You are hereBlogs / Jay Small's blog / Registration evolution at Scripps sites
Registration evolution at Scripps sites
The following item, regarding user registration at sites I work with in my "day job," was also posted to two newspaper industry e-mail lists. My apologies if you've already seen it as a member of either list.
At 13 of E.W. Scripps' daily newspaper Web sites, we just finished a month-long test of changes to our content-access registration protocol. I know many media managers remain interested in how industry peers are doing with user registration. So I want to explain what we learned, and how our sites are evolving registration methods as a result.
When Scripps first implemented registration, the "wall" where users had to register to get to content was set between index pages and almost all articles on our sites. We use a "long form" that asks for contact information, demographics and interests. About half the questions are optional.
With the test, we wanted to address concerns that the wall restricted site traffic growth too much, especially traffic coming from search engines. We softened the wall so that a user (client, technically) that was not logged in could view up to a set number of articles in a 30-day period before being intercepted by the registration form. We did not publicize the test in any way; site managers didn't even know which threshold they got.
We tested a range of different thresholds on Scripps sites, from no change in current protocols all the way to completely open access, and including thresholds of three, four, five, six and seven "free" article views per 30-day period. Our sites in Naples and Denver do not have content access registration, so they were not in the test.
Here are the questions we tried to answer, and what we found:
Traffic: Does softening the wall increase traffic noticeably?
We did not find a consistent pattern for page views under any threshold.
Of our "control" sites, which kept hard-wall registration through the test, one saw 9 percent growth, while the other two saw modest declines from the preceding 4-week period (yes, comparing current and preceding 4-week periods is imperfect, but we tried to make seasonal and incidental adjustments).
At the other extreme -- sites that had the wall completely lifted -- one saw 11 percent growth (much of which we attribute to a single-day, single-story spike) but the other two saw 2 to 4 percent declines.
Of the sites with thresholds set from 3 to 7 "free" article views in 30 days, the best performer saw 14 percent growth; the worst, a 4 percent decline.
What about unique visitors? We did not expect to find a pattern there, and we didn't. The test was not announced to the public, so there would be no reason to expect an influx of new people who somehow become aware the protocol is changed.
Abandonment: Does softening the wall reduce abandonment of our sites when users do encounter the registration screen?
The percentage of users who abandon at the wall remains constant, regardless of threshold. But far fewer users ever see the registration screen when thresholds are used. As such, the real numbers of abandoning users go way down, to in some cases 10 to 20 percent of former levels. That is an expected and desired effect of the threshold model.
These sites have had registration quite a while. Many "regulars" already are logged in. So not that many unique visitors encounter the wall in the first place, which may partly explain why we didn't see a rush of new article page views.
Opt-ins: During registration we offer users a checkbox to opt in to an e-mail marketing program featuring local advertisers' special offers. What effects do different thresholds have on the growth of our opt-in list?
The short answer: New opt-ins declined substantially -- more than half or worse -- with any softening of the wall. This is logical. The less often new site visitors are shown the registration screen, the less opportunity they have to opt in.
We found that thresholds of three or four "free" views per 30 days tended to preserve more acceptable opt-in growth rates. Softer thresholds took opt-in growth rates down to close to 1 percent; lower, obviously, when registration was completely removed.
Next steps
Local managers of the 13 sites that participated in the test are deciding this week where to set their soft-wall thresholds permanently.
To preserve a balance between continued growth of our e-mail marketing programs and optimum customer experience, we corporate folks advised site managers to choose thresholds between three and five "free" article views per user per 30 days. The lower the number, the more people will encounter the wall, but the higher opt-in growth should be.
We believe even a threshold of three should allow most search-engine-referred traffic to pass through to articles unimpeded, given that unregistered user sessions from external referrers tend to be only 1-2 page views. We call this crowd "grazers" or "drive-bys," and it doesn't make business sense to try to capture their profiles in our registration database anyway.
Once we set the thresholds to permanent levels, site teams will be free to announce the changes to their visitors and to other local-interest sites (read: blogs) that may wish to link in. We also plan to notify the major search engines and other aggregators that tack the dreaded "reg. reqd." on links to our sites' articles.
We'll continue to watch site traffic to see if publicity about the changes has any effect. I'll be happy to field questions in comments on this post.
Both blogs
[...] ADDENDUM: An alternative to forced user registration is setting a “soft wall,” where casual website visitors don’t hit a registration requirement until they’ve seen several articles, which allows those people who come in via a search engine link to view an article. I think the voluntary approach described above is better, but Jay Small of Scripps writes about that company’s experiments with soft-wall registration variations. Worth a read if you care about this topic. [...]
[...] Registration evolution at Scripps sites. Jay Small passes along another user registration item that also makes sensed  soft walls. The industry is learning, it seems. [...]
[...] Jay Small and Scripps Newspapers revieled their findings on registration roll back testing. [...]
I hit a wall, I leave. Open access please. Get over yourselves or die with the rest of oldmedia
You make no analyses on how accurate the information is for the people who do register. I invariably lie about EVERYthing when forced to register, giving up only my e-mail address. From anecdotal comments read in passing on the internet, I have to think I'm not the only one who does that. In which case, why bother with the registration if what you're gleaning is that Great Aunt Tillie beamed down from the mothership to access your site?
I hit a Wall, I leave too. I come looking for information not advertizing. Generally I already know the info, I'm just checking its accuracy before I pass it on.
