Could be worse. I recently called my doctor for a refill on a prescription, left a message for the nurse. Before I received a call back, I got a text message from GoodRX with a link about the medication, following the link (I anonymized it) took me to a page about the medication and advertising about where I could get it for the best price.
Of course, this made no sense to me, as I deny sharing of my medical information with anyone. So I contacted the office, went through a couple of phone trees and spoke with a few different people. None of them could see the issue with what happened. Then I explained a phone number is PII, and tying it to a medication seemed like a significant HIPPA issue.
Then things changed, it took a few hours to hear the whole story where a pharmaceutical rep had apparently told the nurses/office staff about this great way to get prescription drug information to patents by simply entering the medication and the patient's phone number in an App...
I’ve been told this is rectified, but the reality is I wish there was more I could do. I can’t imagine how many patients have been affected by this and thought nothing of it.
I feared this story would end with some version of "no one could tell me how I received this message". So, I suppose it's heartening to hear that you were able to resolve it, and relatively quickly.
But, I do think that's where we're headed fast: with so many interconnections, "affiliate" side-deals, opt-out limitations, etc., we'll soon not be able to manage the byzantine trail of entities that have access to our data in a given scenario. Our efforts to even understand who has our data and how, let alone control access, will be near-useless.
1. Geofencing is not very accurate. The range I’ve seen is ~100 yard accuracy. This is potentially a colossal waste of ad spend for lawyers because they get some people in the waiting room, but they also get every single device in the building, parking, and surrounding area.
2. They’re called device ids not phone ids. Proof right there the author didn’t research effectively.
3. The doctors/etc who have to abide by hiipa are explicitly requires to have their vendors (ex marketers) follow the same laws as them. This goes for any industry, but is written in the law for health.
Geofences are awesome for things like sports arenas, neighborhoods, malls, etc. but ineffective for anything smaller.
Edit: For #1 here are two sources. This is coming from a vendor that only sells location data, so their accuracy number of 93 feet is definitely not 100% truthful.
I'm not sure I've ever been to a hospital with anything non-hospital related within a 100 yard radius. And you can always adjust the center of your geotargeted area to make it more effective.
In an urban setting, sure, but most hospitals I've been to are MASSIVE, sprawling facilities.
Seems like the parking lot is close enough to being in the ER anyway.
Regardless, the 'waste' will be more from targeting health care workers, janitors, visitors, etc, but the targeting will still be more effective than not doing it at all.
To expand on that, say you want to market to only people within the ER room itself because you are an attorney for personal injury.
You will most likely get everyone who’s phone was within the overall geofence +100yards or so (sometimes better sometimes worse, really depends on what and how you are measuring).
That will include everyone above and below you with good enough signal to get through. Hospitals are usually tall in urban centers, so 5-10 stories of people?
Now you have your potential list. However, the real segment you just made are hospital visitors + employees/vendors.
Maybe you can exclude those that you see a lot or on a regular cycle as staff/vendors.
You are still not very close to just “personal injury” data.
Usually, you’ll need to combine this with something more to make better segment overall and go from there.
Unfortunately, data vendors will often market the above without much cleaning as “ER visitors” or “people with health problem X”.
There are numerous threads on HN about how bad some of these segments are - people tagged with mutually exclusive segments, wrong info, etc. That’s why FB and Google, and to a lesser extent for now Amazon, are sitting on some really valuable data - we tend to tell them explicitly and/or they have fine-grained location on us.
And let’s not forget that stuff like this has spurred responses like GDPR.
The list doesn't need to be perfect - hell, it doesn't even need to be very good at all. It just needs to convert better than not using the list.
(I'm equal parts horrified and impressed that someone's actually doing this... This is why, as much as it's annoying at work, I really like the GDPR...)
True, but then again Google/FB/etc offer better access to a more targetted audience at a reasonable price.
I’d consider using something like this if I had already deployed marketing capital elsewhere and hit diminishing returns. These are more akin to gambling in my book.
Not to say that polygon-based geofencing doesn’t work. It does, but with larger, less dense locations where the lower fidelity is good enough.
> True, but then again Google/FB/etc offer better access to a more targetted audience at a reasonable price.
I'm not so sure in this specific case. If I were a (cue pejorative stereotype for emphasis) scumbag ambulance chasing lawyer - I doubt there's anything in Facebook or Google's ad targeting demographics that'd give me quite the same opportunities as a rough-and-ready demographic of "people at or close to hospital emergency departments" to try selling to...
I hope (but am prepared to be shown otherwise) than neither Facebook nor Google have an easy way to target ads to people who've just posted "Faaark! Just been in a car accident! #YOLO #GladIBoughtAVolvo #ShouldntHaveBeenTexting"
(Evil-idea: I wonder if anyone's using siome combinations of accelerometer and GPS data to detect potentially traumatic events and selling that data on?)
