| 1:56 pm on May 18, 2007 (gmt 0)|
This can only be a positive step. Forced to 'fess up on what they're really doing they'll both have to get better.
Of course the panel method has some pretty severe limits, especially for smaller sites. You may have stable traffic, but a panelist or two showing up (or not) causes wide swings in their stats for you. On the other hand, counting every single visitor opens you up to swarms of bots trying to skew the stats.
Maybe they'll both have to get larger, broader, more representative panels to compete now. That could enhance accuracy.
| 2:42 pm on May 18, 2007 (gmt 0)|
|comScore and Nielsen, meanwhile, prefer a panel method |
Well - they would, because it's a bigger barrier to entry. But I really think they have no choice but to open up. How on earth can you profess to provide research data without having the statistical analysis and methods open to scrutiny?
Conscore tried to bite back [webmasterworld.com] with a claim abourt coookie tracking, but it's hard to throw stones when you lilve in a glass house.
| 3:40 pm on May 18, 2007 (gmt 0)|
Anything to make panel based methodology and their severe shortcomings more transparent is good.
Biggest problem with these guys....their work panels are microscopic.
| 8:47 pm on May 18, 2007 (gmt 0)|
Well its no surprise most of these metrics have just audited how well certain companies are at hosting the images of some of best free image posting companies.
Mind you "Hits" and human webtraffic are two totally different things.
| 9:32 pm on May 18, 2007 (gmt 0)|
|Biggest problem with these guys....their work panels are microscopic. |
So are Nielsen's TV ratings panels, but advertisers don't seem to mind. Of course, there are a lot more Web sites than there are TV channels, so it's easy to see why the TV-ratings approach might viewed with skepticism on the Web.
I think there are bigger problems than the size of the ratings panel, though. A few years ago, FORBES reported on how About.com had subverted the MediaMetrix rankings by buying HTML popunder windows as ads. MediaMetrix counted the popunders as visits or pageviews. (Come to think of it, X10.com ranked as one of the "most visited sites on the Web" for what I assume was the same reason.)
| 3:03 am on May 19, 2007 (gmt 0)|
People don't watch TV at work, but they do browse. If you have a site that gets a big draw from people in the office, Nielson/Comscore will really get your numbers wrong.
The ironic thing is that this hurts advertisers (the only people who really use these services) because when evaluating sites, ones with a more affluent demographic should have a greater percent of their traffic coming from people in offices and that traffic gets under reported.
And yes there are so many problems with their methodology that its hard to point to one as the big weakness.