A Foundation for Data-Driven UX
At ProductTank NYC, Lindsay Silver talked about the challenges of blending the inspiration of fashion and taste with the guidance offered by huge amounts of hard data at Conde Nast, where he is the Head of Data Technology. Specifically, he walked through an internal data product for that leverages UX data to support creative and commercial decisions, making it easier to build digital products. This talk highlights some of the technical specs of their product, as well as the “data usability” factors they discovered and the requirements for an effective UX data infrastructure.
Conde Naste has a staggeringly huge number of users, and a huge number of interactions with each of those users, which obviously generates an overflow of data! Different teams in the organisation have used various 3rd party analytics tools, and they generally work okay . However, those tools are ultimately general tools, and having data in 3rd party repositories slows down the process of getting the data the teams need and doing the exact analysis they need. So, the team decided to built a custom data framework to solve the specific media product challenges they were facing. Going into this project, they knew the data systems would need to be SMART:
- Scalable – able to expand gracefully
- Malleable – flexible enough to grow and fit future needs
- Accessible – Simple enough to be used operationally, and granular enough to answer specific questions
- Reliable – Exactly what it sounds like!
- Timely – Provide data in a reasonable time-frame, for a reasonable history, at a reasonable granularity.
The team were trying to satisfy varied demands – the data needs of engineers and CXOs, and being able to cope with a huge, varied and changing ecosystem that Conde Naste is pushing data around. The good news is that they succeeded in building a real-time system allowing them to push event data about their products to any system they need, make custom metrics, define custom meta-data, and build a data infrastructure that helps them build and rapidly improve their digital products.
Lessons Learned Building a Custom Data Infrastructure
It is crucial to understand that your product is not a category, and that you are building it to fit a specific need. So, in the case of Conde Naste, they weren’t simply building “a web app”, they were building “Vogue Online”, and so Sparrow needed to reflect that.
Crucially, when you’re dealing with vast amounts of data and thinking about usability, you’re like to arrive at an idea something like:
Accessibility of data = Granularity of data / Age of data
What Does the Right UX Data Make Possible?
Sparrow takes in over 10,000 events per second, and access to that data is devolved out to individual teams, empowering them to make the decisions that matter to their specific challenges. So, for example, their editors are working on a system to dynamically adjust publishing times of new articles to maximum impact, based on the entities in each article and historical data about what’s worked well.
They’re also working on targeting content syndication based on user behaviour, and intelligent content resurfacing based on trends and surges in topic interest. The data is enabling smarter ad placement based on trend spotting, and behavioural targeting.
Basically, having the right data framework in place is enabling Conde Naste to embrace digital publishing as a truly distinct phase of the media industry, understanding how their users are interacting with their products in an incredibly fine-grained way, and then blending that data with creative intuition. The result is that they can build new digital products faster and more effectively than ever before.
If it’s enabling them to change the way they build media products so dramatically, then it’s worth taking a look at your own data infrastructure and analytics options, and asking yourself if you’re getting the right data, enough data, and whether you’re able to use the data in the ways you need. It probably isn’t necessary to build your own infrastructure for it, but you can at least make some smart decisions about how your existing one is set up, and what analytics tools you’re using!