Competing With Data & Analytics
What to Read Next
In just a few years, the sharing (or access or gig) economy is already casting a shadow over numerous industries. But while the “sharing” aspect is emphasized, data and analytics is critical to making the sharing actually work.
Yes, the basic idea of sharing is appealing and feels right. We’ve heard it promoted since we were toddlers with toys or cookies. As adults, the idea still resonates — why squander our unused resources that someone else can use (and give us money for!), if only we knew how to connect with a person who needs them? Not only are there potential monetary benefits, the idea of reduced waste (of time, natural resources, etc.) is enticing.
Travel and transportation are classic examples of this sharing approach, if the word “classic” can apply to anything so new. We can subsidize our costs for cars and homes by renting excess capacity; others get access to cars and homes at lower costs; platform companies match the people who have with the people who need.
Modern information and communication technologies make it cheap and easy to quickly make these matches. Everyone carries devices that let him or her connect with anyone.
But, wait — would you really want to connect with just anyone?
Probably not. Instead, in the sharing economy, “the secret sauce is trust and reputation.”
Without data, sharing would be a single-stage game. Each player would maximize only his or her value in each transaction. When renting out your house, why not exaggerate the view or amenities if it got you a higher price? When giving someone a ride, who cares if the seats are muddy? When renting someone else’s house, why bother taking care of anything? When riding in someone else’s car, why not smoke if you want to? As each player myopically maximizes his or her own value, the other player may bear the brunt of the consequences. Sharing, then, falls apart quickly.
Data, however, turns a single-stage game into a repeated, multi-stage game. When games transition from single-stage to repeated, suddenly players have consequences for how they play each round. The exaggerated view and amenities get reported for future prospective renters to see. The trashed house goes down on a nefarious renter’s permanent record.
As a result, accumulating data changes behavior in every round of the game. eBay knows this; the buyer and seller feedback there form reputations. Amazon knows this; the marketplace review system tries to incorporate consequences for each transaction. Craigslist doesn’t incorporate reputation, and perhaps this is part of its recent decline.
Furthermore, even if this data isn’t visible to everyone (such as the hidden driver ratings of passengers in Uber), everyone still knows the data is being recorded. Back in grade school, we couldn’t see our “permanent record,” but we feared its mysterious secrets.
Despite the potential for improvement, however, data does not perfectly eliminate bad actions. For example:
- Poor reputations can be discarded as long as acquiring a new reputation is quick, easy, and low cost. eBay has always struggled with users dropping one account and creating another if feedback is bad.
- New users can be caught in a catch-22, as they find it difficult to participate without reputation and no way to gain reputation without participation. New property listings on VRBO, for instance, are more uncertain until reviews are available.
- Undesirable behavior on one platform rarely leaks to other platforms. Bad drivers whose reputations have suffered on Uber can move to Lyft without penalty.
- Data can be strategically manipulated. Practically every platform tracking data about reputations deals with fraudulent practices.
Advancing analytics practices may be able to improve the multi-stage game both for sharing-economy organizations and organizations that could benefit from improved reputational effects.
So how do businesses use data to promote the good while weeding out the bad?
Try increasing your data’s transparency. For those organizations still treating each interaction as isolated, start by making data visible to potential customers (encouraging reputational spillover from past customers) and to existing customers (reinforcing positive by showing past interactions such as quick past response time). To be genuine, these should go beyond the ham-handed, cherry-picked results from satisfaction surveys. For example, on-time flight performance data displayed at the time of booking changes airline incentives.
Get savvier about identifying users. With analytical abilities to uncover customer behavior, reputations should be stickier. Simply registering a new user name should not be enough to trick your organization into thinking the underlying person is different.
Consider making reputation information visible so that reputation can spill over between sites. An opt-in approach may alleviate privacy concerns while at least allowing positive reputations to transfer.
We all know that data quality is important and frequently discussed — the trustworthiness of data directly relates to the value it can add to the organization. But data can also influence the trust we have in organizations, creating either a virtuous or a vicious cycle. Data can enhance organizations’ trustworthiness, allowing them to collect even better data — or it can undermine them. The managerial challenge is to make the cycle created by the repeated game work for, rather than against, the organization.