It’s been quite a while since my Philadelphia 76ers have been relevant in the sports world. In the 12 years since Allen Iverson defiantly stepped over Tyron Lue and the mighty Lakers in Game 1 of the 2001 NBA finals, the Sixers got really bad, Iverson started “talking about practice,” and the Fresh Prince bought the team. Baseball’s Oakland A’s enlisted the kid from “Superbad” to use analytics to win games, and in an interesting change of direction, the Sixers followed suit by hiring Stanford MBA (and analytics guru) Sam Hinkie as G.M.
Since the Hinkie hiring, I’ve found myself reading more and more about the analytics movement in basketball, and everything I see keeps circling back to the same three themes: contextual data, strategy, and execution. Not surprisingly, these three themes surface in nearly every conversation I have with online organizations implementing a personalization and testing program.
Contextual data in the NBA means going beyond the stat sheet to find true indicators of value. For example, rather than simple points per game (which would likely overvalue players like Iverson, who shoot more often), we might look at “weighted +/-,” which ranks a player’s net output, accounting for his teammates’ performance, versus the league average player.
In an online world, contextual data means going beyond the performance of a simple A/B test (i.e., “did it win?”) to find more subtle metrics indicative of success. Specifically, today’s successful marketers are drilling down into their test results to evaluate the performance of discrete audience segments (i.e., “for whom did it win?”).
Running an A/B test simply to see if it “wins” for a majority of the audience is akin to an NBA coach determining that running the pick and roll results in success more often than not, and then running it on every play thereafter. Best-in-class organizations (á lá Amazon) are seeking to understand how different audiences actually behave in different situations, and then developing ways to validate and expand upon those learnings.
Which brings us to strategy. In the NBA, contextual data generally drives strategy in the long term (“how I build my team”). Analytically driven teams will ignore common conventions of player evaluation in favor of acquiring the “right” players over time that fit their given system.
Online, contextual data can play an incredibly powerful role in shaping brand and ecommerce strategy. A great example I heard recently was a CMO whose goal was to drive increased sales from repeat visitors, because he did not have to pay to acquire them via an inbound channel. Knowledge of what his repeat visitors have browsed, purchased, carted, or viewed but not purchased (contextual data) gives him a huge advantage in accomplishing that goal. Each of these segments is obviously quite different and must be treated as such (more on this later).
The application of contextual data to long-term strategy is rather intuitive, because of the presence of a time runway. However, a strategy is only as good as its execution on a day-by-day (or minute-by-minute basis), which can be challenging in highly dynamic (and competitive) environments.
To quote Boston Celtics Assistant G.M. Mike Zarren, “A coach might have 20 things he thinks will help win games. They might be able to work on six of those in practice, the players might actually remember four, and then you execute one in a game.” If Zarren’s critique is correct, then how many wins are foregone because teams cannot more readily execute on contextual data?
I ask marketers the same question all the time: How much are you missing out on if you are not able to execute on-site changes needed to serve your higher level strategy? Using the prior example of my CMO friend and his strategy of selling more to repeat visitors, the contextual data about his visitors is somewhat useless if he can not create concurrent, personalized on-site experiences aimed at serving each of them. A great strategy fizzles out because of an inability to execute.
Why does this all matter? In the NBA, there is a quantifiable goal: winning as many games as possible. In the same way, marketers need to look at every visitor to their site as a “game” and attempt to “win” as many as possible. My suggestions:
• Contextual data—Simple A/B tests (win/lose, test vs. control for all visitors) are not enough. Focus on different segments, and how they behave, across the lifetime of their buying cycle.
• Strategy—Where is your business, where does it need to go, and what tools will actually provide the return on your investment to get you there?
• Execution—You have to be able to change the website, quickly, to accommodate as many segments as possible to win them. If you are not, your competitors are… and you are losing.