Apple, FedEx and the cookie apocalypse
ben-evans.com
ben-evans.com
Translation: Apple worded the question plainly and without deception.
However, I think that “track” sounds much more invasive than personalize. Of course we know they’re the same damn thing, but “personalized” sounds more innocent.
Serving personalized ads is the goal and tracking is the means.
https://webkit.org/blog/11529/introducing-private-click-meas...
From a pure privacy perspective, it's odd because it involves the browser collaborating in measuring the results of your behavior. I think it shares several of the concerns with Google's FLoC and "privacy sandbox" proposals.
It only makes sense if you buy into Apple's perspective that there's a level of "acceptable" tracking, and new tracking features under that threshold are acceptable if they give Apple the political capital to kill tracking technology that crosses that threshold. It's a bit much to swallow. But they are at least trying to address that concern.
> "App Name" would like permission to track you across apps and websites owned by other companies
> [Allow Tracking]
> [As App Not to Track]
And then also the app developer can put in their own small description that shows in addition to the above.
This seems incredibly fair and conscience. It is an objective statement of facts that is only biased because the reality is something that most people probably don't like.
I am not a native English speaker, but I am not sure about that. Look at the word "track" -- pretty much no one would want to be tracked, right?
I guess if they replace "track" with "identify", more people would actually allow that. Something like "Allow this app to identify your device across apps and websites". Average users usually don't have problem being identified. They use the same email across websites, and will prefer facebook login on 3rd party websites if it is an option...
End of day, the thing getting blocked by the prompt is iOS IDFA, where letters "ID" mean "identifier".
But still, I don't mind Apply play a little wording game here. The ad industry, on the other side, did way uglier than Apple in the last decade.
I mean if your asking for ID, they Why is always there: to prove age, to track, etc. So, calling it "track" in their (Apple) messaging is good. Its the Identifier to Track You.
Nice demonstration of apophasis there.
Regardless, how else should they frame it? The question seems incredibly concise and matter of fact. Indeed, the question simply claims that it will "ask" the app not to track, which seems pretty broad and permissive.
https://github.com/WICG/floc won't really let advertisers do that, this is what https://github.com/WICG/turtledove is for
(Disclosure: I work on ads at Google, speaking only for myself)
Check if you've accidentally enabled some auto-delete?
Other browsers are working on this general project ("how do you do ads without cookies") too: Edge has https://github.com/WICG/privacy-preserving-ads/blob/main/Par... and Safari has https://webkit.org/blog/8943/privacy-preserving-ad-click-att...
why?
a.example doesn't have the user's identifier yet so it redirects to identifier.example
identifier.example sets a new ID as a cookie and redirects back with the ID
This happens so fast the user doesn't notice and location.replace doesn't generate history entries so there's no record of it happening
a.example saves the ID in its storage
Now the user goes to b.example
b.example doesn't have the ID so it redirects to identifier.example
identifier.example sees the ID in the cookie set when a.example redirected through and redirects back with the ID
b.example now has the same ID as a.example
That's all 1st party cookies
This seems worth raising with browsers as a bug, since this should be easily detectable as an attempt to work around cross-site tracking restrictions?
You can still do this. Car ads on car websites. Watch ads on the Economist website.
Has there been any research on the effectiveness of retargeting?
1. Incrementality is extremely hard to measure due to noise and biases, in particular when trying to account for the multiple agents in the game. Attributed sales are easy to measure. For global, incremental sales, it is much harder.
2. Because of 1, this research needs large amounts of data, which only large ad tech companies have access to.
3. Because the topic is so hard to pin down and so sensitive business-wise, no private ad-tech company would publish ground-breaking material on the topic.
A deep introspection of advertisers, browsers, publishers. Users, the fundamental owners of the data we're talking about, seem to not even be at the table.
Personalized ads were always about showing you ads for diapers just after your child had outgrown them, or ads for a crappy car just after you had bought a good car. Meh.
I think the most interesting "ads are good" argument I've heard is being able to give free tools to people who don't have the discretionary budget to pay for things like premium news. A bit sad about information paywalls over this.
At the same time, I'd still gamble that this 'apocalypse' will prove to be created because there wasn't pressure to perform better with a limited ad capabilities. My bet would still be that things will be fine even after the playing board is tamed -- the markets just need some time to react and correct.
Is there evidence that blasting even very targeted ads at people without discretionary income yields conversion?
