The Advertising Value of Intrusive Tracking

Here's an interesting research paper that tries to calculate the differential value of privacy-invasive advertising practices.

The researchers used data from a mobile ad network and was able to see how different personalized advertising practices affected customer purchasing behavior. The details are interesting, but basically, most personal information had little value. Overall, the ability to target advertising produces a 29% greater return on an advertising budget, mostly by knowing the right time to show someone a particular ad.

The paper was presented at WEIS 2015.

Posted on August 24, 2015 at 5:50 AM • 26 Comments

Comments

proxy filterAugust 24, 2015 6:14 AM

Ads work, and there will always be incentives to track. Rather than depend on legislation, people can easily filter out all the trackers and adware with an automatic proxy setting. There's apps that do this, and installing one at home is easy.

My proxy.pac filter is filled with items that look like:

BadDomains[i++] = "boomtrain.com"; // Tracker - 2015-04-25
BadDomains[i++] = "brandaffinity.net"; // Tracker - 2012-07-12

The ad industry will never defeat blacklists.

ianfAugust 24, 2015 6:46 AM

[…] basically most personal information had little value. Overall, the ability to target advertising produces a 29% greater return on an advertising budget, mostly by knowing the right time to show someone a particular ad.

So, our browsing & purchasing habits are gathered, folded, spindled and mutilated, and our privacy constantly invaded by mindless bots, all to NO AVAIL? Nice to know that, also that the ad placing agencies are being sold old, long-outdated data. A case in point: couple of years ago I conducted a search for getting a teenager a driving license. Since then, some advertising networks (e.g. washingtonpost.com's website) have been showering me with pointers to local driving schools. Because, apparently, "once looking for X, I must always be looking for X." And WaPo pays for it, if mere pennies an impression!

MichaelAugust 24, 2015 7:40 AM

Just to be clear, in the world of business and finance, a 29% greater return is NOT nothing. In fact, it's rather significant, considering they will shave a few millimeters off the bolts in your car to reduce costs, or add 0.9 cents to every gallon of gas.

At the same time, 29% may be small enough that stricter regulations on data collection could make it less profitable or not worth the risk.

ttvdsAugust 24, 2015 8:13 AM

Well, here's a conclusion one could come up with: advertising industry needs more data and more sophisticated algorithms.

JeffPAugust 24, 2015 8:56 AM

I with @Michael and @ttvds. 29% return is significant. The advertising industry absolutely needs more sophisticated algorithms. The example experience I frequently refer to is: I buy a computer mouse on Amazon. For several weeks, I get bombarded with "Hey, look at this mouse." My browsing experience is not enhanced.

WinterAugust 24, 2015 9:05 AM

These 29% increase is the result of lazy thinking. There is a case where detailed data of customer purchases was hugely successful in selling more goods:

Amazon's book suggestions.

That is so good, that people actually go to Amazon to get suggestions. People looking for marketing, who could believe that?

But marketing seems to be so strongly targeted to fleecing the customer that trying to give the customer value back simply does not enter the equation.

Or we see the famous old saying:
Everyone can sell the best product at the lowest price. The real challenge is to sell the worst product at a highest price.

As a side not. I find it very worrying that the best and brightest minds are currently employed to manipulate people to buy rubbish.

daily sampanAugust 24, 2015 9:26 AM

To be honest, I too have been underwhelmed by the accuracy of advertising algorithms.

Not long ago I ordered a bus travel pass from my city's transport office website. The banner I saw immediately after I completed the purchase was an advert for "travel" to Indonesia. Are they basically running simple word searches and throwing up whatever crap is on their list?

I guess it's a numbers game, like good ol' spam. It only takes one sucker or a lucky hit for the entire thing to be worth it.

JasonAugust 24, 2015 9:42 AM

My favorite amazon algorithm fail. We have pet rats and have searched on amazon for supplies (rat bedding,etc). Got an email saying "you may be interested in".... A device that traps and electrocutes rats.

AnonAugust 24, 2015 9:47 AM

I've spent a few years working in the on-line advertising industry, as a developer.

With regard to this study, I may be wrong, but I have the impression they're talking about the value of personal data *given the current methods used in general in the industry to decide which creatives to show to which users*.

The methods in use today are primitive. They revolve around putting users into groups ("segments") and advertisers choosing which groups to advertise to, with some other simple constraints, such as country, time of day, if a given user has seen a creative before, etc. The main other method is to examine the search terms a user has entered into a search engine, and to display creatives based on that (or to place users into segments, based on those keywords, and then advertise to them as before).

Given how unintelligent these methods are, I am not surprised personal data has relatively little effect. It would be like trying to perform heart surgury while wearing rubber gloves; you cannot utilize subtle data when the only means by which you select users is so crude.