Tie into insurance safe driver programs that in your car... Hell tie into the car its self soon... With all the car brand API's coming out. I know it could be done with Tesla, and Mercedes....
Yeah that was my point, you're getting so many employees and unrelated devices in that tiny area. Compared to at a stand-alone box store (home depot) you'd get like 70% shoppers, but the hospital is probably on 50% patients, and few of those are in the emergency room.
> I'm not sure I've ever been to a hospital with anything non-hospital related within a 100 yard radius. And you can always adjust the center of your geotargeted area to make it more effective.
This is somewhat true of my local hospital, but the ER facility is dwarfed by the number of non-ER facilities in a 100-200 yard radius. I personally see two specialists in the facility, receive diagnostic imaging and bloodwork, plus physical therapy within that same area. I additionally drive every day through that radius as I live on the same street as my hospital.
I dont want or need these sorts of ads cropping up.
> I'm not sure I've ever been to a hospital with anything non-hospital related within a 100 yard radius. And you can always adjust the center of your geotargeted area to make it more effective.
The large local hospital in Redmond WA (Evergreen) is right next to a strip mall and shares a parking lot with a movie theater.
I had a former sales person who worked for the Berry company (who produced the Yellow Pages) tell me the two biggest buyers of listings/ads were doctors and lawyers. So not surprised lawyers may be willing to overspend a bit compared to how much those very prominent ads for lawyers were in phone books (think rear cover or side binding). I assume they still spend quite a bit on the Yellow pages since that may be all that is available to you while you sit in jail and need legal council.
Last time I was in jail there was nothing with words in my cell except my booking papers. Calls placed out of that jail were collect and didn't show a caller ID. Thankfully I got the number of a bail bondsman from a man who was in there for battery. YellowPages would have been nice.
>1. Geofencing is not very accurate. The range I’ve seen is ~100 yard accuracy. This is potentially a colossal waste of ad spend for lawyers because they get some people in the waiting room, but they also get every single device in the building, parking, and surrounding area.
Sure, but advertising is a total crap shoot anyways. How well do you think those billboards on the highway are targeted? Anything that ups your chances of getting in front of your target audience is a huge win. This is Facebook's entire business model.
That’s not really true at all. I run digital ad campaigns for a living. Pretty much everything digital (that is being run by a competent agency) is extremely targeted. You can definitely go to a site and fine poorly targeted ads, but there are legitimate business reasons for those to exist.
If built correctly I can target exactly who I want to with minimal effort. Takes some time to prune and optimize to get it there though.
2. They’re called device ids not phone ids. Proof right there the author didn’t research effectively.
Not. NPR is the mainest of mainstream news and radio-originated to boot, so their reporting is commonly simplified to an extreme degree for easy comprehension by a general audience who may well be engaged in other tasks like driving.
> This is potentially a colossal waste of ad spend for lawyers because they get some people in the waiting room, but they also get every single device in the building, parking, and surrounding area.
A 10-50% chance of successfully getting an ad in front of an active patient is tremendously higher than any other advertising method available to them, I'd suspect.
Having seen the sort of incorrect categories Facebook slotted me into via their data broker partnerships, I wouldn't expect that to be anywhere near 100% accuracy.
I've discovered that Google, Amazon and Netflix all slot me into categories based on searches I've done and items I've looked at in the past. They don't seem to understand the difference between gag searches and a legitimate desire to buy something, which is how Amazon kept trying to sell me a 20-gallon drum of personal lubricant for so long.
That's amazingly accurate, this is not for navigation purposes within the hospital but for advertising on the hospital grounds (typically a substantial multiple of 100 yards). It's not like you want to direct someone to a certain wing, though it is probably even accurate enough for that .
Besides, for many phones it is much closer than that, because location services are enabled for instance for Google (because someone used Google maps) or because some app has linked an SDK that lifts the location data from the phone and moves it to some server.
Geo-fencing might not be terribly accurate but it gets you close. If combined with Bluetooth beacons (any chance the hospital is willing to install one for a fee?) or WiFi network recognition (I see the ER's WiFi network -- I am almost certainly inside the hospital) you can get the accuracy down to a few feet which is more than good enough for advertising company to sell that phone's position to the law firms.
Getting the right personal injury case is like hitting the lottery. Lawyers are willing to spend big on advertising to be in the right place at the right time for the right case.
It's the same as any type of advertising. As long as they produce more than they cost, it's a no brainer.
> This is potentially a colossal waste of ad spend for lawyers because they get some people in the waiting room, but they also get every single device in the building, parking, and surrounding area.