Morally, ethically and potentially even legally wrong (technically the ad network should be held complicit if they're promoting a scam), but it does work. "No discretionary income" doesn't mean the mark won't spend money if they believe they have a high probability of making it back + more.
But it also no longer is tge case. AdWords stopped working (at least for us) in April
Not quoting the question makes it easier for the author to frame the formulation as malicious without risking the reader coming to a different conclusion. Here's the question:
"Allow APP to track your activity across other companies' apps and websites?
Your data will be used to deliver personalized apps to you.
[Ask App Not to Track]
[Allow]"
(AFAICT, the second sentence can be different, e.g. "Your data will be used to measure advertising efficiency", depending on the app.)
We've had to start doing horrific things like searching potential customers through the directory and sending unsolicited emails
We used to rely on networks of contacts for B2B marketing. Maintained via junkets to conferences and retreats at exotic locations.
I get that all this information is great for targeting me with better ads, but I think there should be a limit to what data is collected and for how long it is kept.
Personally, I hope Congress (USA) creates some laws to allow consumers to opt out of all this tracking. If it is truly beneficial for users, then they will opt-in
In practice, Apple says they use the following data for targeting: search queries, news articles you've read, any content bought through their various store fronts, App Store searches, and app usage data and using it for targeting. Some of this is un-differentiable for privacy/needs to be stored server side, the Apple docs neither mention this distinction nor differential privacy at all, and there's no public indication any of this is covered by differential privacy
Done well, it's actually ok: you read the article looking for that last breadcrumb from the headline, and find it in the last paragraph with a sort of 'a-ha!' moment.
Here it's done badly, with some vague handwaving about how ad spend might go to Fedex for free returns.
As a consumer, if I were given a choice between intrusive cross-site tracking and free returns, the choice is completely obvious.
This is incorrect. Since the beginning of advertising there have been people who survey those who read a particular magazine, newspaper, watch a particular show. And for those who are willing to share those habits with advertisers PAY THEM MONEY for that. Ever heard the term "Neilson family"? That was a family that agreed to put a special box on their television which recorded which channels you watched, when, and for how long. It would then aggregate that information and send it to advertisers and broadcasters alike. Based on who the "audience" was, they would negotiate advertising rights. And from the very beginning broadcasters try to "game" the results with known desirable content during "sweeps weeks" which were the weeks where the statistics were gathered.
Nearly every periodical I ever subscribed too occasionally sent a "marketing survey" with some compensation for filling it out like an extra year of subscription or a branded bag or something. There were, and ARE, periodicals today where the only "subscription price" you pay is giving up your personal data about why you want to read that periodical.
So the "apocolypse" here, is that these platform owners found a tasty revenue stream by disintermediating all of the surveyors and poll takers and returning statistics with 100% participation. No modelling necessary, here are the exact characteristics of your readers.
They didn't give their users a choice, they just took that private data and sold it.
And now they are whining, loudly, that OMG that value we were stealing? They want us to pay for that? That could ruin us!
Newsflash, it won't ruin you. Does this mean the end of "free" journalism? Maybe, but it was never free to begin with, it only felt free.
The inferring is still just statistical aggregation. Readers of the Economist may be on average wealthier, but a poor college student may still be reading it and seeing the ads for expensive watches. Even if they personally filled out a survey. Which is why the inferring is often wrong.
In contrast, actually search for watches today, and get watch ads for the next 3 months on every site you visit...
Back when I did freelance writing for magazines there was a book that was published annually called "The Writers Guide" which published the demographic information for the readers of various periodicals. That was developed from surveys, not from context. It helped writers like me to understand what the editor would consider appropriate for their readership.
Why do I have to pay subscription to 10 news papers if i only read 2 articles per month from each? Why can’t I have a button at the bottom that says pay 10 cents to read the whole article where payment happens instantly without redirects and filling credit card info.
Why does paypal/stripe have a fixed minimum fee of $0.30 for doing a database update?
Here is an example of what I am talking about that is implemented via the lightning network: https://yalls.org/
That's Visa + Mastercard's network fee.
------
Y'ever noticed how no-one has ever actually disrupted the online payments industry? It's because Visa and Mastercard own the very substrate of e-commerce through their private inter-bank networks. You cannot get anything done without going through them - and their minimum fees.
The fundamental problem is that the US Dollar was not designed for digital transactions. When bank networks came into existence, they stopped trading physical bills and just started flipping bits in their privately owned databases.
