Hacker News Reader

Comment by turdnagel | Hacker News Reader

I’m confused by this. If Safari users are harder to target, and demand for targeting Safari users hasn’t changed, shouldn’t the price go up?

View on HN

The demand for highly-targeted users is what drives up prices. Since Safari users all look the same now and can't be highly targeted, they have lost that high-end demand and prices have fallen to generic display rates.

Thanks for the explanation. The supply has become crappier while demand stayed the same.

Demand has decreased. The curve of constant demand with decreased supply pushes the price higher.

Supply and demand have stayed the same in quantity, but the quality of both has dropped severely. It's the same equilibrium but at a much lower price point.

I would disagree. "People who can afford Apple gadgets" is a desirable high-income audience. I wouldn't be surprised if it was broken out as a special category for purchase by FB and GOOG soon.

iPhones are popular across income distribution and Macs are a very small market compared to phones

Not really at all. iPhones are only "very" popular in US (high average income) and among the wealthy elsewhere.

A 20-second googling produces one example: https://www.forbes.com/sites/toddhixon/2014/04/10/what-kind-...

Google around for more.

Usually ads care about more demographic data than just device type. Unless you are advertising iPhone cases you probably care more about other data.

Real-time bid requests generally already contain the UA, often partially redacted. So bidders can do this already.

(Disclosure: I work on ads at Google)

I feel really comfortable being in that bucket. It’s a very large bucket.

That’s already priced in, and is likely the floor to which prices have dropped.

There are billions of Apple devices now. Just having one is a very weak correlation to any particular action that a marketer is interested in.

It correlates pretty well with "does this person have money?"

At least in my experience, lots of iPhone users are rocking old models they bought used for like $200.

The point is they are less valuable because you can't track them, so you can't take advantage of their potential value.

Lets say an average Android user is worth 100 value units, and you can get lots of uniquely identifying information about them, say 90% of the ideal amount of info, so their effective value to you is 90 value units each (90% of 100).

Uniquely identifiable Apple users info is potentially much more valuable, say 200 value units. However you can only get a very small amount of actionable information about them because they are so hard to track, due to Apple's security and privacy systems. So maybe you can only get 10% of the information you would ideally like to have about them. That makes their actionable value to you only 20 value units each (10% of 200).

> The point is they are less valuable because you can't track them, so you can't take advantage of their potential value.

This is simply bullshit. So-called “targeted” advertising is a giant scam pulled on non-technical marketers who don’t understand it at all and use “measurements” provided by the ad networks themselves.

Targeted ads don’t actually work in any meaningful sense: https://thecorrespondent.com/100/the-new-dot-com-bubble-is-h...

Numbers below are just for making a point:

Let's say someone is trying to sell snowboards using ads. Previously they could target say "young male users who have previously browsed snow sports." Let's say the ad had a 5% conversion rate and they were willing to spend $100 per conversion (sale). This means they are willing to pay $5 per ad displayed. Now they can only target "all safari users" and their conversion rate is now 0.5% since not everyone likes snowboarding. They are still only willing to pay $100 per conversion, but now they are only willing to pay $0.50 per ad displayed since they need to display 10x as many ads per conversion. If the auction rate is >$0.50, they stop buying ads completely which decreases aggregate demand for ad slots.

In theory a "user" is more valuable the more you know about them because the extra info lets you better target them for higher response rates. Better privacy means you know less about the user, meaning you are less likely to bid high for that user's eyeballs, meaning the user is worth less ad spend.

The actual story is a little more nuanced: when you target an ad, you are usually bidding for users in an auction. You're generally willing to pay more for more specific filters because you're more confident of a higher response rate.

In ecosystems with a lot of data on every user, specific filters actually work, so you end up with higher bids on fewer users, which should drive the average spend per user up without necessarily driving total spend up. With less data, you have to use less specific filters, so you end up with lower bids on more users, which drives average spend per user down without necessarily driving total spend down.

There's also a countervailing force, depending on the market. First an example that supports what everyone is talking about:

* User X loves cola

Pepsi and Coke both bid more to serve their cola subscription services or whatever. Ad price goes up. Now add more data:

* User is a Coca Cola fanatic.

Even though this is the same user, the ad price goes down, because Pepsi doesn't want to bid.