J-PAugust 24, 2015 10:18 AM

@ttvds wrote:

Well, here's a conclusion one could come up with: advertising industry needs more data and more sophisticated algorithms.

Maciej Ceglowski made the point in his talk at Beyond Tellerand that this isn't a failing of the existing industry: it's the whole thrust of it.

In his model, the industry helps guarantee its future by being able to tell investors "if you give us more money, imagine how we could build more sophisticated algorithms! Imagine how much better we could be! We're so close!" He calls it "investor storytime", where you sit them down and paint them a picture of how much better it could be, if only....

If that's the case, then it's not in a lot of players' best interests to end up in a state where they can't ask for more money to maybe increase it in the future: always jam tomorrow; never jam today. But more data? Eh, they could always do with access to more data. What harm can having more data do? We'll get the investors to buy us a bigger RAID stack!

what advertsAugust 24, 2015 10:49 AM

I block all adverts from all sites. As the first comment states, block lists will always work. I currently maintain lists on all my devices, but now I have a proxy server at home (Sophos UTM) that will do it all for me once I get it set up.

I know that adverts pay for the content I look at, but with the current state of some adverts at the moment trying to install spyware and other crap on my machine, the only choice I have to to block all of them.

ianfAugust 24, 2015 10:59 AM

Yes, 29% is significant, but let's not forget the point that (that paper's writers and/or) Bruce made:

most personal information had little value. Overall, the ability to target advertising produces a 29% greater return on an advertising budget, mostly by knowing the right time to show someone a particular ad.

proxy filterAugust 24, 2015 11:10 AM

most personal information had little value. Overall, the ability to target advertising produces a 29% greater return on an advertising budget, mostly by knowing the right time to show someone a particular ad.

Based on the ROC curves in the paper, the clear conclusion is: personal information has no value unless ad timing is used.

Ad timing is the only method that shows performance better than chance.

David LeppikAugust 24, 2015 11:14 AM

@proxy filter: Good point. 29% is huge. Who wouldn't want a 29% raise, or would fail to notice a 29% drop in income? But it's an entirely different story if, 90% of the time, you tell your customer, "sorry, I can't take your money right now; it's the wrong time to show your ad."

ianfAugust 24, 2015 12:09 PM

sorry, I can't take your money right now; it's the wrong time to show your ad

That ad salesman(?) would qualify for The Observer's Ethical Awards, thus cavort with Björk & Yoko Ono in the well-catered green room. 2nd year in a row—assuming s/he'd have survived that long—be shortlisted for the Alternative Nobel Prize.

Any takers?

Slime Mold with MustardAugust 24, 2015 4:57 PM

A couple of years ago, a large supermarket chain in my area stopped its customer loyalty cards program. A spokesman for the company explained "knowing everything about our customers did not increase our profits" (emphasis added). I had sort of hoped it was a trend.

For many products, directed advertising is only sensible. I do not think Bruce would realize much return if he advertised his services during "Dancing with the Stars". "Directed" doesn't always mean intrusive. Most of us advertise in trade publications, realizing that it brings few customers directly, but is crucial when a word-of-mouth referral realizes that they have, after all, heard of you.

I have read an interesting alleged fact: Far more people research a product online, and then purchase it a brick and mortar store, than actually purchase online. It's what I do when feasbile.

@ Winter

If you ever again refer to the marketing folks as "the best and brightest" I will:
1. Track you down
2. Send my Marketing Department to live with you. They are nice people. They are all really good looking . Their collective IQ may equal the Cosmic Microwave Background radiation (measured in degrees Kelvin), but then again, it might not.

GrauhutAugust 24, 2015 5:41 PM

Sometimes i like personal offers. Amazong Inc. knows me so well and tracks that well, sometimes they show me personal rabated offers refecting the price i had seen seconds before on an other website. Great feature. Saved me some bucks.

If i had a shop to host, i know on wich cloud i wouldnt put it... ;)

rgaffAugust 24, 2015 7:14 PM

@Slime Mold with Mustard

"research a product online, and then purchase it a brick and mortar store"

Except, I've found that after I've researched the exact make and model I want, I can't FIND it in ANY brick and mortar store any longer... it's simply not there. Sure there are others like it, but what, do I run back home and research those other brands too and compare, then come back and maybe buy? Or just buy online... If I knew I could find it locally though, sure, why wait and pay extra for shipping...

@ianf

So I guess we're interpreting Bruce's original comment to mean "most [individual kinds of] personal information had little value [except for timing related ones]"? Interesting. Cause otherwise, yeah, 29% is not little. But can't almost anything be considered timing related in some way?

