Gamified Analytics: Unlocking Disruptive Genius or Disrupting Data Quality?

Big Data Evangelist, IBM

Games are fun, and fun can be a powerful force for creative genius, superhuman productivity, and, of course, personal satisfaction.

Gamification is in vogue in online user-interface design these days. The term, defined here by Wikipedia, refers to online environments that incorporate systems of engagement, exploration, challenge, competition, and incentive to encourage user behaviors that produce desirable outcomes.

Recently, there have been online discussions of gamification in a business analytics context, such as this article from an Ovum analyst.

Is it wise or foolish to indulge in discussions of "gamification" where business analytics applications are concerned?

Be careful. You don't want to gamify the governance of data and analytics. The responsibilities, roles, and business processes must be cut-and-dried and not subject to stakeholder whim and improvisation. You absolutely must not disrupt your "single version of the truth" by treating data stewardship as a game that might lead to failure. Failure is not an option where data quality is at stake.

For the same reason, you wouldn't gamify most core business applications--such as finance, customer support, quality assurance, supply chain management--that depend on official records, structured workflows, and strict compliance. Business integrity depends on maintenance of top-down, repeatable, auditable controls on the data, transactions, analytics, orchestrations, and rules that drive these processes. These controls are designed produce assured business outcomes. These are core duties, not games. The staff incentives for guaranteed success are stark: enforce these processes and outcomes or lose your jobs.

However, you might want to gamify interactive data exploration, scenario modeling, and real-world experimentation within your data science efforts. These activities are speculative, in the sense their success is not guaranteed and the sequence of activities needed to produce success are not well-understood in advance. Success on many data-science initiatives may be due mostly to the dogged work of very smart individuals using power tools, in which case personal curiosity and professional pride may be the most important incentives. But gamified work environments--in which individual and team collaborators compete for cash prizes (a la the Netflix Challenge or Kaggle data-science contests)--might provide just enough incentive for smart people to smart people to focus even more creatively on a fresh approach.

In leading-edge data-science challenges, perhaps what's needed to unlock the disruptive "eureka" moment in a team of data-science geniuses is a multi-user game designed to put them slightly off-balance, short-circuit their old thought patterns, and encourage out-of-box thinking. This is especially necessary when a project has a core group of data scientists from a common discipline--such as marketing campaign analytics--who all take the tried-and-true modeling paradigms as the default for new projects, even when they are not entirely appropriate to some new challenge. For example, an online game might pit various behavioral scientists against each other, with financial incentives riding on their ability to write algorithms that discriminate fine-grained customer sentiment from the tsunami of noise emanating from social networks, clickstreams, and smartphones.

And you might, within bounds, find value in gamifying some analytics-driven customer service, marketing. and back-office functions when those processes require some ad-hoc, dynamic, and situational human judgment. For example, handling a churn-prone customer relationship in a multichannel environment might benefit from human touchpoints--call center, brick-and-mortar, direct sales, etc.--competing with each other to win "points" on their contribution to day-to-day metrics of customer loyalty, upsell, satisfaction, etc.

One last thought. As I discussed in this blog , you might also consider using game theory to model and optimize various customer-engagement scenarios. Keep in mind, though, that this doesn't necessarily lead to "gamification" of engagements: your channels and customers may not experience game-theory-optimized interactions as a game that they can on some level "win" (other than the usual desirable outcomes of customer loyalty, satisfaction, sales, etc.).

Game-playing can be a powerfully creative activity, as long as you keep it confined to a well-governed playing field.