How Domo powers partner analytics at Sony Interactive Entertainment

Sony Interactive Entertainment sees it as an imperative that its partners are treated as an extension to the wider PlayStation data ecosystem. To that end, the gaming behemoth has turned to Domo to deliver a robust partner analytics offering.

Sony has two types of partner - games partners and media services partners, such as the likes of Netflix, Disney+ and Hulu. Both provide input into the PlayStation data ecosystem and in turn gain value from insights delivered via partner analytics, says Brook Girma, Technical Product Owner in the Partner Analytics Team:

This provision of data to our partners should be perceived as an enabler of competitive advantage, whether it's through delivering truly next gen features by blending partner data with our own, or through empowering our partners to maximize the reach and potential of their content at Playstation. Our partners go through the content creation process where they create, test and ultimately deploy their games. Once it's released, that's where the partner analytics tool comes in.

Objectives

What Sony set out to do with its analytics program was to deliver actionable indicators that, when coupled with the correct levers, would ensure that players are kept engaged with relevant and timely content. Girma explains:

We wanted to provide a single focus data delivery platform for all of our partners and realize and unlock the value of partner and PlayStation data through innovative data-led initiatives. With the changing paradigm around content creation, namely the free-to-play and games-as-a-service business models that have emerged, we needed an offering that can both meet the demands for low latency insights, as well as into our content performance. Our analytics tools become a key component of the wider PlayStation partner program, which aims to make PlayStation not only the best place to play, but also the best place to develop and publish content.

There are two fundamental principles underpinning Sony’s strategy - service delivery and the service offering itself. The first comprises the set of components responsible for the provision of the partner analytics service and is built on a tenet of reliability. As a partner-facing service, it has an active Service Level Agreement (SLA) that operates at multiple levels and ensures that fallbacks exist where applicable.

It’s also based off a well-structured security model, ensuring that partners only access data that is relevant to them, but still flexible enough to cater for complex partner relationships, as well as addressing single account partners.

And it must, of course, be trusted, notes Girma:

The content made available via the partner analytics service must have a high level of data quality focus against it, ensuring that we have both proactive and reactive mechanisms for dealing with data quality issues. An example is how we've set up alerts on our data warehouse to Domo pipeline to let us know when gameplay data volume has exceeded or fallen below a certain threshold from the previous refresh so that we're the first to know and address the issues before users are even aware of them.

It’s a demanding ‘to do’ list and one made all the more so by the need to be able to scale the service from one to 100,000 partners and their users, with only a team of four in the Sony partner analytics team. The analytics project kicked off in November of 2019, when the data architecture and preliminary work began. By February 2020, an architecture had been defined with a UX and embedding approach agreed on. The partner analytics team took shape in March - and then came COVID. Work continued and, over the next couple of months, the curated content that would eventually appear in a beta release was developed. Girma recalls:

In July of 2020, we went live with our beta where we picked six active partners to get early access to the tool. The success of the beta increased both the partners and our confidence in the tool and production environment. It allowed us to test internal operating procedures with real partners, and gain end user feedback by providing hands on experience with the tool so that we could eventually hit the ground running with our full rollout in August. In August, we expanded to a much wider audience and then also introduced net new PS5 assets and reporting metrics during the PS5 launch in November of 2020.

In practice

As for the service offering itself - AKA how partners perceive the platform - the first essential is that it is multi-faceted, says Girma, ensuring that a partner is provided with the right data asset via the most relevant mechanism. The second is that it's action-oriented, with the service providing assets that inspire action, not simply provide data for data's sake. Finally, it must be fit for purpose and cater to different consumption preferences for different partners. Girma notes:

Some want to access it via the standard portal, while others want direct access via API. We have data sources from across PlayStation console generations, including PlayStation specific services like PS Now, PS Plus in the PlayStation Store and of course, all this from our game and media partners.

He goes on:

We offer the analytics tool as you'd expect, but also couple it with service data using Domo stats, so we can see exactly how our users are engaging with the tool and adjust based on their behavior. The different access mechanisms via the PlayStation portal are also via direct API access and we also have indirect access mechanisms via integration with other tools available in PlayStation portal. For example, if a partner wanted to create an official news post, which is a feature that allows partners to make announcement posts within the PS5 UI, they can actually deep link from the post creation tool directly into analytics so that they can see how many impressions or CTA clicks the posts received.

In terms of implementing partner analytics at scale, each partner on the platform is given their own instance along with a unique ID which is attached to partners respective data in all datasets. Those datasets from the Sony data warehouse are then loaded into the main instance in Domo.

Standard sets of dashboards are created for game partners, and another set for a media services partners. To get these curated dashboards in their respective datasets from the main instance to partners, Sony uses Publish V2, which allows the creation of a view of the content that is sent to subscriber instances. When dashboards and data sets are updated in the publisher instance, those changes are reflected in subscribers instances upon refresh.

Usage and feedback

The out-of-the-box Domo stats dataset has been really helpful in understanding how partners are  using the tool, says Girma:

We can confirm whether our assumptions were correct and can make tweaks to our curated content if needed. Or understanding when users are logging into the app the most versus when they're on at the least. So in the scenario that we need to make any upstream changes or if we're migrating between AWS environments, Domo stats has been really key in understanding when they're on the tool so that we can choose times where it's low traffic.

Overall, the partner analytics program has had what Girma describes as an “overwhelmingly positive” response from the start:

We've had one major partner say it was ten times better than our nearest competitor's offering. We had another one talk about how much they love the speed and responsible responsiveness of the tool. So really, the only feedback that we got that could even be perceived as negative, was that the tool was so new to them, they just wanted to get more out of it. They were hungry for that contextual information. That's kind of what inspired that decision to create description cards and put the help articles within the tool itself.

Feedback sessions with partners have also helped influence enhancements to the program, he adds:

We had one partner describe their standard Monday of logging in and wanting to view their sales information, but not knowing whether the data that they were looking at was up to date. So we actually took that back and we decided to surface both the refresh times and schedules from our data dictionaries, along with the latest available data in each of our datasets, into one card and that ended up working out so well that we ended up sharing it with all of our partners. So being able to experience that feedback loop between hearing that pain point from a partner, being able to come up with a solution and quickly deploy that so everyone benefits, is exactly what we're about and how we want to do things.