What needs to happen is the Federal Reserve needs to create a digital Dollar system that is secure, is accepted everywhere, and can be transacted quickly with zero fees. Erase the bank networks and turn banks into customer service centers who are given custodial powers (credit, chargebacks, rewards, etc) over your wallet in return for fees.
The government could even levy taxes over the entire US economy this way. No more tax dodging and loss of trillions in public funds. No more TurboTax.
I think there is a lot of content that people love to consume (cat videos, food porn, sports, etc.), even if it is not particularly “worthy”, and vice versa content that is important or worthy, and that you might want to have (even if you don’t always want to actually consume it!), such as deep investigative journalism exposing misdeeds of the powerful, operas, what have you.
With bundles, like The Economist or any other newspaper, or TV channels, you can support or cross subsidise the latter with the former.
With micro payments, not sure that would work. It might lead to more clickbait.
It’s a bit like in the experiment where you could preorder your food at the work canteen for the entire week - people would pre-order a nice balanced healthy diet, while when you make the choice every day, you end up with pizza most days.
Or how people would buy the Encyclopaedia Britannica for home for a substantial amount, and then rarely if ever use it.
However, one thing in favour of micropayments versus ads: they allow you to support an article without having to read it…
TLDR: subscriptions and bundles might allow “the better angels of our nature” to support better and more valuable content than micropayments or ad financing.
I wrote an article about it some time back: https://atodorov.me/2021/03/07/please-support-web-monetizati...
But really, the winners are Facebook, Apple, Google, Amazon, LinkedIn etc. Nearly all of our digital ad purchases have been directly on the big platforms. In-app or network ads have increasingly been absolute dogs for user acquisition anyway.
Meanwhile, I can pay Facebook directly to access their users and they don't have to give up any of their valuable data to anyone to do it because it's all in a closed system.
Maybe I'm wrong and the targeting really does work that well that it should displace practically all other forms of advertising, but it just feels like a situation where it's to nobody's benefit within the ad buying/selling system to really take a hard look at whether this is worth it.
People will pay more to target fewer because it costs them less to get sales.
The problem is that the markets are so danged efficient. Because everyone can so efficiently target the right people, you get exactly what you pay for. So you can blow a lot of money to find out that your message isn't right. Ironically, the amount of targeting data available makes it harder to justify A/B testing just because of the cost involved anymore.
I really don't understand the point of that.
It's not like breakfast cereal. I'm not going to buy another TV because I just bought one.
But it's worth noting that ad networks are more interconnected than most people realize. You can go to Facebook and see the "off-Facebook activity" they have received about you from partners, many of which are retailers. I imagine purchase-history is part of what's shared if you make purchases with those retailers using a email/phone number that's also associated with your Facebook account.
First, I would bet that multiple stores 'know' you searched for a television, while only one 'knows' you purchased one. The stores that don't know you already made a purchase will keep advertising in the hopes that you choose them, and the one you bought from may not have implemented a feature to 'de-target' you, because their conversion rate is low enough that it doesn't cost them much to keep targeting you.
Also, you may return the first one in favor of another, 'better' model; or you may be purchasing for other people or business use.
Further, the extreme trackability of actual ad engagement I think has devalued the sense of value in just "your ad in front of eyeballs": I don't know that an ad you don't click on but that causes you to buy through other channels later really gets valued (though with the amount of tracking that happens maybe people are actually capturing this too).
The other thing that doesn't necessarily get tracked would be "your ad is so annoying/invasive/battery-draining/spookily-targeted that it actually turns people off from buying": they're just non-converted impressions same as any other. (There's also clearly a large component of "very few people really get turned off by ads, if they noticed you even out of annoyance that's good on balance" which might be true, I don't know.)
We'd have seen a shift in ad dollars regardless just owing to how much time people spend online and on mobile devices especially, but how much of the shift is driven just by the availability of the metrics? Again, the metrics are one of these things that's good for both theoretically-opposing "sides" of the ad market: the ad networks can obviously pitch them to marketing, and marketing can in turn use them to justify the ad budget.
semi related, with lower fidelity data a lot of niche products wont be made since their wont be effective ways to get the word out.
What does it mean for something to be a dog?
It actually has a context in marketing for products where you will have a small stake in a shrinking market:
https://www.marketingstudyguide.com/wp-content/uploads/2015/...
My best guess is that's how GP was using it.
Don't this new tracking restrictions favor content creators at the expense of advertising networks?
