Online news registration may not be working [Thanks paidcontent!].
Among the reasons offered are that there’s a lot of false information being posted and people are sharing registrations. Some ten to twenty percent of registered email addresses are bad, and no one knows how much of the demogrpahic data is phony. But newspapers seem to think they need demographics to sell ads.
“Our view is that we need help from you: We’ve got to pay for what we do, we’ve got to convince advertisers into looking at us and tell them that these are the demographics we now know about our readers,” [Atlanta Journal-Constitution ombudsman Mike King] said. “The old standard advertising geared to people who live in the areas we cover doesn’t work anymore.”
But Google built the largest, fastest-growing advertising business on the Web without any demographics at all. And a huge chunk of that business is never touched by a sales person.
The other part of the justification is that it’s becoming more expensive to publish a newspaper on the Web. That’s weird. It should be getting cheaper.
The goal behind registration is targeted (read: more expensive) advertising. Whether that program is successful or not depends almost entirely on whether advertisers believe the underlying registration data is accurate. If so, they’ll step up to the CPM premium. If not,…?
I’m not convinced that the demographics is all that useful on the Web, especially such crude measures as age and sex. That’s not very useful compared to behavior, which is one reason Google’s doing so well selling ads.
I found the AP story to be poorly sourced (except for the inclusion of Fred Mann, who DOES know what he’s talking about) and poorly quantified. Why would an ASNE administrator know diddly squat about Web registration? What actual error rates are showing up in the data?
I’ve spent quite a bit of time looking at actual newspaper site registration data and I have found under one percent “outright liar” rate. What I mean by that is people who claim to live in silly places and use silly names. The rest are reasonable names, reasonable actual local addresses, and reasonable ages. That may seem surprising, but then, most people are not as nutty and paranoid as it might seem when you look around on the Internet.
Also, our more recent registrations (Athens and Topeka) seem to be “cleaner” than our earlier registrations (Augusta). I suspect people are simply becoming more accustomed to the concept.
ZIP obviously is important, but age and sex also turn out to be very useful targeting tools. Belo did some interesting work adjusting ad campaign creative by gender; Wes Jackson discussed it at the E and P conference recently.
One last point. Even if 10 percent of the registrants are liars, you have 90 percent targeting accuracy. That’s a whole lot better than nothing.
I agree that the story was poorly sourced and there could have been more data.
I actually support the idea of registration as a way of knowing your readers and creating some sense of accountability in forums when they’re offered.
I still think that demographics are a poor way to sell online advertising and that it’s not sustainable in the long run. It may be a necessary evil while advertisers learn the medium, but I don’t get the sense that publishers are learning any faster. Google is ahead of the market here and they’re being rewarded for their vision.
Having said that I have to acknowledge that Google’s ad model is not directly transferrable to publishing because their volume is orders of magnitude greater.
However, I still feel that publishers are still trying to sell online advertising like ROP, demographics is a crutch, and registration is not a cure for what ails online advertising by a symptom of another disease.
Demographics driven by registration systems are just one component of a targeting strategy. Expressed interests can be married with implied interests (derived from on-site behavior — if you’re searching the cars classified AND you’ve expressed an interest in cars, we’d qualify you as interested in making a car purchase) and measured based on leads/clicks/calls to action.
Google’s targeting mechanisms are query and context, not implications of behavior. And Google’s mechanisms obviously work, especially the query methods more than context.
So we need to add those mechanisms, smartly, to interest and behavior targeting. Demographics help, too, but advertisers seem to want “gross” demographics (males or females) vs. “fine” demographics (males 25-34 in ZIP code 00000 with $x income and declared interest in cars).
I usually don’t go thru the registration process unless it is an article I just have to read. Usually I can find it “syndicated” somewhere else or possibly cached. The registration is actually a major obstacle to reading the articles and I think most people don’t want to register to read articles.