You can influence me (and others through me) but the odds are indistinguishable from zero that I'll ever purchase anything through an ad in a newspaper 1000 miles from home.
Do you want NATIONWIDE influence or $0.02? That's your choice.
2 things to say about registration for newspapers...
1. If a google search indicates "registration required" I will often not select that link unless it is a local story that doesn't have wide coverage. Most stories are covered by numerous sites. And a great deal are just word-for-word repeats of a Reuters or other wire story. Why register?
(And google does say registration required on most stories, even if you have some "free" threshold set.)
2. When I am forced to "register" I never give accurate information. Specifically, I never give an email address that I care about. How many Chuck U. Farley or Jim Shoe registrations do you need? Are they of any use to you?
But usually (95%) when I hit a registration form, I look for another source. But like I said, if registration is required in the first place, I usually look elsewhere.
To NahnCee:
In my five years of working with local news site registration systems, certain patterns have been consistent:
1. The vast majority of users who encounter the "wall" fill out the form. So we must have at least something good going on behind the wall -- something that can't be easily found elsewhere -- or everyone would do as Mike, Joe and Zendo Deb report they do.
2. The vast majority of completed registrations appear to our eyes to be addressable profiles from real people. When we send e-mail marketing campaigns on behalf of advertisers to those people in our registration database who opted in for those programs, the response rates are very good -- better if we use the registration data to target the messages. That fact alone suggests enough of the data are accurate to lend real market value to the profiles.
3. The obviously spoofed profiles, 90210 ZIP Codes and known BugMeNot trojan entries represent only 2-3 percent of the total database. (As for BugMeNot IDs, I long ago lost interest in trying to remove them, figuring that having a few known, unaddressable profiles in the data was just easier to manage than trying to escalate an arms race.)
To Mike, Joe and Zendo Deb:
Your use cases are exactly why Scripps sites implemented the registration threshold. From what you describe, you grab bits of news from here and there. The odds are remote that you would conduct enough sessions, or deep-enough sessions, on any one of our sites to ever encounter registration.
It serves no purpose for us to collect profiles of people who spend only rare, brief time on our sites. It is not likely such people would respond to any of our advertisers' attempts to market to them.
How do you reconcile imperceptible traffic growth with reduced abandonment? It would seem to me that this would indicate greater quality of traffic, which should result in a higher value to your advertisers.
You do mention that "drive-bys" aren't a real target for your local advertisers, but doesn't Scripps take any national/international advertising on their Web sites? "Drive-bys" should still be valuable to those folks. Doesn't that indicate that you could optimize your ad engine to prefer advertising with a wider scope to unregistered readers using their 3-per-month free views?
This, I think, should make you worth even more to your local advertisers, since you give them a hook to identify people with more than casual local interest, and not waste time marketing to the rest. National advertisers, too, could have some control over their local marketing vs. national marketing, which they might want to pay for. Even if they decreased their advertising because of low national reach, they would at least have some confidence in the numbers they're getting, which sounds like a problem in the newspaper world today.
As a frequent "drive-by" who is irritated by the gajillion registrations he must keep up (see the Web site link for more on that), I hope you continue to see some value in catering to our needs.
Thanks for writing, Jeff:
Traffic growth is not imperceptible across the board; it just follows no consistent pattern and not all sites that softened their walls for the test saw growth. In a longer test, they all probably would have.
Also, as a percentage of total unique visitors, not that many people encounter the wall nowadays anyway -- with or without the "free" views. So reduced abandonment just doesn't give us that much of a lift, because abandonment wasn't that big a problem when compared to site traffic overall.
Scripps sites do get a modest volume of nonlocal advertising, as I know many other local media sites do. It isn't nearly enough to fill out our inventory of "untargeted" advertising positions by itself.
Two things we've observed:
1. Probably more than half the overall page views on our sites go to areas that do not require registration at all: home and index pages, and Classified ads. We can, however, measure the traffic generated by registered users in those "open" areas, and get pretty reliable demographic/behavior data there over time. We couldn't do that without content access registration.
2. Whether the advertisers are national or local, some are far more compelled to use behavior, demographic or interest targeting than others. Registration and the resulting targeting mechanisms don't do much for advertisers who just want high-volume impressions.
You are absolutely right when you say allowing the drive-bys to avoid registration makes the database more valuable for local advertisers interested in targeting. That's one very good business reason to do what we did.
It's really all about balancing the business needs with best possible user experience -- some compromises, to be sure, but that's the world we're in.
And to your last point, now that we've decided to soften the wall, I think it'll stay that way a long time.
[...] Their experience shows that there are trade offs between registration and free content. This article was originally posted as a blog at: http://smallinitiatives.com/2006/04/19/registration-evolution-at-scripps... [...]
Measuring the Free Pageview Threshold
Jay Small has a good luck at some tests using 13 of E.W. Scripps' daily newspaper Web sites. Basically, it sounds like they varied the number of free pageviews a visitor got until they got a registration screen. (What Steve Outing has called "soft wa...
[...] Scripps decided to covertly (in a nice way) test how people react to hitting the wall. Jay Small tells us all about it on his blog. We softened the wall so that a user (client, technically) that was not logged in could [...]
[...] good sense for news organizations. In my opinion, it does not make sense to require registration just for reading stories on the site — but for the privilege of posting a comment, it makes a world of sense. (There [...]