Slashdot Mirror
Effectiveness Of Online User Databases Questioned
Aleatoric writes: "According to this article from the NY Times, advertisers aren't exactly buying into the claimed effectiveness of targeted online user databases. Not to get complacent, though, it also includes comments from many sites that gather user information concerning their efforts to try and change this attitude." Amusing. It seems Web advertisers are just now learning lessons direct mail and print advertisers learned long ago.
Internet business must necessarily be low markup. Your competitor, who sells your product for 1% less, is just a URL away. Why bother looking at advertisements when you can find out product information yourself with one click on your favourite search-engine? Economics 101 says that this is a recipe for everyone just breaking even.
My impression of the Internet is that it is modelled more along the lines of a Public Library, rather than a department store. There is so much free information that one tends to ignore anything that is for pay.
So, at last, advertisers are finding that banner ads don't work. Maybe the real problem is far deeper. The Internet is wonderful for broadcasting information about every conceivable subject, but I see no evidence that it can generate the huge cash-flows that would be necessary to support the outrageous market values of most e-businesses.
I don't need to be targeted everywhere. As anouther poster mentioned, the best ads are not targeted - I'm not a target of a Jaguar based on anything I have (I've driven S10s all my life - which isn't very long), but with some advertising I might look at one for the next car instead of anouther truck. Maybe.
OTOH, if I walk into a music store I want to see music I like. Imangine some face recigniction that sees my face, and adjusts some robotic shelves. Suddenly instead of 10 bluegrass CDs out there are hundreds. Today if a music store has the latest from IIIrd Time Out, it is not on display, but if they could put it on display when I walk in (Taking away Jazz CDs that I's never listen to anyway) I might buy it. But I probably wouldn't ask since I don't expect them to have it. However they do need to account for wanting other things. They won't know that my brother wants some Heavy Metal CD for his birthday next week, even though they have it. SO long as I can find plenty of the type of music that I want and something of other styles I'd like to be targeted. It would make my shopping expirence more useful to be targetted.
half of the internet companies seem to tout the idea of target advertising (especially the niche or community sites) as being a major source of revenue for them. esp the companies that dont have any revenue at the moment but will Real Soon Now we Promise. doesnt this article in the nytimes suggest that these people are full of crap?
more importantly, if its in the nytimes, which probably has a readership only surpassed by slashdot (hee hee) that suggests that this is no longer something only the underbelly of dissatisfied IT workers (like many slashdot readers) know about.
anyone else think the stock market is going to fall quite a bit more?
unc_
Sadly, the rate card is not the entire story, or even a small amount of it.
Most sites sell their ads for substantially less than what's on their rate cards. It's not easy to sell 500,000 ad impressions a day.
Don't bet on Slashdot making as much as you think.
D
----
That's not all that much in today's economy, considering the kind of value they've brought to the table. However, the $ 1.5 million downpayment plus the $ 4.5 million they've been promised in the future is enough to keep them in Jolt colas, gadgets and shiny new real estate for quite some time.
But the site itself has to make money for it to continue operation long-term, so I'd still say it's a good idea to lend it support.
D
----
It's really very simple and if marketing gurus weren't deluding themselves about their profession they'd understand this perfectly:
Marketing is mostly lies and obfuscation. But you can't lie to people who already know the truth. Therefore marketing to people who are familiar with an industry is useless (unless you have a message with real content).
--
Have Exchange users? Want to run Linux? Can't afford OpenMail?
Linux MAPI Server!
http://www.openone.com/software/MailOne/
(Exchange Migration HOWTO coming soon)
The guy from AOL had a great point. Why not simply focus on doing what the web marketing does best -- providing information to the consumer?
The interesting thing is that this seems to be part of the long running interface wars between people promoting intelligent (e.g. analytical, predictive) behavior in systems versus the people who say just make the interface responsive (I tend to fall into the latter camp, myself, but I'm not dogmatic). The idea is that an intelligent agent with a good enough database will turn the web into a surgical strike advertising medium. Like a lot of the intelligent agent stuff, I think it would be wonderful if it would actually work, but it is unfortunately not all that promising.
