Rocket Readings: Here’s how smartphones, tablets and huge databases will upend market research

Here’s how smartphones, tablets and huge databases will upend market research:

from GigaOM by Derrick Harris

If you’re tired of those annoying 8 p.m. phone calls asking questions about where you shop, or of carrying an Arbitron sensor to provide radio ratings, your omnipresent smartphone or tablet might well turn out to be your savior. And all you have to do is give up a little privacy.
Our mobile devices are amazing at capturing real-world data — location, temperature, movement, sound — that just goes to waste if we don’t put it to use. It’s easy enough to get a personalized experience on the web, but these types of data might make it possible to get one in traditionally more-static places such as retail and radio as well. At the least, perhaps we can expect content, price tags and experiences that cater more to our actual tastes than those of station programmers and a fashion designer’s idea of what people should be willing to pay.

Location is the key to everything

Retailers already have a pretty good sense of what people are buying and even how they’re moving through stores, but they don’t really know where customers are going once they leave. This knowledge could be very useful, however: If you want to improve your store or figure out how to market your company, knowing what else your customers are up to could go a long way. This type of data is starting to become available thanks in part to a Seattle-based startup called Placed.
We’ve been covering Placed for about a year, since it launched its first product targeting developers interested in learning where users were accessing their applications and mobile sites. The company has since expanded its operations to include a Panels service that lets the company track around the clock, on behalf of paying businesses, the physical location of customers who have downloaded the app (usually in exchange for a small monetary reward). It also has its own Panels app, unaffiliated with commercial customers, that allows Placed to provide market data on the physical movements of some 70,000 consumers.
This week, the company released a report highlighting some national findings from the first quarter, including, for example, what departments stores are most popular with what demographics, what business categories experienced the most increases in traffic, and what businesses have the highest and lowest affinities (i.e., people who visit one also visit, or don’t visit, the other). If you’re willing to pay, Placed will tell you pretty much anything you want to know, founder and CEO David Shim told me, broken down by geographic region, business type, demographic, you name it.

Shim noted a couple of actual users and potential users that I think highlight why this type of data is so valuable. One is a high-end retail business that found out that while female millenials enter its stores a lot, they don’t buy a lot. Rather, the stores they visit next are usually discount retailers such as Burlington Coat Factory and Ross. The suggestion is clear: These shoppers want to see what’s hot and then buy a reasonable facsimile at a lower price.
He also noted that some Las Vegas casinos are interested in running their own Placed panels to figure out what restaurants their guests are eating at once they leave the casino grounds. Now, if casinos can figure out where else on the Strip people are spending their money, they can make better choices when it comes time to swap out their own restaurants and shops.
In both cases, it’s possible the answer to the question of how to get more of these customers’ money is to drop prices. If a 10 percent price reduction leads to a 14 percent increase in sales, that’s a win-win situation.

Rethinking radio

Location data becomes even more valuable when combined with other data, though, such as sound. Consider the implications of knowing not just what radio stations people are hearing — which is essentially how the Arbitron ratings work — but what songs they’re actually listening to. Just because you hear the Latino station for an hour at the taco shop during lunchtime or the top 40 station at the gym, that doesn’t mean you’re listening to them or like listening to them.
But the songs you choose to listen to in your car, for example, probably tell a lot about what you actually like. And the technology exists to determine that. Last month, I wrote about how Gracenote is able to use the internal microphones on tablets and smartphones to recognize the songs playing on people’s televisions or stereos. It can also detect reactions such as cheering or booing, and likely whether someone turns up the volume.

Arbitron’s Portable People Meter

Now, all of a sudden, one can envision a world in which programming managers at radio stations can figure out on a song-by-song (or artist-by-artist) basis what people are actually listening to, and when and where they’re listening. If all it involves is someone downloading an app, they can presumably do it at a larger scale than requiring people to wear special additional sensors or fill out a diary. Broadcast radio can never be as personalized as something like Pandora, but it could start sounding a lot more like what listeners would choose if left to their own devices.
Digital radio could get downright great, even better than what Pandora can currently offer. I might never add Disney theme songs or the Sesame Street favorites to my preferences list, but if that’s all I listen to when I’m in my car between 4:30 p.m. and 5 p.m. — and it is — maybe a service could hook me up with some new songs every day. If I’ve turned up the volume on a Taylor Swift song three times this week while I was at home, maybe I actually like it and want to hear more even if I won’t admit it.

Not just data, but good data

As great as all this might sound (it does to me), it’s the advent of big data that makes it all possible. Placed’s analytics are so accurate because it has special algorithms to determine where a person actually is — even if there are numerous options within a small area — and its models are constantly being trained. Shim said his company gets 15,000 responses a day to surveys asking Panels users whether it had them at the right location, and it has already validated 3.5 million of the the 13 billion locations in its database.
Gracenote, for its part, has audio and metadata for millions of songs that it keeps in memory so it can access them in a hurry for the sake of real-time recognition. It can group music into dozens of categories based on genre, artist, geography or even just how the songs sound. It wants to build an in-car system that can change songs based on driving conditions fed to the stereo from the car itself.
I acknowledge this all sounds a little creepy, but, ironically, it also sounds like the beginning of the end for some concerns over privacy. Heck, Shim said, about 500,000 people have already downloaded the Placed Panels app.

Really, it all comes down to value. If handing over a little bit of data actually provides value in return — in the form something better than just targeted ads — it appears people will be willing to do so. People tell Amazon about their purchases, let Google Now access their email and tell Placed which store they’re at out of five possibilities because they think they’re getting a worthwhile service in exchange.
The data-collection genie is already out of the bottle. Now it’s just a matter of making it work for us instead of at our expense.
Feature image courtesy of Shutterstock user Vadim Georgiev.