Let me explain better: If I understand well, now The Economist could sell exclusive access to their readers by selling advertising directly to the advertiser? I believe this gives media with a certain volume of visitors and engagement an entire new leverage over "legacy adtech"
There will be more layers of aggregation on top of that, such as SSPs (supply-side platforms, which aggregate access to publishers' inventory) or PMPs (private marketplaes, which are effectively funnels of select inventory per some critera for some client), but the general principle is that targeting capabilities are shifting from the demand side to the supply side, closer to the viewer. So instead of picking the right segment, effective advertising will be about picking the right publishers. That's good for Google and Facebook because their 1P data is so rich.
Ads are not inherently good, consumers don't need them, and ad companies don't have any right whatsoever to our attention or data. If it can be abused, it will be, and Google and Facebook have shown you can make tens of billions off of personal data and attention. To me it's fundamentally immoral. To see the terrifying endgame to all this, check out "The Age of Surveillance Capitalism" by Shoshana Zuboff.
Some level of targeting is fine. Pushing denture cream to 5 year olds makes no one happy.
Otherwise yea, creepiness is off the charts, wouldn’t mind most of it going away.
This guy says this like advertisers have an inalienable right to show ads to people. Get the heck out of here.
Hopefully the worst offenders here (Samsung Smart TV ads?) are rightfully punished by their users here.
And then of course there is the problem of leakage (Cambridge Analytica et al).
So it's not quite so benign as you suggest. Every marketer I've ever met has a voracious desire for any/all data about anyone they might consider marketing to.
There is the angle of Apple's hypocrisy in terms of Ads and Data Collection, which is of course out of scope of this article. All while they are expanding their own Ad network [1].
And just want to say Thank You, for your newsletter and blog post.
[1] https://twitter.com/eric_seufert/status/1399833956082761728?...
Edit: You may be interested in this thread https://news.ycombinator.com/item?id=27344667
It seems to me very unlikely.
Amen!
"here’s a bunch of reasons why this might not work, not least that no-one except Google and Apple want Google and Apple to have that much control over publishing and advertising. "
The enemy of your enemy is not necesarily your friend.
Things SHOULD change. Moving the power from one big corp to another big corp is not the answer. No matter how convenient it may seem.
So why is it ok for apple to do that, and non of their vendors?
https://www.macworld.com/article/342763/apple-opens-up-its-f...
I'm not sure how that's relevant to your privacy as a user, although obviously the tags and other trackers raise concerns around abuse.
[1]https://nakedsecurity.sophos.com/2021/05/14/apple-airtags-ha...
Apple can't know that it's not a legitimate communication because they don't know who you are, what device is calling for help, and who the device owner is.
And if you want to use my phone to upload a few bytes/second while I'm passing by, as long as the data isn't being stolen from an air gapped system, be my guest.
How are these two sentences at odds with each other? Apple is absolutely "protecting me against evil social media companies" because I give them a lot of money for their products that offer privacy as a feature. They are looking out for Apple by providing a high value, sadly unique service in the industry, and their users obviously appreciate it.
It is possible for there to be aligned interests between a user and a provider. Pointing out that they're doing it for their own reasons (which in this case is their product positioning) isn't the slam dunk you think it is.
I'm saying Apple is ok with exempting themselves. How come their apps are not using the same dialog? Why do they hide behind "location services", when in reality, they are used for similar purposes?
Example, Airtags. https://www.gizmodo.com.au/2021/06/remember-apple-airtags-an...
How can AirTags highjack ANY apple device to report their location, without asking the device owner? Do IOS owners get asked in a dialog about allowing air tag tracking? NOPE
Apple isn't sharing your data with other companies / data brokers. They aren't selling your information, or buying your information. It is a silo. They aren't "exempting" themselves, it isn't relevant to them.
For that matter, maybe you haven't noticed but Google apps don't show this warning either. For whatever criticisms I might have about Google's data siphoning, they silo it and aren't data whores like Facebook and companies like that.
Again, this is Apple's business model, and it's one that many of us appreciate and will pay for, and more for. It is sadly a rare quality now.
"How can AirTags highjack ANY apple device to report their location"
Hijack is an odd way to phrase it.
And again, you may not understand the ad space, as apple is doing what you think they are not doing. [2]
[1]https://nakedsecurity.sophos.com/2021/05/14/apple-airtags-ha...
[2] https://www.adexchanger.com/mobile/is-apple-giving-its-own-a...