Is there a competing alternative that does any better? I wouldn't count on shady would-be advertisers to wait for perfect targeting solutions, if I were you (or else there'd be basically no advertising anywhere!).
How this likely works is the geofence is one of tens of different tactics that attempt to reach the audience who will soon need a lawyer. No way to tell unless you work for the ad agency, but from the outside this doesn't seem like an effective tactic to me.
Do you see anything else wrong with the story in the article other than perhaps the advertising might not be targeted enough?
Are you concerned that it may not be the right thing to do even if it is technically legal? I am guessing you work in adtech, I am wondering if you or your workplace has a model to address those kinds of questions?
Keep in mind that this type of correlation is baked into the system. Lawyer hires an ad agency to get the word out. Agency does SEO on inbound clicks. Finds that there are geographical clusters where a lot of searches originate from. Agency responsibly increases bids on ads for devices in those places. The agency could take the next step and identify what those places are and why they're popular with people looking for lawyers, but what would be the point? All they're interested in is maximizing the value of every dollar they spend on advertising and if the data says do it that's what you do.
The book "Weapons of Math Destruction" goes into some detail on how these models are being used to exploit people when in some cases they could be used to help them.
Here in Argentina my mother had a car accident. I received a call from the hospital. I went to the hospital and talked with the guy that called me. He was dressed like a nurse. He gave me the news about her accident and that her friend died in the accident. Then he gave me his card. He was a lawer.
No. I was overwhelmed by the whole situation and public hospitals here are kind of crazy. On the positive side they're public and if anyone (citizen or not) needs medical aid it's free up to a certain point (they're not going to give you a peacemaker for free but the surgery is free).
The problem is that folks want either 'no regulation' or 'more regulation'. Neither is the solution.
We need regulation that targets specific problems, nothing more and nothing less. Instead we get regulation that may loosely address the problem it was originally drafted to address, but has since grown into a mess of compromises added at the request of lawyers, lobbyists, and uninformed politicians.
I'll start:
"You cannot send advertising to people when you suspect (e.g. by geofence) they are suffering from a medical emergency."
> "You cannot send advertising to people when you suspect (e.g. by geofence) they are suffering from a medical emergency."
Doesn't this risk making things worse? If you advertising based on a metric that could lead to that conclusion (even if it isn't your intent), you might feel obligated to geolocate people to avoid being accused of breaching it. The loose wording could also lead to companies trying to exploit loopholes - "we weren't advertising to people suffering a medical emergency, most of whom _aren't_ suffering a medical emergency. We had no reason to suspect _they_ were suffering a medical emergency.". You either need to be stricter (no advertising specifically targeted at hospitals) or specify precisely what it means to "suspect".
That doesn't seem to help anything. Wouldn't this be better:
> Personal data held by a hospital about their patients may not be used for any non-medical purpose, and may not be shared with any third parties unless directly necessary for a medical purpose.
Neither do I, I don’t want to overstretch but I’ve seen a lot of blowback on GDPR lately and actions like this are what I think will get the US there eventually too.
Correct me if I'm wrong, but there's got to be at least one app on the person's phone that's reporting GPS info that is feeding information to the marketing company. Is the name of the app(s) mentioned?
i doubt it. do carriers really sell advertising directly to their clients? ie. does the company have a advertising division where you can spam SMS to their clients within a certain radius?
Actually, yes. I was pitched by ad reps from the largest carrier - they claimed to be able to send an SMS when one of their customers enters a geofenced area.
In a population centre you can probably map to 50m accuracy or less based on the wifi networks one can see - can apps report back the specific networks? Isn't that kinda how Google's enhanced "location services" works?
In America, don't victims of injury caused by other people have to sue to get compensation for their medical bills if they're under-insured? That would mean it's quite helpful to have lawyers offering their services right when you need it. In my country, hospitals will pressure you into taking money from the state accident compensation service when you're injured. So what's the difference? Patients are being informed of the assistance that's available to them.
It's weird that targeted advertising is so widely criticized now, but when it was made popular by Google a couple of decades ago (we'll read your email and offer you things that you'll like!), it was hailed as an improvement over the blanket ads that were polluting the internet and of no use to just about anybody. I still think targeted ads are fantastic. Advertisers are doing the leg-work of informing people about how to buy what they want, instead of just cars, shoes, and travel which is what we used to see.
If it were only compensation for their medical bills that would probably be OK. But the way it really works is that the the typical personal injury lawyer takes 30%-50% of the award as a fee, so they are highly incentivized to inflate the damages, often with "pain and suffering" claims that money really doesn't compensate for and are hard to disprove.