Traditional marketing techniques are used very effectively with broad stroke broadcast media; what these people are promising is the exact opposite -- narrowcasting. With broadcasting, you can count on two things -- building your image with the public and reaching a statistically certain number of people ready to buy your product right now. Sure, you want to target the right magazines and the right TV shows so you don't pay to reach people who would never buy your stuff in a million years, but you're still reaching a very broad audience. With narrowcasting, you forgo the wide image building capabilities of broadcasting, as well as the statistical certainty of reaching so many people ready to buy, with the hope of dropping the lure right on top of the fish.
The problem I see with this model is that perfect timing is essential. Sure, I may be the kind of person who buys minivans, but that does you no good if I bought a different one a few months ago.
You can still do the same kind of broadcast style advertising on the web (like I presume companies do on Slashdot <gloat> -- although thanks to junkbuster I don't see them</gloat>), but the Internet is unique in that the user is more in charge than he is in other media. So, with the web, you're out of psycho-marketing mumbo-jumbo and you're back into capitalism 101: providing products and services.
Post may contain irony: discontinue use if experiencing mood swings, nausea or elevated blood pressure.
I work for a DSS/Data mining firm in the retail business. Retailers are not interested in what a consumer wants so that they can charge you more; they want to know what you *don't* want so they don't buy it in the first place. Excess inventory (i.e. wasted money) is the retailers' worst enemy.
They are also not very interested in individual tastes: it's very expensive (technologically) and, as the article points out, the law of diminishing return applies. They really want to know more about seasonal and regional buying patterns, again in an effort to minimize excess inventory.
Keep in mind, if a retailer has excess inventory, they're not likely to just absorb the cost; they will make up for it by marking up the rest of their inventory (they have to make a living after all). So, actually the inverse applies here: if a retailer knows what you (in real-life, their average customer) want, you're gonna end up paying less for it.
Look at the kings of supply-chain optimization: Dell online, Wal-Mart offline (Wal-Mart unbeknownst to many, pioneered data-mining and supply chain management, and they still lead the pack). Neither are known for high-prices or bad customer service, and they have actually achieved fatter profits and higher customer loyalty.
Data Mining is *good* for the consumer, as long as private information is kept private (e.g., names are thrown away, but not data like gender, ZIP codes, or types and frequency of purchases).
engineers never lie; we just approximate the truth.
Net Perceptions would claim that that isn't always the most effective way to sell products. One example I heard was that if someone buys a towel in a B&M retail store, NP would recommend that the customer be asked "Would you also like to look at our vacuum cleaners?" There were claims of increased sales through this method (can't find the article that used this example).
From http
--
The shareholder is always right.
I, personally, enter fake information into 90% of the forms I fill out on the 'net. There's no reason these people need detailed information about me, and I suppose I get some innate pleasure from entering 'Joe Smoe' '123 Main St.' 'Los Angeles, CA' '90210' Anyone else do this?
The format of the cookies file is fairly .netscape/cookies file, and remove any that .logout, or a cron job, like
transparent. After visiting the sites that
you want to keep the cookies for, make a copy of
the
don't come from the sites you want to recognize
you (I keep nyt & slashdot.) Then put something
in
cp ~/.my-cookies ~/.netscape/cookies
Alex.
The rate cards for UserFriendly and Slashdot/Andover are available, and show their rates ranging between $22 and $70 CPM. So your million dollar a year estimate is around 4-10 times too low. Furthermore with a third of a million registered users, and each one viewing at least 5 stores per day, (I'm understimating here, I must generate 100-300 views per day, depending on how many comments I read), we're talking another factor of 3-10.
So by my estimates, SlashDot could bring in anywhere from $12 million to $100 M per year.
That's enough to make me want to get in on the goldrush!
An internet browser should be released that records your activities on the network and sends the records to advertising companies that gain profit from this information. All people must have micro-computers implanted into them and the browser (that also can be installed on those micro-computers) will be setting cookies in humans' brains and reading their minds (as an undocumented feature, this browser will impose its will over the users' minds and will force them into impulse buying sprees, will destroy all violent behaviour in its roots, and will work as a credit-card authentication system.)
On the bright side this technology will save users' time. A user will not have to go to courts any longer when he/she runs a red-light or speeds too much, since to prove innosence or guilt all that would be required will be using the browser to access user's memory core and to read what actually happened. (this of-course does not mean that there will be no more lawyers in our world but their function will change from squizing their clients through a pin-hole of technicality to proving that the client's browser was not installed correctly
You can't handle the truth.
What kind of people actully ever click on banner ads?
