My buddies and I are pretty devoted fantasy football participants. I should come clean that I’m rather poor at the whole exercise: it’s complex enough that it requires a lot of research, which I don’t do, and then I lose more games than I’d like (and often, in a season, more than I win).
The other day, as I sat watching the flurry of off season trades on my Instagram feed, I started thinking about that complexity and how the immense volume of data involved is translated into what becomes a very engaging and simple user experience. Pulling myself out of the clouds back to the master data governance business case I was working on, it struck me that there’s a similarity between fantasy football and asset data/information management.
As most rational people would, I’m sure that you’re now thinking: physical equipment and assets are FAR more complex than a game like fantasy football. But, consider the following characteristics of the NFL fantasy environment (using a narrow focus on just NFL.com’s platform):
- Core object: The player is the core of the activity, with a wide range of statistics being captured about their performance, actions that need to be taken against them (e.g., add, drop, trade), and related unstructured data the must be captured and made available (news, pictures, video) from many internal and external sources
- Hierarchies: Players belong to teams, which belong to divisions, which belong to conferences; the players can, and do, change their position in the hierarchy throughout most of the season and offseason
- Large numbers: In 2017 there were roughly 689 players on the platform, across 32 teams, each team played 16 games, with 10 statistics being tracked for every player on the field, in near-real time as the games were played; in addition to actual results, the platform must calculate projected performance for each player based on their upcoming games, considering seasonal performance and updates such as injuries
- Offering scale and complexity: In 2015, and estimated 75M people played fantasy football, if we assume that NFL.com only had 5% market share (which I would bet an expensive cup of coffee is FAR too low an assumption), that’s about 3.75M users; many users log in many times a week, with heavy users logged in every day; these users build leagues that can have standard rules and calculations, or they can customize those rules and parameters
- Multi-device delivery: Users access the platform across any device, including phones, tablets, and personal computers (but never during work hours of course)
If we then consider the assets and equipment that make up a typical industrial facility, and consider it not from an operational perspective, but rather from a data perspective, what we realize is that it’s not actually that different. However, it’s still very rare to find a facility (particularly in oil and gas) that is leveraging analytics to anywhere near this degree. What’s being missed, as a result, is the insight that’s potentially available to help optimize operations and lower costs.
Now certainly, there are challenges that asset-heavy organizations face that fantasy football platforms don’t:
- Assets are very different from one another (whereas there is, for example, a limited universe of roles on a football team) and so differences exist in applying analytics across an entire facility (e.g., different types of sensors, different algorithms, etc.)
- Lots of facilities have been around for decades, and many are missing master and transactional data, not to mention sensors and infrastructure to capture data
- It’s not uncommon to not always know where all pieces of equipment are located (it’s somewhat – though admittedly not impossible – more difficult to misplace an NFL star)
Even with those challenges, I think that it’s a good example of what’s possible in a complex analytics scenario.
We could be doing more with the industrial internet of things, and while it certainly could be hard, it’s not impossible.
(I don’t
think I need to defend the value proposition of having a better understanding of asset performance, but if I do, it’s probably something related to decreasing the cycle times between an event occurring, an insight being realized, an action being taken, and the next event being measured.)