Surely the solution to that is not having fewer victims sue the people responsible? Sure, that would reduce the total pain and suffering payments, but in an unfair way by giving it all to a few who get lawyers while the others get nothing, not even the cost of medical bills.
If the lawyers were smart, they wouldn't use geo fencing--they'd use IP Targeting instead. IP Targeting allows you to target an IP address that's been matched to a physical street address. All they have to do is set up IP Targeting and target the hospital's network. Anyone who uses the hospital wifi would see the lawyer's ads.
Only if you connect to the hospitals wifi, to be contrary, a marketer may want to exclude the hospital network from that geo fence so that they do not target doctors/nurses/admins.
In many cases the doctors/nurses/admins and employees of the hospital typically connect to another network, a password-protected network. So probably not an issue.
An IP only targeting strategy limits the amount of people a lawyer may target to those who are on the hospital wifi, which will be a smaller audience. There's also a higher risk that the people on the hospital wifi and IP-range work at the hospital.
While it will be good for known IP ranges to be used for targeting, it should be an additional tactic, not the only one. Geo-targeted ad-buys within an area where there may be higher concentration of hospitals, layered on with additional targeting parameters, will be the optimal setup.
Been using El Toro (el toro dot com) for years now, they have a really good system for matching IPs with physical street addresses. So you can give them a list of addresses and they'll tell you which of them they have IPs for.
I agree, it's their ad agencies/SEM agencies who are using geo fencing, which isn't ideal. If the marketing agency was more advanced, they'd be using IP Targeting instead.
Have used IP Targeting to target conferences/events and several hotels around the conference and it worked really well.
It doesn't take much imagination to use geofencing for lots of sleezy advertising. Like lottery, gambling and loan shark ads around pawn shops and food banks.
Also, I imagine that televangelists who want a shiny new airplane probably target low-income neighborhoods with their free enchanted water 'guaranteed to bring wealth and prosperity'.
The article doesn't say where the advertisers get the geolocation data -- is that from the data that the carrier is selling, or is it coming from some app(s) installed on the phone?
> is that from the data that the carrier is selling, or is it coming from some app(s) installed on the phone?
Could be both, could be some website you viewed, could be google maps, anything that shares your location. Location data is pretty valuable to advertisers, OpenRTB supports it explicitly during the real time bidding process.
All manner of cleaning is applied to the data to ensure that it isn't faked or useless before the bidding starts.
Probably the real value here is that if you can hook them in the ER right after something terrible happened, they will have a very hard time being rational and walking away later. If you get to them at any other time, they will be vastly more rational. But planting this idea in their head at this critical time will serve to essentially make many people a captive audience who cannot escape later when it seems they would be more able to think rationally.
The first step towards AD supported healthcare. Watch ads in the ER - accrue $ towards your bill (which is silently and automatically padded to make sure that any 'money' you make is sucked right back out). Or maybe Pfizer will buy ad - space on the hospital guest LAN allowing more dollars to flow into the hospital execs pockets - essentially allowing to profit over the patient's illness and misery.
Interesting, at one point it seems like there may have been technical reasons for that [1], but not any longer. I've had to wait in hospital waiting rooms for hours, and not being able to use a phone/laptop _at all_ would kind of suck.
It isn't all about personal injury. Hospitals are where life-changing things happen. They are where many people are born and die. People coming in and out of hospitals, be them patients or family members, may need all sorts of legal services, wills and estate planning being top of the list. Not every lawyer advertising near hospitals is an ambulance-chasing leach.
I'm not sure that makes it any better - targeting the bereaved or dying (for wills and estate planning), whilst they are still at the hospital, feels potentially more tasteless than targeting people for personal injury claims.
>> targeting the bereaved or dying (for wills and estate planning)
Or the young parents who want to add their new kid to their will? Or the very ill person who realizes that their will/life insurance leaves everything to their ex-wife, not their current family? Getting that changed asap will certainly be a relief to them. Or the parents who's kid was just brought to the hospital by the cops and need someone to talk to about it?
There are times and places where lawyers do good things, they just aren't as sexy as the evil things.
Of course, this made no sense to me, as I deny sharing of my medical information with anyone. So I contacted the office, went through a couple of phone trees and spoke with a few different people. None of them could see the issue with what happened. Then I explained a phone number is PII, and tying it to a medication seemed like a significant HIPPA issue.
Then things changed, it took a few hours to hear the whole story where a pharmaceutical rep had apparently told the nurses/office staff about this great way to get prescription drug information to patents by simply entering the medication and the patient's phone number in an App...
I’ve been told this is rectified, but the reality is I wish there was more I could do. I can’t imagine how many patients have been affected by this and thought nothing of it.