I'll admit I occasionally glance at them, but I sure as hell won't click on them.
Caveats: This rate of return is ridiculous compared to the current reality. I don't think anyone claims that, currently, targeted ads are 55 times more effective. But if it could be done ... and that's the chimera people have been chasing.
The Mongrel Dogs Who Teach
The danger is that those few persons out there that would like to do these subversive things can now do it much easier than ever before.
Lets not forget that history also shows wherever there is a new business oportunity to be exploited, someone will. We probably can't imagine all the _new_ and unwonderful things to come from all this personal detail.
smellthecoffee
...................
...................
...................
does that mean that my secret of visiting 'special' sites is no longer?
no wonder i've been getting catalouges for 'squirrel action' and 'rodent paradise'
i guess i'll have to bury my racoon suit.
*sigh*
Be you Admins? nay, we are but lusers!
I found this article most amusing once I remembered this story from Wired, which got all hot and bothered about the number of registration profiles the site has managed to collect.
We have an article about the inefficiency of user databases that requires registration in order to be read... ;)
I think it's self-defeating to say, "Oh, I'll never, ever click on a banner". Sometimes the products and services offered through them are of value. When I think clicking would be worth my time, I'll click.
I don't click on ads often, simply because I'm normally focused on the content of the page I'm reading. I'm much more likely to click on an ad when it's what I'm looking for.
For instance, say I'm looking for a good deal on a Mercedes-Benz automobile. I go to Yahoo's Mercedes-Benz section and see a banner: "Buy your Mercedes today from CarsDirect.com". I click, because it's something I really want to check out.
Now, Yahoo could then track my click to, say, the real estate page. I'm looking for real estate in Malibu, and a "CarsIndirect.com" banner pops up: "Buy your Mercedes at CarsIndirect for the best prices!". But then I'm focused on my real estate, so I don't click.
This, in my opinion, is why targeting doesn't work. We're all focused on a particular task. Right now, I'm posting a message to Slashdot. Unless the banner ad blinking above me was totally compelling, I'm not going to click on it because I'm focused on my message. This is Yahoo's problem; Yahoo email gets negligible clickthrough no matter what because people are thinking about their email and what to say in response to it. So if Yahoo sells a few targeted ads in the Mercedes section, they can't sell them when people are reading their email. But if they sell sitewide ads, Mercedes would have to buy some email placements to get the one they really wanted.
D
----
The answer was in the article - as little as $ 5 per thousand impressions. So if your site got 1,000 impressions a day, you'd earn $ 5 a day, or $150/month.
Of course if you get 200 million-odd impressions a day, you'll get about $ 1 million a day.
Slashdot gets about 500,000 impressions a day, so it earns a minimum of about $ 2,500 a day. This means gross annual income is around $ 912,500 if they sell all their ads at $5 CPM [ad industry jargon for cost per thousand]. Fortunately, Slashdot doesn't have all that much in the way of expenses - I believe Rob and Hemos make about $ 90k per year each, and there are probably a few others at lower salaries. The people selling ads get a commission, so it's fairly likely that Slashdot in and of itself either is fairly profitable, or could be.
The reality is that most of that profit is most likely absorbed by andover.net corporate overhead, but Slashdot is probably doing better than ever in sales terms thanks to that same corporate overhead. - a dedicated ad sales force is a healthy part of it.
The picture might not be so rosy, since sometimes a lot of ad inventory remains unsold. There's probably some sort of deal to pick up unsold inventory, but usually it's for next to nothing.
If you want to support Slashdot and make sure it grows and thrives, it's probably a good idea to click on the banners. They are closely tracked, so if there's a company you'd like to learn more about, by all means click. It can't hurt.
D
----
For the privacy freaks, or just the lazy ones out there, keep in mind one can always use NYTimes's partners link (eg replace www. with partners.).s onal/050700personal-privacy.html ..
In this case the URL would be http://partners.nytimes.com/library/financial/per
This will bypass the required registrtation forms
-- Chris Chabot
"I dont suffer from insanity, i enjoy every minute of it!"
The problem here is much more with the medium than the information. In fact, Roblimo's comment on the front page is inaccurate. Direct mail and print advertisers spend 5-10x more for targeted subjects. You can buy direct mail names for US$5 CPM, but it'll cost you US$80 CPM for a list of mothers who recently gave birth.
.
