through to those revenue conversion points.

The thing about metrics regarding that - it's the 80/20 rule, which means that it's the small percentage of changes that you can make which will give you the biggest alterations. There's not really a point where you'll say: "The metrics have just shown us that the whole thing's rubbish."

It's just a few points that people aren't getting to, aren't engaging with, that you can fix. But obviously without a good, customisable metrics solution, you can't find out what they are.

At the moment we use Playmetrix a lot in Home - we've got a lot of spaces using it, a lot of clients in there. To that degree, it's battle-tested, with fairly significant traffic loads. The analysis, the graphing systems are all really tried and tested.

Consoles - particularly the PlayStation 3 - are our landing platform at the moment, although we are in talks and moving forward on an early adoption programme with social games and so on. The good thing about Playmetrix is that it's platform-agnostic, so it doesn't matter if you've got a console game or a Flash game - it could be on Facebook, but if it's not, that's fine.

Essentially, if you can trigger an event, we can track it and produce useful information and business intelligence based on those events.