You are misunderstanding this entire topic, and neither of your citations remotely support your claims.
Fact: Google apps do not show the tracking permission. Now as homework I want you to go and study why that is, and then figure out how that same scenario might just apply to Apple.
Do you even understand what "third-party" means? Clearly you don't.
Companies were literally using Facebook's ad targeting to discriminate against certain groups of people [1]. There is also the Cambridge Analytica scandal to concern ourselves with too.
1: https://www.npr.org/2019/03/28/707614254/hud-slaps-facebook-...
An advertiser may care about more than one of these things at a time ("interested in baby wipes" + "what culture do they identify with" so I can target things culturally, for instance), but even a very highly-targeted ad will be ignoring all sorts of attributes.
If you are really applying attribute tags to individuals, you are not "grouping" them in the usually-understood meaning of stratifying.
In truth, you are still building an individual profile for each person, but describing it in a disingenuous way.
Under most circumstances this would be a distinction without a difference, but in this specific case it turns out to be important. If advertisers are interested in "groups" than they may still think of themselves as needing identity, and demographic breakdowns, and wanting addresses for those groups, etc... but if you conceive of all they care about as being "in the market for baby wipes" there is no reason to give advertisers anything more than "here is an advertising opportunity for baby wipes", and no other information. Not an aggregation of "individuals", with all the extraneous baggage that comes with, but aggregations of fine slices of attributes that they have no need to correlate to anything else. (Only residual desire because they're probably still used to thinking about "individuals".)
>there is no reason to give advertisers anything more than "here is an advertising opportunity for baby wipes", and no other information.
That's exactly what advertisers are offered. Do you believe otherwise?
That's true for advertisers that are discriminatory as well as ones that just want to advertise what they're selling to people most likely to buy it.
People say ‘privacy’ so often that one can lose sight of the fact that advertisers don’t care who you are - they just want to show ads to people that might be interested in them, and not to show ads to people who won’t be interested. They don’t want to ‘violate your privacy’ - they want to show diaper ads to parents, car ads to people who want a car and watch ads to rich people.
I find this to be untrue many times, as advertisers also want to convince people to buy stuff that they do not actually need or want as well.
They are also very concerned about fraud. A great way to ensure a real user saw your ad is to... wait for it... invade your privacy!
I have no sympathy for these parasites. Surveillance capitalism needs to die in a fire.
The HUD case was a bit unique because there are specific laws about how housing can be advertised.
The Cambridge Analytica scandal is the real jarring one because they were given data normal Facebook advertisers couldn't even dream of getting a hold of.
Current online adverting targets the individual - irrespective of where they are (what site or app) or what they are looking at.
Also, unless you manage your own lead database, targeting to the individual level doesn't really happen in modern marketing systems. You build cohorts like "everyone who downloaded this app and put food as an interest". While it's certainly possible to de-anonymize people occasionally with the advertising data you work with, a) ad platforms never want to give it to you and b) there's an unbelievable amount of noise in the system so you just build programs with the assumption that ~50% of the data is going to be flat out wrong.
You know who was “given” data by Facebook that others couldn’t even dream of? The Obama campaign.
A fact that is very much forgotten by the NYT and every other Cambridge analytica hater.
I'm sure the bull in the china shop doesn't want to break any porcelain, either.
These "marketers", "ad-tech" companies, and tech giants could not possibly GAF less about the consequences for anyone else as long as they get their money.
It is more like "See no evil" than "don't be evil".
The solution is right there on the page, just not in the article itself.
> When those publishers are in the news business, this is another trade-off - pay walls (and privacy!) conflict with reach and public purpose
But there are other ways. One which is already successful is community supported free content, just like this article's author is already doing: they release it early to paying people and then release it publicly. They still get paid, and they still get the reach. I even support quite a number of people on Patreon who provide no benefit to the payer. I like the content they produce, so I pay them to make it, regardless of who else gets to view it.
I think this is one of the most harmful effects of advertising-based business models: it kills other business models. If advertising pays well, then there is incentive for someone else to "steal" your content and slap ads on it. But if we destroy advertising as a viable business model, then that incentive disappears, and we can start exploring other business models.
At times it feels like we're stuck in a local maxima. Momentum plays a part and it can take a significant shift (e.g. from privacy abuses and shifting expectations), before we bother to try alternatives. This _takes time_, and technology moves faster than policy, business, or culture seem capable of tracking.