I think you missed the point here. While it is true that a mail-order house can often pay up to 16 times more for a targeted mailing list than a shotgun mailing list, the mailing list doesn't represent the bulk of their direct mail advertising dollars. In fact, this represents a rather insignificant cost compared to the color catalog (about $3-$4 in 10K+ quantities), or mailing them (about $.20/$.50 per unit).
Frankly, the discount you can get from the US postal service for presorting the mail by delivery route is often greater than the cost of the mailing list. So even paying $0.08 per name (which is what $80/CPM works out to) is nothing compared to sending out a $3-$4 catalog, and could potentially be saved by properly presorting the mail anyways.
My point is this: if you work in mail order, the cost per person to advertise between a "shotgun" mailing list and a "specialized" mailing list is something like $3.58/person for a specialized mailing list verses $3.51/person for a shotgun mailing list. If the specialized mailing list yields a 10% increase in market effectiveness, then you're talking about a 10% increase with a 2% increase in marketing costs.
The on-line model, however, is quite different. The $/CPM model is the total advertising costs, and not one (rather insignificant component) in the total advertising campaign costs. (Yes, you need a web site to redirect traffic to. But most companies already have a web site, and unlike a color catalog, doesn't cost twice as much if you have twice the viewers. So the incremental costs of additional eyeballs on a corporate web site is insignificant.)
So if you're talking about charging $80/CPM to target an audience by which you have little (if any) evidence that they are well targeted as your statistical profile is incomplete, verses $5/CPM for a shotgun list--you're basically asking advertisers to pay 1500% more for advertising that may or may not yield an additional 10% in results.
If I were an advertiser, I'd happily pay an additional 2% to get an additional 10% yield. I'd call you bloody f-cking insane if you asked me to pay a 1500% premium for the same incremental yield.
Yet that's the direct marketing internet advertisement model in a nutshell. Worse: because the statistics are incomplete, you cannot even show that I'd get a 10% additional yield for my additional 1500% investment!
Magazine interstitials (the full and half pages in the midst of your articles) must always be at least glanced at so you can discern whether or not it is part of the article)
Magazine interstitials also serve the purpose of providing more information when a person goes back to the magazine to find more information about a product he is interested in. That is, advertising in a magazine serves two purposes: first, to encourage someone who hasn't thought about buying something to think about it. Second, to provide more information to someone who is interested in buying something how he may go about it.
For example, I advertise a product in a magazine for a bug tracking program I developed and sell under my own banner. People who search out and buy my product use the ad to know the URL of the web site they should visit to purchase the product. I'd say the vast majority of the people who visit my web site and purchase my product don't immediately see the ad and go "I must have Bug Tracking Software!!!" Instead, they save the magazine on the shelf for a few months, and then when someone says "we're starting a new product, and we need bug tracking software"--they pull the magazine off the shelf, flip through it, find my ad, and visit my site for a demo.
Banner ads don't work because they are not persistant. That is, while I'm paying approximately $15 per 1000 readers of that magazine per month for the advertising, the advertising is persistant--they can view the ad again and again, and go back to that ad months later, even after I stop the campaign. Banner ads, on the other hand, cost just as much (if not more) than a quarter page magazine ad, yet are not persistant, have less are to provide information, and for a product like mine, do not drive traffic effectively.
Television is not persistant--but it's a hell of a lot cheaper than a banner ad per view. My brother ran a series of television ads for a local political campaign. One way he ran his 30 second spots was to run them "shotgunned" through the day on several major local television stations: for about $20 per insertion. Per viewer, that worked out to be less than $1/CPM--and on some news programs, a hell of a lot less.
Banner advertisements don't work because (a) you can ignore them more readily than television ads, (b) they cost a hell of a lot more than magazine ads which are persistant. So basically banner ads take the worse of television (lack of persistance), add a terrible feature (making them easy to ignore), and jacks the price up by about 8000% per view, on the theory that their interactivity (that is, the ability to instantly click on the ad and get more information) is somehow worth it--dispite a less than 1% clickthrough rate.
This is why the email firms (yesmail and MyPoints jump to mind) are going for so much. It costs US$60-80 CPM for emails to MyPoints members (volume discounts), a factor of 100x over banners, and I believe they just had price hikes. The fact is, if you belong to those lists, those are ads you must read. This just points to the medium is what makes the difference, not the targeting/data.