Dirk PraetAugust 24, 2015 7:32 PM

@ proxy filter

... people can easily filter out all the trackers and adware with an automatic proxy setting

That may not be sufficient. I recommend installing browser add-ons such as Adblock Plus, UBlock Origin, Privacy Badger, Ghostery, NoScript and the like. Hopefully EFF's Privacy Badger will be included in TBB soon.

Ad injectors are a real plague. They drive people nuts and I'm seeing this stuff on nearly every device friends and family are complaining about. Google has recently declared war on them and has pulled over 200 Chrome extensions that were actually ad injectors.

WinterAugust 25, 2015 12:09 AM

@Slime
"If you ever again refer to the marketing folks as "the best and brightest" I will: "

I was referring to all the top math, physics, and psychology people hired to do the data mining. The people that use these data centers that melt the pole caps.

My whole point was that all these resources are then squandered by people trying to sell you items you already bought.

BTW, I am opposed to cruel and unusual punishment as a matter of principle. Nothing these marketeers have done could justify such treatment.

tyrAugust 25, 2015 1:38 AM


I find the whole thing puzzling. 29% of what ? Does that say
thhat targetting ads when the people actually want to buy the
item works better than when they don't want it.

I'm reminded that Pavlov noticed after conditioning dogs to
salivate to his bell, that it only worked when the dog was
hungry.

The real horror of the advertising wars on the net is the
tracking cookies and following you around out of prurient
curiosity because you might need an ad for some crap one of
these days.

Running enough add-ons to have a reasonable browser actually
is a partial solution to the most egregious stuff. Set
Ghostery to block everything, only temporarily allow what
No-script blocks, Privacy Badger knocks out all those iconic
tracking buttons from the worst offenders (FB,Twitwit, Giggle
etc.). None of this is perfect but it cuts the extraneous
crap. As an ancient Net user I shudder to think of the wasted
bandwidth all of this tracking and pushing unwanted ads takes
up.

The real value of intrusive tracking is to irritate users to
the point of mounting active defenses against it. I buy stuff
I locate on line but never from anyone who pushed me an ad.

BoppingAroundAugust 25, 2015 9:08 AM

Slime Mould,
> A spokesman for the company explained "knowing everything about our customers did not
> increase our profits" (emphasis added). I had sort of hoped it was a trend.

Have you considered the possibility that they had found a more reliable data source? Or something else along the lines.

CallMeLateForSupperAugust 25, 2015 12:55 PM

I used Amazon for a while - about six times over about 18 months - and stopped cold-turkey about six years ago, for a number of reasons. Which is to say, I don't know the current quality of Amazon's product suggestions.

Maybe suggestions improved in the mean time, but my experience ranged from unfavorable to meh. Many were off-subject (e.g. browsing DVD burners elicits DRAM suggestion) and too many were, to be kind, strictly-from-hunger (e.g. I targeted a particular book- *by*title* - about bicycle wheel building, Amazon suggested a variety of bicycles).

"Do what you love and do it better than everyone else."
because
"Build a better mouse trap and the world will beat a path to your door."
and embrace and cherish your customers
because
"The best form of advertising is word-of-mouth." (I would add, "from someone you know and trust")

One trouble with producers today: they are hit-and-run capitalists. All too eager to sell-sell-sell, "build the enterprise", sell it and make a killing so they can reinvent themselves and do it all over again, Thus we have a plague of sub-standard products that are here-today-gone-tomorrow.

ianfAugust 26, 2015 5:51 AM

@ rgaff “… 29% is not little”

The thing to remember is that basically all sales promotion-based surveying is JUNK SCIENCE. If your livelihood depends on selling allegedly vital marketing services, you can prove anything with any given set of data (statistics do not lie) to anyone with a budget. Just pick a strategy & arguments that will allow the client (advertiser) to justify it upstreams. So that figure of 29% has to be seen in relation to overall standard infinitesimal online ad conversion rates … here at best translating to 13, rather than 10, people noticing the product in some ways (not necessarily 3 more sales), for each say 10000 paid-for impressions or eyeballs.

CallMeLateForSupperAugust 27, 2015 8:21 AM

Short article in 8/26/2015 BostonGlobe(dot)com:
"As ad blockers rise, will the Web business model fail?"

Doesn't answer the question, but there are some interesting quotes from several persons on both sides of the issue.

Leave a comment

Allowed HTML: <a href="URL"> • <em> <cite> <i> • <strong> <b> • <sub> <sup> • <ul> <ol> <li> • <blockquote> <pre>

Photo of Bruce Schneier by Per Ervland.

Schneier on Security is a personal website. Opinions expressed are not necessarily those of IBM Resilient.