Big Data Is Here To Stay. So Can We Use It To Make Recalls Actually Work?

Image courtesy of (me and the sysop)

Sometimes products are unsafe. From bacteria-filled food to shrapnel-shooting airbags, on occasion even the most conscientious company will find itself needing to recall a product if it turns out to be harmful to consumers. But recalls are a big pain in the butt all around. One of the biggest issues? Actually letting consumers know that the stuff in their hands or on their shelves has, in fact, actually been recalled.

The key challenges with recalls are timeliness and effectiveness. In other words, you need a recall announcement to be made as quickly as possible, in order to prevent the problem item from causing more harms — but you also need to be accurate and to reach as many people as possible.

It’s a thorny problem. Regulatory agencies, and the companies that sell affected goods, can put out press releases until they’re blue in the face but that doesn’t mean anybody reads them or takes away the right information from them. How often do you visit the website of your 8-year-old car’s manufacturer? Hang out on the CPSC’s site? Read every single fine-print notice on the bulletin board near customer service at your grocery store? For most of us, it’s “never.” Even a regular recall roundup from your favorite consumer news site can’t capture everything.

So how do you get information about faulty or dangerous products to the people who need to know, and make sure they find out?

A panel of safety experts, regulators, and advocates discussed the challenges of product recalls at a consumer event in Washington, D.C. late last week. And the most popular solution, among advocates, was simultaneously dead simple and incredibly complex: use all the data we’ve already got.

Sometimes, it already works…

One retailer that’s already had great success using data for smart recalls? Costco. Jennifer Thompson, an executive with the warehouse chain, explained how the company can use its customer data to reach consumers swiftly and effectively. Because Costco requires a membership, all their customers have some sort of registration information — address, but usually also an e-mail address or even a mobile number for texting — on file, and the retailer can easily reach them.

Having robust databases of information not only lets Costco see who to send recall notices to, Thompson explained, but also lets them know how effective those notices are. They can measure the rate of mail returned undeliverable to the lists of mailing addresses they use, and they can see how many e-mail or text notices bounce. They can also track how many members who received recall notices actually bring the item back to the store.

If it sounds like Costco uses a potentially scary level of data, well, they do. But at least as a membership store, consumers know they’re signing up and handing over information. Even so, here’s the thing about that: Costco is far from the only retail chain that collects the information. They just admit they have it all, and make use of it not only for marketing purposes but also for making recalls fast and effective.

…so let’s expand on that

If you’re like many American adults, you have a half-dozen loyalty and rewards cards dangling from your keychain right now. Grocery stores, pharmacies, even more targeted retailers like pet stores or IKEA: everyone’s got some kind of loyalty or rewards program they beg customers to join. Even if you don’t have the physical card, you have an account and you give the cashier at each store your phone number to get sale discounts.

That data is all captured. That’s how you get targeted coupons, both digital and paper. Because you buy cat food, you end up with “special offers” for cat litter. Because you bought baby cereal, you’re suddenly inundated with coupons for Gerber everything. Your purchase history is there anyway. So why not leverage that information, and specifically target the consumers who bought that item with timely recall notices?

Jean Halloran, a food safety advocate with Consumers Union (the advocacy branch of Consumerist’s parent company, Consumer Reports) praised the potential of big data to help consumers. Targeted advertising, she pointed out, is everywhere. Keywords in GMail, or any search history or tangential topic discussion on Facebook, brings up ads targeted to the specific individual the algorithms think you are.

So why not put recall ads in those spaces, as a public service? If you mention in an e-mail, “I got sick after dinner,” perhaps the information about a chicken recall should show up. If you’ve “liked” your region’s dominant grocery chain on Facebook, why not see a sidebar ad for a product recall at that store?

Kids’ products are the hardest

Another advocate, Nancy Cowles from Kids In Danger, spoke to the particular challenges of children’s product safety recalls. Sure, every carseat, crib, and stroller comes with a registration card but even if you do remember to fill it out and send it back, there’s no guarantee that you won’t move six months later, making your old contact information useless.

KID, Cowles reported, finds that on average less than 5% of all children’s products that get recalled are ever returned or fixed. That means 95% of all the dangerous things out there that could hurt kids are still being sold and used. So there’s plenty of room for improvement on reaching parents and caregivers.

What if you could use an app to take a photo of a product you buy, Cowles suggested, and have it automatically registered to an account, so that you can get push notifications in the event of a recall? Or what if other participants in the retail chain participated? Your credit or debit card company knows if you have a transaction from Babies R’ Us. What if, in the event of a recall, the recalling company could get that mailing list to reach out to customers who probably bought the items?

It doesn’t even have to be that targeted, Cowles pointed out: in a 2014 study, KID found that 75% of the companies that recalled products have a Facebook presence, but only a quarter of that 75% ever once mentioned product recalls or other safety notices on their Facebook pages. Just basic use of social media, not even targeted, privacy-challenging use, could drastically improve the reach of recall notices.

But cars are low-hanging fruit

In other product types, there are even easier, less potentially nefarious data sources that could be readily mined for safety reasons. Clarence Ditlow, from the Center for Auto Safety, repeatedly pointed out to the panel how effective a simple data query could be for vehicle safety recalls.

All cars — new and used — are registered. You can easily track them. A federal or state-level regulator could query DMV databases and find out exactly how many, say, 2005-2010 Chevy Cobalts are registered in each state, and would immediately have the home addresses of all those cars’ drivers to make a mailing list for a critical recall notice. It’s easy for a used car on its second or third owner to fall off the manufacturer’s radar, because Honda or Ford or whoever only made money on the first sale.

But any car on the road will be registered. And registration isn’t just something you do once: you have to renew it every year or two. So why not, Ditlow asked, run a car’s VIN through a recall search before issuing a registration renewal? The DMV could, in that case, notify the consumer that they need to take their car in to a dealer for free repairs.

Or, he suggested, insurance companies could do it. Car insurance is mandatory in most states, and usually has a 6- or 12-month renewal cycle. Insurers could do a VIN check for recalls at the time of renewal and request or require that their customers get their cars up to date. After all: a safer car can mean an accident is less likely, or, if one does occur, that its harms will be mitigated. Fewer accidents with less injury means money saved, for an insurance company. And that’s just good business.

So can we make it happen?

The sheer depth and breadth of the information gathered about every consumer is staggering. But the ability to track someone’s every purchase both online and off is a pretty new innovation, and the data’s still scattered.

There are private data brokers who have some information, and not all. Different companies sell, trade, and hold different levels of data. Making sure that data is both accurate and current is a separate set of challenges all its own.

On top of that, recalls are all handled differently depending on type. Food recalls, handled through the USDA and the FDA both, take different approaches than vehicle or car parts recalls, which are handled through NHTSA, and other product recalls, which are handled through the CPSC. Each agency has a different level of staff, resources, and authority, and a different level of ability to work with companies that have products that need recalling.

Consumers are rightly worried about the sheer volume of unregulated personal data out there in the world. But if it’s going to be there anyway — and it is, even if what companies are allowed to do with it changes — businesses and regulators might as well work together to use it for good.