I skim my e-mail. Yes, the advertising is interstitial, but the mail is only as persistant as my delete key--it's not like a magazine which I may store for a couple of months in the magazine rack out in my living room, in part because I paid real money for the rag, and in part because I subscribed to the rag for about a month or two worth of "spare time" reading. So at best we have something with the same properties as a television ad: zip for persistance, but still costs a factor of 80 more per view.
Great.
Sometimes I think Silicon Valley is full of people who couldn't balance a checkbook if they used all the fingers on both hands and all the toes on both feet.
Because I think demanding an 8000% increase in advertising costs over a basic advertising model with questionable effectiveness in the first place, in order to get (maybe!) a 10% increase in eyeballs is beyond silly.
It's pretty refreshing to read an article like this. I've seen from the inside the efforts that some .com startups make to analyze their clickstream data, when they should be spending their money elsewhere. It seems like a Silicon Valley phenomenon. The thinking seems to be, "If we have all this database power, it MUST be useful somehow." But the marketing folks at the very same company, when they go to buy advertising, are interested in raw CPM's.
The article contains the wonderful quote, "If you want to sell cars you advertise on the cars page." That little bit of wisdom is a whole lot cheaper than hiring a bunch of DBA's to tell you the same thing (or worse, to tell you something different). But in the Valley, technology rules, even when it's wrong.
I wouldn't mind corporations having a lot of my information if it was passively used: I walk into a store and the things I would most likely buy are pointed out to me and put on sale (as proposed by National Cash Register/ JCPenny alliance), or as with webpages advertising especially for me. I'm going to be seeing ads anyway, right? They might as well be for something I want.
What I fear, though, is stigmatic stereotypes emerging from patterns in consumption habits. How would you like the government to think you're a Communist, or a homicidal maniac because you shop like they do? Or you try to get a job, but they think you're a pedophile because you ordered the Playboy channel and Cartoon Network (props to the Drew Carey show).
There are certainly things we will have to look out for, but I will not instantly condemn massive information gathering.
Instant Crisis
"These people looked deep within my soul and assigned me a number based on the order in which I joined." --Simpson
That's the thing about this directed marketing stuff: if you gathered all the information in the world about me, how can you know what kind of car I (or my wife) may buy in the future?
And further, if you already know what kind of car I'm going to buy based on my past history, socio-economic status, and other purchasing habits, then why bother to advertise to me in the first place?
Advertising is about influencing the buyer, not about telling the buyer about what he already wants. If I want to purchase a Miata, I don't need a Mazda advertisement to tell me how wonderful a Miata is; I'll probably just go down to a Mazda dealership and test drive the thing. That is, if I want a thing (an MP3 player, say, or a new computer), I'll probably search it out. At that point, all the advertisements in the world won't matter to me.
So targeting me based on my past statistical information is close to worthless. If I'm inclined to do something, I'll go out and kick the tires--advertising won't change my mind. Targeting me with ads which I'm statistically inclined to do--even if it's statistically correct--won't change my mind because I'm already past the "get my interest" stage where advertising generally appeals.
(The example in the article--that of showing you your stock portfolio in an ad to encourage you to trade: it won't work. If I have an account, I'm already past the "get my interest" stage, and I have the account. No amount of reminding me of what I already own is going to encourage me to manage my money differently.)
Further, based on my and my wife's past statistical information, you could probably guess that I'd want a small sports car. (Male, 34, married but no children nor plans to have one, upper-middle class neighborhood, professonal, college graduate.) But you'd probably never guess in a million years that my wife would fall in love with an S-class Jaguar.
She fell in love with the Jags, by the way, because of a rather effective television commercial starring Sting--which, frankly, was a shot-gun ad and not targeted in any way.
Shotgun ads are great because they get you thinking about things you'd never normally think about--such as wanting to check out a large luxury car when you're used to driving inexpensive subcompacts. If Jaguar would have used nothing but targeted ads, they'd overlook us--and probably lose a potental sale.
I'd happily "drop my shorts" so to speak and provide advertisers all the information about me they want, including my tax returns if they're so inclined. It's worthless crap. My wife and I have no kids; we don't plan to have any--but it doesn't preclude me from shopping in Zany Brainy for Legos, Beany Babies, and little toys for my relative's kids. And we have never owned a luxury car ever, and we're not even in the right target age range--but that doesn't preclude us from kicking the tires on a Jag...