I was shopping around for a doctor recently on my health insurer’s website; never mind what for exactly. This isn’t going to be that kind of story. Suffice it to say: Mom, if you’re reading this? Know that I am just fine.

The point is this: I live in Indianapolis, a major U.S. city rich in medical resources. I entered my criteria in the search boxes, selected a radius of five miles from my home, and still came up with hundreds of hits.

I’m not a guy who loves making phone calls; realistically, I figured I was willing to make about five. I checked out specialties, degrees, and where the degrees came from. My insurance does offer ratings for doctors—if they’ve received any—but most of them haven’t. Those who have received one, maybe two reviews: too small a number to be meaningful. I ran a few likely prospects through independent online physician review sites (like Yelp for docs) but hit the same snag.

We are living in the decade of data. We’re awash in the stuff. I go to sleep at night to Amazon or Netflix and wake up to Facebook. The Internet may know more about me than I know about myself. So why is choosing a doctor still so hard?

The answer (I think) is that a lot of this data I’m looking at is formless. It’s not been shaped in any truly intentional way. In some cases, like the doctor review websites, its formlessness is incidental: there aren’t enough reviews of the doctors I’m looking at to be meaningful, and even if there were, a lot of the criteria being assessed (things like wait times, difficulty getting an appointment, the comfort of the exam room) aren’t the things that are most important to me.

Data journalists are digging into all kinds of data like this. From reinterpreting medical studies whose conclusions were apparently flawed, to doing textual analysis on large-scale sets of documents like those obtained through Freedom of Information Act requests or from WikiLeaks, their investigations are not just startling, they’re indicative of both the potential and the limits of our newfound data bounty.

A recent article on Quartz is telling, I think. It teases out the apparent correlation between home values and Starbucks locations, one interesting finding being that Starbucks locations may be driving up home prices, rather than the other way around.

But perhaps a more important takeaway from that same article is this: as much as Starbucks has invested in the data analysis that drives their locating process, they admit it’s not all data driven. It’s art as well as science. They just try to make the science as consistent and reliable as possible, to cut down on error where they can.

A similar notion was advanced just recently by writer Tim Wu. Wu looked at programming decisions made by Netflix, whose chief content officer, Ted Sarandos, is a proponent of data-driven decision making. Kevin Spacey, star and executive producer of Netflix’s House of Cards, has said Sarandos ordered two whole seasons of his show because he was so convinced by the data analysis. Yet Wu, who has reported on Netflix before, senses that something more is going on.

Their “biggest bets,” he writes, “seemed ultimately driven by faith in a particular cult creator.” He advanced this theory to Sarandos himself at a Sundance Film Festival panel, and Sarandos conceded to the importance of judgment. He put the data to judgment mix at about seventy-thirty.

So you either come up with a hunch and then use your data to field-test it, or you use data to get you in the neighborhood, then start knocking on doors. In my own example, I can use the limited data I have to narrow the field of doctors, but ultimately I’ve got to meet them and make my own decision.

Do I have time to meet five doctors? I don’t know. It still seems like there should be a better way. And as I’m sitting here writing this story about data, it occurs to me that there is someone—someone I’ve already mentioned—who knows more about this list of doctors I’m looking at than almost anyone. Except God maybe, and the doctors themselves.

It’s my insurance company. Data central. They don’t just know all about me, they know a lot about these doctors, too. They know how many of their customers have been to see them, and for how long. They could tell me if a particular doctor has a large number of one-time patients, which might be a good indicator of a doctor who doesn’t inspire a lot of patient loyalty. They might even be able to tell me the kinds of issues patients were most referred for, which might help me make a decision about the best doctor for my case.

So why don’t they? There are a number of reasons why they mightn’t, ranging from the ethical to the practical. For one thing, there’s that last stretch of subjective interpretation—Sarandos’s thirty percent. It might be okay, even necessary, for a company to make a bet on buying a TV show, but for an insurer to make bets on the best doctors for its customers could be a liability risk, not to mention a conflict of interest.

In April of last year, Lisa Rosenbaum wrote a piece for the New Yorker about the Affordable Care Act, a vast public release of Medicare data, and how it might, and might not, lead to greater transparency in physician selection and care delivery. The data dump was pretty broad and hadn’t been well analyzed. One of the doctors in Rosenbaum’s article had been singled out by reporters as being the seventeenth-highest Medicare biller in the country. What wasn’t in the data was this: This doctor supervised an innovation and care project, sponsored by Medicaid, and was responsible for three hundred and eighty primary-care practices serving a million patients.

Rosenbaum concludes that while data has the potential to root out some flagrant waste and abuse, a direct connection between billings and quality care is difficult, especially when it comes to looking for the good things. And it will be particularly poor at looking for unquantifiable factors like bedside manner and promoting good health in the first place.

So where does that leave me? For now, right where I started. Making calls and going on hunches. But the data is out there and so, increasingly, are the data crunchers. Sooner or later, someone is going to put together something that’s sort of useful. Will it be error-proof? No. Will I still have to make five phone calls? I sure hope not.

There’s opportunity, too, for doctors to close the gap. Rather than just exist as a name on your clinic’s website, why not tell us something about yourself? Your care philosophy, particular experiences, or strengths. Maybe even open a Facebook account. I found a bit of this on the Internet, but not enough. Perhaps because referrals are really where it’s at.

In the meanwhile, I guess, you’ll have to wait for my call.