Now add another Pepsi fanatic for symmetry, and you can see how "more data about users for better targetting" can in some cases reduce auction prices (while of course still increasing advertiser value.)

But if users were trending towards either side then the number of users in the middle would presumably be diminishing, driving up prices to target them even more.

Your scenario just seems to single out one particular example where more information can reduce spending on one group, while increasing it on another.

I think this is only on fake-scenario. In a real DMP, the user would be "loves cola","loves coca cola","loves fizzy drinks","drinks sugared drinks", and about 30 other related segments, and one of those would definitely sneak by.

Humans create audience segments, and it's in the interest of dataset producers to have each user be a part of the maximum number of members they can be to maximize monetisation.

> Humans create audience segments

Eventually there will be machine learning models involved (maybe there are already? I don't know ads.) Those models might move closer to "perfect targeting" where advertisers could compete less over certain users.

And no matter the incentives of the data providers, "increase customer value" is probably always a Nash equilibrium. If better targeting leads to less revenue, eventually you'll probably do it anyway.

Imagine you have two users. One who likes apples and one who likes bananas. Both like oranges. I sell apples and am willing to pay $1 per conversion. You sell bananas and are willing to spend the same. OP sells oranges and is willing to spend $0.75 per conversion.

If we can't tell which user is which, we know 50% of our views will lead to a conversion, so you and I will both spend $0.50 per view. The orange seller beats us and the price is $0.75 per user (assuming an old fashioned auction).

If we can tell which user's which, I can spend my $1 on the apple eater, and you can spend yours on the banana eater. Both of us will spend more than the orange seller, who is still only willing to spend $0.75 per view. The price rises to $1 per user.

Dollar per conversion is the constant, and the pile of converting users doesn't change. With less targeting, there's just more hay in the haystack. Dollars per viewer goes up as the number of viewers goes down.

This makes it clear why it's in the individual advertiser's interest _and_ advertising network/website's interest to get more data on the users.

The cost to the advertiser has stayed the same (or in reality, has probably gone down slightly, as it's likely fewer people are bidding on the impression) while the revenue per impression has overall gone up.

Pricing on targeted ads is linked to confidence that a person will perform an action (buy, sign-up, etc). This is a function of the "richness" of data collected -- does the data about you contain demographics, engagement, behavior, so forth.

So a shoe company may pay a lot for a "male/18-35/basketball-fan/shoe-blog-reader". But Safari does not easily allow for that richness of data to be collected, so the price is low.

It's still trivial to target users where browser="safari".

But with Safari's recent changes you can no longer target users on Safari where "non-empty cart"=true, because that relied on cookies and tracking which are no longer available.

On a per-impression basis advertisers are willing to pay (way, way) more if they know they are reaching someone who has already visited their site and been engaged enough to add-to-cart. There are plenty of other examples but abandoned-cart-retargeting is one of the most obvious ways advertisers can see good ROI on expensive ad impressions.

Me too.

Maybe its because the data that can be collected about safari users isn't worth as much.

You have to look at it as the companies buying ads pay more for users that are more likely to result in sales.

Exposure increases this value just by playing a numbers game; therefore, the larger the ad network, the more they can charge.

Targeting increases it more, by allowing companies to choose demographics more likely to be interested in their product, so again, they can charge more.

An interesting aside: While working as a mobile developer, I learned that many apps do not have an in-app-purchase to eliminate ads because it tanks the value of the in-app advertising. Why? Because the people most likely to drop $5 or $10 to not look at ads? They're the ones with the most expendable income, and therefore, users more likely to impulse purchase products advertised.

In other words: By allowing people willing to spend money to support the app, even if there is NO OTHER EVIDENCE they'd be interested in a product, decreases the value of your advertising.

I imagine it’s not that it’s harder to target whether someone is using Safari, but that it’s harder/less effective to try targeting demographics within the group of Safari users.

Advertisers look for a certain ROI on their ads. If Safari users are harder to target, then the absolute return on ad spend will be lower (apparently enough to offset the affluence of Safari users). The way to bring the ROI of Safari ads in line with the ROI of ads on other browsers, then, is to pay less for Safari ads.

Reply on news.ycombinator.com