Smart Disclosure

You Should Own Your Usage Data

Have you noticed that music and video streaming services have gotten pretty good at predicting what you might like? It can almost feel like they know you better than you know yourself. But here’s the thing: in some ways they do!

Consider Netflix in its DVD-by-mail days. It asked customers to rate movies and shows to help predict their tastes. Viewing history offered another source of information: did someone finish Breaking Bad or stop after three episodes? Did they watch occasionally or binge through a rainy weekend? A service can keep records that a customer may not remember. We humans can sometimes get all the cop shows confused.

There is a general point here. Netflix has possession of something valuable, which is your own past behavior. That gives them an advantage over any competitor that is starting fresh. Obviously, Netflix owns whatever algorithms it has created in order to convert your viewing history into useful recommendations. It need not (and should not) be required to share that bit of intellectual property. But shouldn’t you have a right to make use of your past viewing behavior as well?

This question sits at the heart of a broader proposal that we once gave the clunky name RECAP and that was later rebranded, with help from the Obama administration, as Smart Disclosure. A White House document defines the idea this way:

“smart disclosure” refers to the timely release of complex information and data in standardized, machine readable formats in ways that enable consumers to make informed decisions. Smart Disclosure will typically take the form of providing individual consumers of goods and services with direct access to relevant information and data sets. Such information might involve, for example, the range of costs associated with various products and services, including costs that might not otherwise be transparent. In some cases, agencies or third-party intermediaries may also create tools that use these data sets to provide services that simplify consumer decision-making.

The proposal has two components. First, complex product information should be available in a standardized, machine-readable format. Second, organizations that record people’s purchases or usage should generally give those people access to their records. The authors describe this as ownership of purchase history; the practical proposal is access and portability. Most consumers would not need to inspect the files themselves. Software could use them to compare options.

Consider the parents of a child with a nut allergy. Suppose they do most of their shopping at a large supermarket in their neighborhood. If it is part of a chain, there is a good chance this supermarket has some kind of “shopper’s club” that keeps track of what each household buys. We think that parents—and all other customers—should have the right to access that shopping history, including from online shopping services. Why might they want that data? Well, suppose that with just a couple of clicks, they could download all their purchases over the past six months in a format that allows them to be uploaded to a website that offered to search through all their purchases and find the ones that they might want to take off their future shopping lists (on the basis of specified criteria, such as nuts, high calories, lots of sugar, or high prices). The website might even recommend good substitutes.

The proposed No-Nuts website would need accurate, current ingredient records and a reliable way to match them to purchases. Machine-readable ingredient lists and product identifiers could make this task easier. The resulting tool could flag products for review and suggest alternatives to investigate. Its output would depend on the completeness of the records; matching a purchase to an ingredient list would not by itself establish allergy safety.

Similarly, shopping for a new smartphone and accompanying plan is a perfect example of a choice that cries out for Smart Disclosure. For most of us, the only aspects of the phone decision about which we feel even remotely confident are its color and size, because we can do a physical inspection. After that we would be flummoxed. How many GB of storage do we need? What is a GB anyway? What in the world does “1792-by-828-pixel resolution at 326 ppi” mean? Choosing a plan is even worse. How many calls or texts will I make? What will it cost to use when I travel? How much data do I use, and will that change if I get this new phone? Yikes! Can I just go back to choosing the color? Here, the second component of Smart Disclosure would help enormously. A choice engine that had access both to all the pricing features of a plan and to the consumer’s past usage data could make tailored recommendations. A choice engine could also estimate how usage might change after an upgrade, while showing the assumptions behind that estimate. If consumers gave the choice engine continued access to their usage, they could also receive suggestions on switching to a new plan later on.

The authors describe banking and energy initiatives in the United Kingdom as efforts to put parts of this proposal into practice. The examples below reflect the period covered by their account.

Regular payments for rent, utilities, and credit cards can make switching banks feel like a chore. Just reading about setting them up again makes Sunstein break into a sweat. The authors describe UK Open Banking reforms introduced from 2018, allowing customers to authorize data sharing with regulated providers. At the time of their account, the nine biggest banks and building societies were enrolled in the Open Banking Directory, with others expected to follow. They cite Lumio as an example of bringing financial accounts together. The point is that authorized access can let another service work with data held by the original provider.

The other UK example concerns energy comparison. The authors describe suppliers making customers’ usage data available through QR codes on bills, allowing comparison services to apply competing prices to that usage. They cite Uswitch as one such service. Having data in a common format reduces the work of making a comparison; the quality of the result still depends on the offers covered and the assumptions used.

More like that, please!